08:30 - 09:00
09:00 - 09:30
09:00 -
Welcome, ESA’s EO Programme
Borgeaud, Maurice - ESA/ESRIN, Italy
N/A
09:20 -
WS Objectives and Logistical Information
Berger, Michael - ESA/ESRIN, Italy
N/A
09:30 - 11:10
09:30 -
UNCCD LDN and GEO’s LDN Initiative
Wheeler, Ichsani - OpenGeoHub Foundation [UNCCD] Wageningen University & Research, Netherlands, The
N/A
09:50 -
GloSIS: Towards a Global Federation of Soil Information Systems
Yigini, Yusuf (1);
Vargas, Ronald (1);
Viatkin, Kostiantyn (1);
Luotto, Isabel (1);
Baritz, Rainer (3);
Kempen, Bas (2);
van den Bosch, Rik (2);
Olmedo, Guillermo Federico (4);
Duque Moreira de Sousa, Luís (2);
Mendes de Jesus, Jorge (2) - 1: Global Soil Partnership, Food and Agriculture Organization of the United Nations, Italy;
2: ISRIC - World Soil Information, The Netherlands;
3: European Environment Agency, Denmark;
4: Instituto Nacional de Tecnología Agropecuaria, INTA ·EEA Mendoza, Argentina
The Global Soil Partnership (GSP) is a global platform to bring key stakeholders together with a common vision to improve the sustainable management of limited soil resources. To achieve this mandate GSP implements various activities under 5 Pillars of Action. The Pillar 4 (“Information and Data”) aims to enhance the quality of soil data and information in order to organise, empower and facilitate soil knowledge, data and impact to address the current environmental and societal challenges. Pillar 4 Global Implementation Plan (P4GIP) is based on the endorsed Pillar 4 Plan of Action (P4PoA) and aims to develop an “enduring and authoritative global system to monitor and forecast the condition of the Earth’s soil resources”. The Pillar 4 Implementation Plan sets out activities to establish a federated global soil information system (GloSIS). The GloSIS will be housing three types of data products; soil profile data , global soil polygon coverage and global soil grids. It proposes a two-tier model for the soil profile and point data. Tier 1 as a ‘comprehensive federated database’ with minimal data requirement and no stringent requirements. Tier 2 will be a ‘world reference database’ with well-described, harmonized and quality-assessed. The Global Polygon Coverage is foreseen as an updated and improved version of the FAO/UNESCO Soil Map of the World. Three types of global grid products are identified: i) an updated Harmonized World Soil Database (HWSD), ii) fine resolution soil grids version 0 which will be a collection of existing grids (1 km resolution) with no attempt at harmonization, iii) fine resolution soil grids version 1, which will be a collection of consistent, standardized grids of soil properties with global coverage at high resolution (100 – 250 m). While the GSP, GSP Soil Data Facility (ISRIC World Soil Information), INSII (International Network of Soil information Institutions) and Pillar 4 Working Group working on the design of the infrastructure of the GloSIS, the first data product of GloSIS was launched in 2017 on the World Soil Day. The data set is the first gridded soil map product produced for GLOSIS through a country-driven approach at 1 km spatial resolution following a semi-standardized approach.GLOSIS is envisioned as a federation of soil information systems, which will be relying on national soil information systems and share interoperable soil data sets via web services. Since there are large differences between data holders in how soil data are stored, managed and disseminated, the GloSIS will give soil data holders to choose between different levels of participation to GloSIS, acknowledging the differences in technical level, technical skills and resources of data holders as well as differences in ambitions that data holders might have for setting up and maintaining their own SIS. Besides the above-mentioned basic soil data products, the P4GIP outlines a system for monitoring, forecasting and status reporting of the soil resource called ‘SoilSTAT’, using soil indicators, as rasterized spatial assessments and derived statistics. This system will be relying on reported country statistics and GLOSIS data elements as input.
10:10 -
EU’s Soil Thematic Strategy and EIONET Task Force Soil Monitoring
Baritz, Rainer - European Environment Agency [EEA], Denmark
N/A
10:30 -
LUCAS Topsoil Initiative and the European Soil Data Center (ESDAC)
Jones, Arwyn - European Commission Joint Research Centre, United Kingdom
N/A
10:50 -
The Role of Remote Sensing in Global Digital Soil Mapping: the Example of SoilGrids
de Sousa, Luís Moreira;
Poggio, Laura;
Kempen, Bas;
Heuvelink, Gerard;
van den Bosch, Rik - ISRIC - World Soil Information, Netherlands, The
Soil is key in the realisation of a number of UN Sustainable Development Goals by providing a variety of goods and services. Soil information is fundamental for a large range of global applications, including assessments of soil and land degradation, sustainable land management, and environmental conservation. It is important to provide free, consistent, easily accessible and standardised soil information. SoilGrids is a global product that fulfills two main goals: 1) it is a source of consistent soil information to support global modelling, and 2) provides complementary information to support regional and national soil information products in data-poor areas. The modelling approach is based on state-of-the-art statistical and geo-statistical approaches, in particular machine learning. It includes the assessment of the uncertainty associated with the predicted soil properties. SoilGrids routinely uses remote sensing to derive environmental covariates to model soil properties at the global scale. The main covariates considered are morphology (e.g. elevation, landform), vegetation information (e.g. NDVI and other vegetation indices), climate (e.g. precipitation, land surface temperature) and human factors (e.g. land use/cover). These covariates are derived from a mix of sensors, such as MODIS and Landsat. Currently we are exploring the integration of Copernicus products such as Sentinel 1 and Sentinel 2. This work will present some of the challenges, advantages and potential limitations of using the existing Copernicus products in the SoilGrids methodology.
11:10 - 11:30
11:30 - 13:10
11:30 -
Soil Erosion as a Global Threat: Translating Science into Policy and Action on the Ground
Lefèvre, Clara (1);
Vargas, Ronald (1);
Borelli, Pasquale (2);
Bazza, Zineb (1);
Bottigliero, Fiona Maria (1);
Caon, Lucrezia (1);
Cuevas Corona, Rosa (1);
Luotto, Isabel (1);
Rodriguez Eugenio, Natalia (1);
Sala, Matteo (1);
Stanco, Giulia (1);
Tong, Yuxin (1);
Verbeke, Isabelle (1);
Viatkin, Kostiantyn (1);
Yigini, Yusuf (1) - 1: Global Soil Partnership, Food and Agriculture Organization of the United Nations;
2: University of Basel, Switzerland
Soil degradation due to erosion was identified as the main soil threat in the 2015 Status of the World Soil’s report (FAO & ITPS, 2015) and is the 2019 identified theme for the Global Soil Partnership (GSP), over the motto “stop soil erosion, save our future”. The starting point of the GSP action and implementation of erosion activities at global level was the Global Symposium on Soil Erosion (GSER19, 15-17 May, FAO HQ, Rome). Activities to implement were identified by the GSP Secretariat in collaboration with its Intergovernmental Technical Panel on Soils (ITPS) and working groups of experts on different topics (erosion mapping and assessment, erosion control practices and policies and economics of soil erosion) and discussed all along the event with all participants. Three main actions will be implemented: (1) development of the Global Soil Erosion Map (GSERmap), (2) establishment of a policy plan of action towards better policy implementation and (3) a global study on the economics of soil erosion. The global soil erosion map will be developed following an interactive and participatory approach, using three phases of development, based on respectively global, national and local datasets and the establishment of monitoring sites at local level. To build the maps, different guiding documents will be developed in collaboration with the ITPS and the working group of experts (i.e. detailed concept note, technical specifications and guidelines), and training and workshops will be organized to enable countries on erosion modelling (all supported by a cookbook manual). The global map will enable the comparison of erosion status between geographical regions and countries in order to trigger governmental action to mitigate this global threat. First version of GSERmap is expected by the end of 2020. In parallel, a global analysis of existing policies and policy gaps addressing soil erosion will be made in coordination with regional soil partnerships and working groups and will allow the definition of a plan of action, regionally defined, to fill policy gaps. The plan of action will be presented at the 24th Conference of the Parties of the United Nations Convention to Combat Desertification (UNCCD COP24, India, September 2019) and will include the publication of information documents (e.g. policy briefs) and multi-stakeholder workshops on policy definition and implementation. A global assessment of the cost-benefit of soil erosion and soil erosion control will be held, based on the compilation of existing works (e.g. The Economics of Land Degradation (ELD) initiative) and will aim to inform policy-makers and land-users on the benefits provided by Sustainable Soil Management (FAO & ITPS, 2017) and the economic cost of inaction.
References
FAO & ITPS. The Status of the World’s Soil Resources (Main Report) (Food and Agriculture Organization of the United Nations, Rome, 2015).
FAO & ITPS. Voluntary Guidelines for Sustainable Soil Management. (Food and Agriculture Organization of the United Nations, Rome, 2017)
11:50 -
The World Atlas on Desertification
Weynants, Melanie - EC - Joint Research Centre [JRC] World Atlas of Desertification, Italy
N/A
12:10 -
GlobalSoilMap: a Scientific Framework for Global Soil Mapping and Monitoring
Mulder, Vera Leatitia (1);
Roudier, Pierre (2);
Libohova, Zamir (3);
Lagacherie, Philippe (4);
Grundy, Mike (5);
McBratney, Alex (6);
Arrouays, Dominique (7) - 1: Wageningen University, Wageningen, The Netherlands;
2: Landcare research, Manaaki Whenua, New-Zealand;
3: US department of Agriculture, Natural Resources Conservation Services, Lincoln, Nebraska, USA;
4: INRA-IRD-Supagro, UMR Lisah, Montpellier, France;
5: CSIRO, Australia;
6: University of Sydney, Sydney, Australia;
7: INRA, InfoSol Unit, Orléans, France
Soil plays a crucial role in global issues, such as food and water security, climate regulation, energy sustainability, biodiversity protection and land degradation. Soil security achieved by sustainable management is a global issue in itself, and effective actions require high-resolution soil property data. Detailed localised soil knowledge is also critical for the planning of multifunctional landscapes. Digital Soil Mapping (DSM) has been developed by the scientific community as a methodology and platform for generating high resolution spatial information on key soil properties using legacy soil data and ancillary spatial information. Earth Observation data have been demonstrated widely to be crucial in the DSM methodology and provides key information for improving the accuracy of digital soil property maps.
An operational outcome of DSM has been the GlobalSoilMap initiative (GSM). GSM, launched in 2006, which aims to deliver the first generation of high-resolution (~ 90 m) soil property grids for the globe, following a bottom-up approach (from country to globe). Over the years, an international consortium of soil scientists has developed the GSM specifications which ensure the convergence of the different DSM products that have been developed across the globe toward a unified and interoperable global soil information system. Recently, these GSM specifications have been adopted by the United Nations FAO Global Soil Partnership, Pillar 4. As a by-product of these specifications, guidelines for DSM methods that can achieve these products have been proposed. These guidelines have been transferred widely and are updated constantly so that the best (minimum uncertainty) possible maps can be produced, whatever the location across the globe. However, these guidelines still do not pay enough attention to the use of available earth observation data. In order to keep the scientific guidelines suitable for stakeholders’ needs and considering the progressing state-of-the-art, new updates of guidelines of DSM methods are necessary. These involve the use of spectral libraries, proximal and remote sensing data for the prediction of soil properties, and soil condition and functioning in space and time. In this context, Earth Observation data with high spatial, spectral and temporal resolutions are essential, mainly because the characterization of soil properties by their spectral signatures is complex due to the spectral mixture of organic and inorganic soil constituents. For this, Copernicus and especially Sentinel data have proven to be key for advancing our prospect of a global soil monitoring system. A potential pathway to further improve soil research activity and intensity to support UN priorities is to align some GSM activities with GEO and the Global Earth Observation System of Systems (GEOSS).
12:30 -
Soil data, activities and collaboration within the EuroGeoSurveys: the Earth Observation and Geochemistry Expert Groups contribution
Kopackova, Veronika (1,2);
Ladenberger, Anna (2);
Négrel, Philippe (2);
Demetriades, Alecos (2) - 1: Czech Geological Survey, Czech Republic;
2: EuroGeoSurveys, Brussels
EuroGeoSurveys (EGS) is a non-profit organisation representing 37 National Geological Surveys and some regional Surveys in Europe, an overall workforce of several thousands geoscience experts. Taking into account the focus of this user consultation workshop organised by ESA, we will present the possible contribution of two expert groups working within the EGS: the Earth Observation Expert Group (EOEG) and the Geochemistry Expert Group (GEG). First, we will present soil-related data, which can be shared with the scientific community. In this part of the presentation, we will describe data sets collected under two pan-European Geochemical projects that included soil and were carried out by the GEG: The FOREGS and GEMAS projects. Such data sets can be possibly used for ground-truthing of Earth Observation-based models. In addition, we will present soil proximal Remote Sensing data; these spectral data (about 600 samples in total) were collected under a well-defined protocol for the whole soil profiles going down to 1-m depth). In order to normalise and align the spectral measurements the internal soil standard (ISS) concept, in which a soil standard sample exhibiting stable spectral performance, was used. We will demonstrate how these normalised spectroscopic data, together with mineral and geochemical analyses, can be employed to model different soil parameters taking into account spectral resolution of Sentinel-2 (S-2) and EnMAP sensors. In the second part of the presentation, we will present a new approach developed for mapping acid sulfate soil, which is either natural or anthropogenic soil containing iron sulfide minerals (predominantly as the mineral pyrite) or their oxidation products. They can possibly cause environmental problems especially in places with fluctuating water table and we will demonstrate how S-2 data can be used efficiently to detect seasonal changes in these soil types.
12:50 -
Modeling and Monitoring Soil Variables in Spacetime: Towards Global 3D+T Soil Information Systems
Hengl, Tomislav;
Wheeler, Ichsani;
MacMillan, Robert A. - OpenGeoHub Foundation, the Netherlands
OpenGeoHub, through it's LandGIS initiative (https://landgis.opengeohub.org), has been promoting global soil mapping based on the global compilations of legacy soil data. We have recently produced a comprehensive stack of global spatial predictions of key soil properties and classes, usually at 250 m spatial resolution. Here we should especially emphasize global predictions of the USDA great groups (https://doi.org/10.5281/zenodo.1476844) and initial predictions of soil available water capacity (https://doi.org/10.5281/zenodo.2629148). We use “old legacy soil profile data” to produce new added-value information immediately and affordably without new significant investments, then boost access to this data by using diverse web-services (Geoserver, Web Coverage Service, REST API). But many soil properties such as soil moisture, salinity, nutrients, including soil organic carbon vary not only in 3D but also through time. We have recently tested using Ensemble Machine Learning spatiotemporal models to predict soil organic carbon for periods of 50+ years (https://envirometrix.github.io/PredictiveSoilMapping/SOC-chapter.html#deriving-ocs-using-spatiotemporal-models), and also 3D+T models where predictions can be made at any location, depth and time (https://peerj.com/articles/5518/). The initial results look promising and 3D+T modeling clearly provides many benefits over simple 2D or 3D modeling: models can be used to explain and predict impact of climate change on future soil properties and distinguish main factors of soil change. This is especially interesting for projects focused on mapping and monitoring soil carbon, such as the UNCCD's Land Degradation Neutrality (https://www.unccd.int/actions/achieving-land-degradation-neutrality). Not all soil variables need to be mapped in a spatiotemporal continuum. Soil texture classes, soil minerology and similar do not change at scale of few hundred years. Likewise, for some soil properties there is probably not enough training data for any 3D+T modeling. We are only about to start collecting spatiotemporal training points, and it might take decades until we will be able to have anough training data to build comprehensive and usable 3D+T models.
13:10 - 13:30
13:30 - 14:30
14:30 - 15:50
14:30 -
Passive (VNIR-SWIR-TIR) Remote Sensing of Soils
Ben Dor, Eyal - Tel Aviv University | TAU, Israel
N/A
14:50 -
Potential Contributions of passive and active microwave observing systems for Global Soil Mapping
Wagner, Wolfgang - TU Wien, Austria
In contrast to optical remote sensing data, synthetic aperture radar (SAR) data have so far been hardly used for mapping of soil properties. The reason for this probably is that SAR backscatter measurements are more difficult to interpret than optical reflectance data, and hence less straight forward to relate to vegetation and soil properties. This is a missed opportunity as the electromagnetic pulses emitted by SAR sensors are able to penetrate much deeper into vegetation and soils than optical rays. Therefore, SAR sensors can provide more direct information about vegetation biomass and soil properties than optical data, and hold hence a significant potential to enhance remote sensing based approaches to soil mapping. In this contribution, the information content of a global high-resolution Sentinel-1 backscatter data cube as made available by the EODC Earth Observation Data Centre is examined. This Sentinel-1 data cube allows to extract different value-adding data products in a relatively straight forward manner, thereby allowing to compress the information content of all available Sentinel-1A and 1B data acquisitions for subsequent input into soil mapping algorithms. Amongst the most promising Sentinel-1 value adding data products are the following: (i) global backscatter statistics (mean, quantiles, etc.) carrying information about surface roughness over non- or sparsely vegetated areas and vegetation biomass in the more densely vegetated regions, (ii) monthly cross-ratio vegetation index data that reflect seasonal vegetation dynamics, and (iii) maps showing the presence of shallow rocks and gravel in the first few decimetres of the soil. These maps can be generated by analysing Sentinel-1 time series together with modelled soil moisture data, because at very dry soil conditions the SAR pulses penetrate deep into the soil interacting with strong sub-surface scatterers like rock surfaces and stones.
15:10 -
Supporting Soil Health and Sustainability by utilizing Hyperspectral data – the need for standardization
Zalidis, George C. - Laboratory of Remote Sensing, Spectroscopy and Geographic Information Systems (GIS),Aristotle University of Thessaloniki, Greece
The health of the soil ecosystem is of paramount importance as it helps maintain a diverse community of soil organisms which in turn control plant disease, insect and weed pests, form beneficial symbiotic associations with plant roots, recycle essential plant nutrients, improve soil structure with positive effects for soil water and nutrient holding capacity, and ultimately improve crop production. A healthy soil also contributes to mitigating climate change by maintaining or increasing its carbon content.
In the past decades, the use of VIS-NIR-SWIR (visible, near and short-wave infrared) to monitor and support the soil ecosystem has gained significant traction. Large soil spectral libraries have been developed throughout the world, and numerous efforts have concretely contributed towards the effective estimation of soil properties via VIS-NIR-SWIR spectra. Notwithstanding the above, the current efforts remain largely fragmented, with each researcher following his own protocol for acquiring the measurements, cataloguing them, processing the information, and producing the final outputs. This disparity must be addressed and to that end the scientific community and ESA must together form a working group in order to develop guidelines and international protocols of specific standards and best practices (including example of usages of the Copernicus Programme) on how to systematically record, catalogue, harmonize and process soil spectra in real field conditions. If everyone adheres to these new standards, this will facilitate the effortless future processing of the data for e.g. identifying the most important wavelengths for the prediction of key soil properties pertaining to soil health. This will pave the way for the overarching objective to develop a soil monitoring system by integrating in situ and satellite earth observation measurements.The specific points that must be elucidated and clearly defined are: a) how to acquire and standardize soil spectra in the field and in the laboratory able to integrate with existing soil spectral data archives, b) what information and associated metadata must be collected, c) how to efficiently catalogue and develop soil spectral libraries, d) what are the best tools and practices (e.g. machine learning methods) which are the standard state-of-the-art in terms of data processing, e) how specific methodologies (e.g. bottom-up approach) can lead to the operationalization of Sentinel-2 data. In this work, we present some initial thoughts upon this matter and propose a way forward in order to achieve a set of standards and specific practices.
15:30 -
The Brazilian Soil Spectral Library (BSSL): from Construction to Community and Consulters
Demattê, José Alexandre Melo;
Paiva, Ariane Francine da Silveira;
Dotto, André Carnieletto;
Contributors BLSS https://esalqgeocis.wixsite.com/geocis/besb, Complete list of - University of São Paulo (USP), Brazil
The Brazilian Soil Spectral Library (BSSL) started with a collection of soil samples in 1995 at the Department of Soil Science, University São Paulo (ESALQ-USP). Currently, BSSL has gathered data from 26 Brazilian states, with a total of 39,000 samples from 65 contributors and 41 institutions.The objective of this studyis to present the dynamics of BSSL construction and utilization by community. The spectral data range was from visible to shortwave infrared (350 to 2,500 nm) inlaboratory. All collaborators sent soil samples to a unique spectral laboratory, which acquired the information. The delivery of researchers was exectuted by personal communications in order to speed up the generation of the BSSL. We determined how many spectral patterns were required to represent the entire Brazilian soils (6 types). It presents the utility of national spectra to predict soil attributes, such as organic matter (OM), sand, silt, clay, cation exchange capacity (CEC), and pH. The models indicated fairly predicted values for clay (R2 between 0.55 and 0.70). Spectra also indicated important information to assist on soil classification. This study demonstratedthe potential of BSSL for tropical soil evaluationand indicated differences in spectral segmentation by regional charcteristics such as biomass, geology,and vegetation. We constructed three approaches in a site as for practical means: a) interaction between researchers and users. For this porpose we designed an interactive website (URL), where you can locate partners for joint research development and get direclty in contact with them; (b)quantification of spectra: we constructed a link where the user introduces his spectra and gets the soil attributes quantification analysis; (c) soil classification: the user inserts the spectra of A and B horizon and gets the most probable soil classification. These applications can be used in soil mapping, soil analysis, precision agriculture, , soil classification and other approaches. The website provides a unique interface to manage the BSSL, which will be presented in this lecture while sharing ideas and suggestions for a similar worldwide utilitiy.
15:50 - 16:10
16:10 - 17:10
16:10 -
Regional Characterization of near Surface Soil Properties via a Combination of Methods from Multispectral Sentinel-2 Data, Hyperspectral Data, Geophysics and Field Data
Frei, Michaela (1);
Fries, Elke (1);
Meyer, Uwe (1);
Scheper, Simon (1);
Waldmann, Frank (2);
Chabrillat, Sabine (3) - 1: Federal Institute for Geosciences and Natural Resources (BGR), Hannover, Germany;
2: Regierungspräsidium Freiburg – Dept. 9 State Authority for Geology, Mineral Resources and MiningFreiburg i. Br., Germany;
3: German Research Centre for Geosciences (GFZ) Helmholtz Center PotsdamPotsdam, Germany
Detailed soil information is a valuable resource for various approaches like agricultural management, soil protection, or spatial planning. Preserving, using and enriching soils are complex processes that fundamentally need a sound regional database. Many countries lack this sort of extensive data or the existing data must be urgently updated when land use changes in major patterns. The projects "ReCharBo" (Regional Characterization of Soil Properties) and “BopaBW” (Soil Parameter Baden-Wuerttemberg) aim at the combination of methods from remote sensing, geophysics, pedology, and digital soil maps, in order to develop a new system to map soils on a regional scale in a quick and efficient manner. This system could be very useful especially in countries where comprehensive harmonized soil data are scarce. First tests are performed in existing soil monitoring districts and on existing and newly developed soil databases, using newly available sensing systems as well as established techniques.
High resolution digital soil maps do exist in Germany, as for many countries basic soil data and soil maps in even medium resolution are lacking, however, more sharpness of detail and a higher data density and monitoring enables more precise planning. Therefore, the integration of remote sensing products is a powerful source for spatial mapping of near-surface soil parameters (NSSP). Among such NSSP are sand, silt, clay contents, soil density, soil moisture, and soil organic carbon, presence of carbonates, and surface-stone cover. Sentinel-2 remote sensing data are combined as a multispectral complementary data source with existing digital soil maps, and especially hyperspectral data measured from satellites or airborne platforms are systematically correlated with gamma-ray spectroscopy.
The results may demonstrate that the combination of traditional field surveys with remote sensing data can lead to a quick and comprehensive understanding of soil properties and their regional interactions. Furthermore, the integration of remote sensing data in soil maps enables to keep soil maps up-to-date and integrate new information cost-effectively and timesaving. The goal is to generate a system that enables users to map soil parameters and patterns on a local to regional scale using remote sensing data and to calibrate the data with only a limited number of soil samples.
16:30 -
Soil Organic Carbon Mapping in Croplands using LUCAS Topsoil Database and Sentinel-2 Data
van Wesemael, Bas (1);
Castaldi, Fabio (2);
Chabrillat, Sabine (3) - 1: Université catholique de Louvain, Belgium;
2: Ilvo, Vlaanderen, Merelbeke, Belgium;
3: GFZ German Research Center for Geosciences, Potsdam, germany
The spatial and spectral characteristic of the Sentinel-2 sensor are promising for soil applications, especially the presence of two SWIR bands coincide with the spectral signature of soil organic carbon (SOC) and texture. The short revisit time of the Sentinel-2 sensor increases the likelihood for collecting images during the narrow time window in which croplands are bare after seeding. The collection of images for the same area during several years mosaicking in order to increase the coverage of the croplands. The main challenge for SOC estimation of the plough layer in croplands using Sentinel-2 is the automatic selection of pixels of bare soil without vegetation residues and mimicking the conditions of a dry soil sample with a reduced roughness such as the ones in the LUCAS spectral library. Here we aim to map SOC in croplands within a large area (100 x 100 km) in northeastern Germany. A random forest model was build using c. 9000 LUCAS spectra in croplands resampled according to the S2 spectral resolution together with 80 S2 spectra extracted at the points for which samples were taken previously. The model performed quite well on an independent validation set (RPD = 1.56). Subsequently, we apply the model to the bare soil pixels in the image. The bare soil pixels were selected using the ESA cloud probability layer and NDVI < 0.35, reflectances in B3 > B2 and B4 > B3 (green vegetation) and finally we tested different values for Normalized Burn Ratio 2 (moisture and crop residues) and found that the RPD decreased sharply for NBR2 >0.2. Thus, the SOC content of c. 25 % of the area could be mapped. This represented the largest part of the bare soils (27 % of the area). This example illustrates that a reliable SOC map can be produced applying different indices for selecting the bare soil pixels and that multiple images combined with a digital soil mapping approach are required for a continuous SOC map for the croplands.
16:50 -
Research And Products For Soil Applications At Barcelona Expert Center
Pablos, Miriam (1,2);
Turiel, Antonio (1,2);
Vall-llossera, Mercè (2,3);
Piles, María (4);
Chaparro, David (2,3);
Portal, Gerard (2,3);
González-Haro, Cristina (1,2);
Camps, Adriano (2,3);
Herbert, Christoph Josef (2,3);
Portabella, Marcos (1,2) - 1: Institute of Marine Sciences (ICM), Spanish Research Council (CSIC), Passeig Marítim de la Barceloneta 37-49, 08003, Barcelona, Spain.;
2: Barcelona Expert Center (BEC), Passeig Marítim de la Barceloneta 37-49, 08003, Barcelona, Spain.;
3: CommSensLab, Universitat Politècnica de Catalunya (UPC) and Institut d’Estudis Espacials de Catalunya (IEEC/CTE-UPC), Jordi Girona 1-3, 08034, Barcelona, Spain.;
4: Image Processing Lab (IPL), Universitat de València (UV), Catedràtic José Beltrán 2, 46980, Paterna, Spain.
The Barcelona Expert Centre (BEC, http://bec.icm.csic.es) is a joint venture of the Spanish Research Council (CSIC) and the Universitat Politècnica de Catalunya (UPC) to support activities related to the European Space Agency (ESA) Soil Moisture and Ocean Salinity (SMOS) mission. More than one decade after, it became a sound reference in microwave data processing and applications, carrying-out research focused on data production and interpretation through several satellite technologies.
BEC activities over land are mainly devoted to remote sensing of soil moisture (SM) for developing added-value applications. Currently, SMOS Level 3 (L3) and Level 4 (L4) SM products are operationally generated and distributed (sftp://becftp.icm.csic.es:27500). The global SMOS L3 SM maps in a 25-km Equal Area Scalable Earth (EASE)-2 grid are a composite of all orbits within a certain time period. They are obtained after a carefully designed quality filtering and binning of ESA L2 SM [1]. The regional SMOS L4 SM maps in a 1-km EASE-2 grid are obtained by applying a downscaling algorithm based on the synergy of microwave and optical visible/infrared data fusion [2].
Many SMOS SM-derived applications were developed using L3 or L4 products. The L3 was used to monitor agricultural drought [3] and to assess the dominant modes of SM [4]. The L4 was employed to derive an index of wildfire risk [5], to analyze the decline of drought-prone forests [6], and to estimate agricultural drought [7], gross primary production [8] and root zone SM [9]. Furthermore, other research studies based on data from the Soil Moisture Active Passive (SMAP) mission were also performed, as the estimation of crop yield from SMAP vegetation optical depth (VOD) [10].
[1] BEC Team, “SMOS-BEC Land Products description v1.0”, Tech. Report: BEC-SMOS-0002-PD-Land, Barcelona Expert Center (BEC), 2018.
[2] Portal et al., “A spatially consistent downscaling approach for SMOS using an adaptive moving window”, IEEE J. Sel. Topics App. Earth Obs. Remote Sens., 11(6):1883-1894, 2018.
[3] Sánchez et al., “Integrated remote sensing approach to global agricultural drought monitoring”, Agric. For. Meteorol., 259:141-153, 2018.
[4] Piles et al., “Dominant features of global surface soil moisture variability Observed by the SMOS satellite”, Remote Sens., 11:95, 2019.
[5] Chaparro et al., "Predicting the extent of wildfires using remotely sensed soil moisture and temperature trends", IEEE J. Sel. Top. App. Earth Obs. Remote Sens., 9(6):2818-2829, 2016.
[6] Chaparro et al., "The role of climatic anomalies and soil moisture in the decline of drought-prone forests", IEEE J. Sel. Top. App. Earth Obs. Remote Sens., 10(2):503-514, 2017.
[7] Pablos et al., “Temporal and Spatial Comparison of Agricultural Drought Indices from Moderate Resolution Satellite Soil Moisture Data over Northwest Spain”, Remote Sens., 9:1168, 2017.
[8] Sánchez-Ruiz et al., “Quantifying water stress effect on daily light use efficiency in Mediterranean ecosystems using satellite data”, Int. J. Digital Earth, 10(6):623-638, 2017.
[9] Pablos et al., “Assessment of Root Zone Soil Moisture Estimations from SMAP, SMOS and MODIS Observations”, Remote Sens., 10:981, 2018.
[10] Chaparro et al., “L-band vegetation optical depth seasonal metrics for crop yield assessment”, Remote Sens. Environ., 212:249-259, 2018.
17:10 - 19:30
Scaling up Land Restoration Approaches to Reclaim the Hardpans of Niger for Agriculture using Sentinel 2 Imagery
Ahmed, Mohammed Irshad;
Hoskera, Anil Kumar;
Sanoussi, Laminou;
Mohammed, Ismail;
Bado, Boubie Vincent;
Whitbread, Anthony Michael - International Crops Research Institute for the Semi-Arid Tropics (ICRISAT), India
Niger is severely affected by land degradation which threatens the food security and resilience of much of the majority rural based and vulnerable populations. Soil degradation, in the form of hardpans (HP) and surface crusting, result from water and wind erosion process on soils made bare by overgrazing and unsustainable farming practices. Due to increased demand for food and livelihood opportunities, the reclamation of hardpans or crusted soils have become a priority and a number of successful pilots using indigenous practices led by women farmers, such as the biological reclamation of degraded lands (BDL) approach, have been demonstrated. Remote sensing based identification and mapping of hardpan or crusted soils can quantify the potential for scaling out of such approaches. Multiple spectral indices derived from sentinel 2 imagery based on hardpan characteristics were modeled using multiple regression for the Matamye department in Zinder region of Niger. Hardpan identification used relevant spectral indices from Sentinel 2 imagery: Normalized difference built-up index (NDBI), sensitive to bare soil, barren lands and urban surfaces; Normalized difference water index (NDWI), a measure of water content in vegetation canopies; Modified NDWI which enhances the detection of water bodies while suppressing noise from built-up land, vegetation and soil; Soil brightness index (SBI) which is sensitive to brightness of soil correlated with humidity and surface salt. The complementary use of these indices increased the spectral sensitivity of land cover features in the study area, classifying the hardpans into three broad categories: 1. Large unclaimed hardpans surrounded by bushy vegetation or quarried for brick making soil. 2. Hardpans or crusted soils (sandy loams) within the agricultural fields and often abandoned due to lack of investment for reclamation. 3. Red lateritic gravelly pans of wasteland, sometimes planted with bushy vegetation and indigenous trees. Ground information was collected with a hand held GPS from 700 locations along semi-arid agroecology of southern Niger representing 626 non-hardpan (NHP) locations including 174 croplands. Median of spectral index value for each hardpan and non-hardpan location was extracted to reduce the effect of extreme values. The multiple regression yielded a low r2 of 0.3 and the residual SE was 0.41 with a highly significant p value. This unique methodology is able to remotely detect hardpan areas which can be potentially reclaimed through land restoration initiatives by communities and encouraged by government and development agencies. By quantifying location, extent and potential for reclamation, such interventions can be better targeted and impacts measured.
Estimation of Soil Moisture in Bare Soils of the Northern Dry Zone of the Deccan Plateau, Karnataka, using Sentinel-1 C imagery
Hoskera, Anil Kumar;
Ahmed, Mohammed Irshad;
Whitbread, Anthony Michael - International Crops Research Institute for the Semi-Arid Tropics (ICRISAT), India
This study attempted to retrieve volumetric soil moisture (mv) from bare soils using C-band SAR imagery (5.42 GHz) acquired from sentinel-1 European satellite platform at 10 m spatial resolution and 10cm soil depth. A semi-empirical model was developed by using backscatter from the C-band SAR data and validated with in situ soil moisture, surface roughness measurements (hrms) collected from the Siruguppa taluk (sub-district) in Karnataka, India. This model overcomes the problem of overfitting, which was observed in previous studies by removing dielectric constant of soil. The backscatter coefficient σo was extracted from SAR image (VH, VV) at the 62 locations where ground sampling was carried out and volumetric soil moisture at 10cm depth measured during 2017 and 2018. Six images were acquired between March 04, 2017 to May 22, 2017 and a further seven images March 11, 2018 to May 22, 2018. The incidence angle varies from 30 to 35° covering the study area in VV and VH polarization. The relationship between σo and mv over the selected dates with co-polarized signal (VV) ranges from an r2 of 0.62 to 0.75 in 2017 explaining the high contribution of soil moisture for VV polarization. Similarly, VH polarization has lower r2 (0.47-0.6) values with σo than the VV signal relationship during the same date because of the contribution from surface roughness. The validation of the model (2017) was explained with r2 and rmse of the corresponding dates during 2018. The r2 value ranged from 0.58 to 0.79 and the rmse value was mostly 0.02 m3/m3 for all the dates except 16th April where it increased to 0.04 m3/m3. The mv in 2018 during pre-monsoon season is the validation output of the semi-empirical model developed using the data collected during 2017. The model has shown a good fit to the validation dataset collected in 2018. The r2 value between σo and mv are higher (0.51-0.71) for VV than VH polarization (0.31-0.50) during 2018. This again indicates the higher contribution of VV signal backscatter during 2018 than that of VH (which is lower than in 2017). The inter-seasonal variation in the σo values behaves almost similar for both polarization, except for the energy response indicating a bare soil condition and the importance of VV polarization along with VH in increasing the accuracy of mv estimate on bare soil. The estimation of soil moisture in bare soils will significantly help in understanding the planting dates, irrigation scheduling and planning for a short duration crop between rainy seasons and soil penetration capability.
The Mediterranean Soil Spectral Library: An Example of an Effective Way to Exchange Soil Spectral Libraries Originated from Different Sources
Ben Dor, Eyal (1);
Ogen, Yaron (1);
Tsakiridis, Nikolaos (2,4);
Tziolas, Nikolaos (3,4);
Zalidis, George (3,4) - 1: School of Earth Science Department of Exact Science Tel Aviv University;
2: Department of Electrical and Computer Engineering, Faculty of Engineering, Aristotle University of Thessaloniki, 54123 Thessaloniki, Greece;
3: School of Agriculture, Faculty of Agriculture, Forestry, and Natural Environment, Aristotle University of Thessaloniki, 54123 Thessaloniki, Greece;
4: Department of Electrical and Computer Engineering, Faculty of Engineering, Aristotle University of Thessaloniki, 54123 Thessaloniki, Greece
Soil provides ecosystem services, supports human health and habitation, stores carbon and regulates emissions of greenhouse gases. Unprecedented pressures on soil from degradation and urbanization are threatening the agro-ecological balance and food security. Soil is a complex system that is extremely variable in physical and chemical composition in space and time. Soil spectroscopy across the VIS-NIR-SWIR-TIR (0.4-14mm) spectral region is less complex than the soil body itself and hence can reduce the uncertain dimension of the soil system. Furthermore, a proxy data-mining approach to extracting spectral information that refers to soil attributes is possible and has gained much attention over the past decade in the precision agriculture discipline. A suitable data-mining algorithm enables extracting a mixture of spectral information that is hidden in the spectrum: it cannot be seen or extracted with the naked eye or even by ordinary computation. As the spectral data-mining approach is statistically driven, a large set of samples ("big data") is required to establish a precise soil spectral model for a given soil attribute. Additionally, appropriate transfer learning techniques can leverage these large spectroscopic databases to achieve calibration model transferability among laboratory sensors and Copernicus multispectral data. Accordingly, over the last 7 years, extensive effort has been devoted to developing, sharing and using soil spectral databases (also termed soil spectral libraries—SSLs) on national, continental and global scales. SSLs consist of metadata (location, geology, climate, topography, horizon, area description and location) plus chemical and physical attributes from wet laboratory analyses. Generating soil spectral libraries is a mission that is widespread over the globe by many large initiatives that combines several spectral libraries from different sources (e.g. the European SSL (Stevens et al, 2013), The World SSL (Rossel et al 2016), African SSL (Shepherd and Walsh 2002) and Brazilian SSL ( Dematte et al., 2019). Apparently those SSLs are generated using different protocols and hence merging the data from different sources provided uncertainty that can affect the proxy model afterward. Recently in CSIRO a standard and protocol procedure followed with an Internal Soil Standard (ISS) was developed and demonstrated an effective way how different protocols can still be used to harmonize measurements from different sources (Kopackova et al., 2016, Gholizadeh 2017 and Romero et al. 2018 ). Nonetheless, the CSIRO ISS protocol was never applied to a large data based composed of different countries. Recently within the GEO-CRADLE H2020 project, a Mediterranean-Balkan SSL was generated using the CSIRO ISS protocol. To that end, about 2000 soil samples from nine countries were collected. The SSL was checked for harmonization and for generating a proxy models for several soil attributes and found to hold very high stability and accuracy. The Mediterranean Balkan SSL is now under enlargement process as well as extending the spectral range to the LWIR (7-12mm) region.
Retrieval of Agricultural Topsoil Properties from Hyperspectral Data: Assessment of Spectral and Spatial Resolution Effects
Pignatti, Stefano (1);
Pascucci, Simone (1);
Pepe, Monica (2);
Casa, Raffaele (3) - 1: CNR IMAA, Rome, Italy;
2: CNR IREA, Milan, Italy;
3: DAFNE, University of Tuscia, Italy
Agricultural soils are among the most important environmental resources, not only supporting crop production, but also providing a range of ecosystem services, strongly influencing global biogeochemical cycles. A detailed, accurate and systematic estimation of topsoil properties and conditions, with an up-to-date and spatially referenced monitoring from remote sensing, from the regional to the field scale, would benefit not only the scientific community, but also the farmers and the policy/decision-makers.
Hyperspectral remote sensing in the visible–near-infrared and shortwave infrared (400–2500 nm), has been shown in the last decades to offer interesting perspectives for the quantitative prediction of key topsoil properties. In this context, the launch of current and forthcoming hyperspectral satellite sensors (e.g., PRISMA launched on march 2019, GF-5, ENMAP and CHIME - Copernicus candidate still under evaluation), is going to transform these perspectives into operational products and services, meeting also the increasing demand for global soil mapping and monitoring for a sustainable agriculture.
The objective of this communication is to show a preliminary investigation aimed at assessing the influence of spatial and spectral resolution on the capability of the current/forthcoming hyperspectral satellite missions, in terms of potential in topsoil properties retrieval in agricultural fields. An application case is presented, based on AVIRIS-NG airborne data, acquired in the framework of the Grosseto (Italy) 2018 campaign, resampled to the CHIME spectral bands with a spatial resolution of 20 and 30 m, as well as to the Sentinel-2 bands at 20 m. Block kriging was applied to soil ground data to spatialize the laboratory retrieved topsoil properties with the relative uncertainties, to the 20 or 30 m spatial resolutions. The results of spectral features analysis, based on characteristic bands or spectral features, and of two different multivariate methods (partial least square regression - PLSR and random forest - RF) were evaluated for topsoil properties estimation from AVIRIS-NG and CHIME-like datasets.
The retrieval of clay provided the best accuracy among all soil properties, for the Grosseto pivot agricultual field, as a consequence of a good range of variation of this variable. The retrival of topsoil properties was slightly better for the 30 m resolution CHIME-like dataset than for the 20 m resolution. The 20 m Sentinel-2 spectral resolution dataset provided slightly worse results than the 20 m CHIME-like data.
These preliminary results highlight the advantage of hyperspectral imagery for field-scale topsoil mapping and monitoring, as compared to multispectral data. Furthermore they seem to confirm that the spectral and spatial resolution chosen by the current/forthcoming spaceborne hyperspectral sensors would open up interesting possibilities also for field-based topsoil properties estimation.
Predicting Soil Organic Matter Content using Machine Learning Models based on Sentinel-2 Imagery
Ćirić, Vladimir I. (1);
Brdar, Sanja (2);
Lugonja, Predrag (2);
Marko, Oskar (2);
Crnojević, Vladimir (2) - 1: Faculty of Agriculture, University of Novi Sad, Serbia;
2: BioSense Institute, University of Novi Sad, Serbia
Soil organic matter (SOM) is one of the most important parameters of the soil as it is a good indicator of quality, fertility and the overall health of the soil. Additionally, SOM strongly affects carbon cycle and thus has a considerable impact on climate change. Currently, the most widely used method for assessment of SOM content is laboratory analysis of soil samples. This method is very precise, but not scalable. Also, using analytical methods based on different principles (e.g. wet oxidation, dry combustion, loss on ignition) leads to different results regarding underestimation or overestimation of SOM. Spatial interpolation based on point measurements can yield aerial maps, but they are localised to the region of the experiment. In this paper we present a method for large-scale SOM mapping based on satellite imagery.
Ground truth was collected in Vojvodina (21506 km2), an agricultural region in northern Serbia situated on the Pannonian plain. In order to build a general model, soil samples were collected from 142 locations that included most common reference soil groups in the research area (Arenosols, Chernozems, Phaeozems, Vertisols, Cambisols, Gleysols, Fluvisols, Solonetz). For the analysis of SOM in soil samples, dichromate wet oxidation method was used. We analysed three Sentinel-2 images from the past three years (2016-2019), which satisfied the following criteria: 1) minimal cloud coverage; 2) acquisition in winter 3) no crops or weed on the fields (NDVI < 0.35); 4) homogeneity of soil moisture across the fields. Even with the careful choice of images, there was a total of only 68 sampling locations for which all these criteria were met for all the analysed images. The database thus comprised of 68 samples with 36 features (3 dates x 12 spectral bands), while the output variable was SOM. The database was used for training a random forest regression model. For this, we used scikit-learn implementation of random forest in Python with 100 trees and default parameters.
To estimate the model performance we used leave-one-out cross-validation. The model achieved the coefficient of determination of R2=0.38 and Pearson correlation coefficient of r=0.62. Considering the generality of the model and the fact that it was tested on various soil types, the results are promising. Future work will include testing of separate models for different soil types, use of other machine learning models and enrichment of the database with new soil samples or new satellite images.
This is the first step in comprehensive mapping of SOM on a national level in Serbia. It will serve as a benchmark for long-term ecological monitoring of soil and an important factor in optimising fertiliser rates with the aim of maximisation of yield and minimisation of environmental footprint.
Prediction of Soil Microbial Biomass C in Italian Vineyard Soils by Artificial Neural Networks
Elisa, Pellegrini (1);
Nicola, Rovere (1);
Stefano, Zaninotti (2);
Irene, Franco (1);
Maria, De Nobili (1);
Contin, Marco (1) - 1: University of Udine, Italy;
2: Vitenova, Italy
Soil microbial biomass (MBC), the mass of living soil micro-organisms considered as a single pool, is an important soil quality indicator. MBC plays a decisive role in ecosystem functionality, being responsible for organic matter turnover, humification of organic inputs, nutrient cycling and stabilization of soil aggregates. Because of its close relationship with soil functioning and the sensitive and prompt response to anthropogenic perturbations or climate changes, MBC has been proposed as useful indicator of soil degradation in soil quality monitoring tasks.
Artificial neural networks (ANN) and fuzzy inference systems are successfully used to build predictive models that estimate parameters which maybe be lengthy or difficult and costly to measure. ANN models are employed in many fields as alternate statistical tools to linear models. They may also be used to calculate expected values for variables that are linked to other soil parameters in a complex and non linear way. Neural networks can model dynamic, non-linear, phenomena that are too complex to be described by analytical methods or empirical rules, or that relationships between cause and effects are vague, and thus be implemented for diagnosis and predictions.
The aim of this work is to achieve prediction by a self-trained artificial neural networks model of the expected value of MBC for a given soil from its chemical properties. The second and consequent aim is to select from a large spectra of parameters those which most influence MBC in order to allow selection of a minimum set of suitable parameters for soil analyses.
One linear and one non-linear regression models were run based on all 11 input variables. Outputs were compared to test the fitting ability of data using two different techniques: a multivariate linear regression analysis and a non-linear regression. The latter was carried out by using a set of feed-forward ANN which can act as black-box mathematical functions, able to build themselves up by learning from noisy data, without requiring any a priori knowledge about the physical or biological laws underlying the problem. The non-linear regression analysis was carried using feed-forward ANN approach.
The ANN method showed the better fit with data. Divergence between measured and predicted MBC was evidently restrained using the non-linear approach, testifying the ability of the ANN to adapt to the highly variable dataset. The best model minimizing the RMSRE included as input parameters: pH, organic matter, total N and Na. On the other hand, the best fitting model with the lowest RMSE included: organic matter, C/N ratio, Cu, EC and active carbonate. Errors between measured and predicted MBC are quite sizeable, being the RMSRE of the best model approx. 50%. However, differences among models were restrained both considering RMSRE or RMSE. The ANN analysis confirmed the primary importance of SOM for MBC prediction, being present in all of the ten best models with the lowest RMSRE and in 8 models (of 10) with the lowest RMSE.
Citizen Observatory based Soil Moisture Monitoring
Dobos, Endre (1);
Kovács, Károly (1);
Hemment, Drew (2);
Woods, Mel (3);
van der Velden, Naomi K. (4);
Xaver, Angelika (5);
Zappa, Luca (5);
Burton, Victoria J. (4);
Garrett, Natalie L. (6);
Giesen, Rianne H. (7);
Pelloquin, Camille (8);
Skalsky, Rastislav (9) - 1: University of Miskolc, Hungary;
2: University of Edinburgh, UK;
3: University of Dundee, UK;
4: Permaculture Association, UK;
5: TU Wien, Department of Geodesy and Geoinformation, Austria;
6: Met Office, UK;
7: HydroLogic Research, Delft, The Netherlands;
8: Starlab, Barcelona, Spain;
9: International Institute for Applied Systems Analysis, Laxenburg, Austria
Soil moisture is a critical factor for agricultural activity, but also an important factor for weather forecast and climate science. Despite of the technological development in soil moisture sensing, no full coverage global or continental or even national soil moisture monitoring system exist. There is a new initiative in Europe to develop a demonstrational system of citizen based data collection and interpretation. This approach has several limitations compared to a systematically and scientifically designed system. The aim of this study is to characterize this new monitoring approach and provide provisional results on the interpretation and system performance.
GROW Observatory is a project funded under the European Union’s Horizon 2020 research and innovation programme. Its aim is to establish a large scale (>20,000 participants), resilient and integrated ‘Citizen Observatory’ (CO) and community for environmental monitoring that is self-sustaining beyond the life of the project. This article describes the how the initial framework and tools were developed to evolve, bring together and train such a community; raising interest, engaging participants, and educating to support reliable observations, measurements and documentation, and considerations with a special focus on the reliability of the resulting dataset for scientific purposes. The scientific purposes of GROW observatory are to test the data quality and the spatial representativity of a citizen engagement driven spatial distribution as reliably inputs for soil moisture monitoring and to create timely series of gridded soil moisture products based on citizens’ observations using low cost soil moisture sensors, and to provide an extensive dataset of in-situ soil moisture observations which can serve as a reference to validate satellite-based soil moisture products and support the Copernicus in-situ component. This article aims to showcase the initial steps of setting up such a monitoring network that have been reached at the mid-way point of the project’s funded period, focusing mainly on the design and development of the CO monitoring network.
The GROW Observatory (GROW) will create a sustainable citizen platform and community to generate, share and utilise information on land, soil and water resources at a resolution hitherto not previously considered. The European Space Agency’s Sentinel‐1 is the first mission capable of providing high‐resolution soil moisture information, but a proper validation of Sentinel data remains a challenge given the scarcity of available in situ reference measurements. Establishment of a dense network of in situ measurement can bridge the gap in spatial resolution of between in situ and satellite‐based soil moisture measurements enabling validation between ground and remotely measured soil moisture observations. The potential exists to answer scientific questions including the validity of satellite data, the impact of climate change on land management thus supporting the needs of growers and integrating citizen and scientific research to be more directly applicable and relevant.
RS Based Soil Diagnostics Mapping and Use for Soil Property Estimation
Dobos, Endre (1);
Vadnai, Péter (1);
Kovács, Károly (1);
Micheli, Erika (2) - 1: University of Miskolc, Hungary;
2: Szent István University, Hungary
Traditional soil maps present soil information in the form of categorical classes of soil types classified on the appropriate level of the applied classification system corresponding to the scale. Soil complexes and associations have been used to describe polygons. This kind of data structure is useful to characterise an area by explaining its soil resources. However, it is difficult to convert these complex categorical units into a simple digital variable, the usage of this kind of data in a digital environment is limited. Modellers and users often need certain properties instead of the complex classes.
Additionally, the problem becomes more complicated when soil information of different origin, based on different classification systems has to be integrated into a common, harmonised database.
The presented methodology is part of the efforts to develop a global SOTER (World Soil and Terrain database) coverage and contribute to the Global Soil Observing System. The aim is to determine and map the relevant soil properties, horizons and materials following the diagnostic concepts of the World Reference Base (WRB) for soil resources and derive the occurrence probability of soil classes (WRB reference soil groups) of certain spots with the application of remote sensing and digital soil mapping tools. The developed method is referred as the e-SOTER approach and is capable of producing a stack of soil diagnostic element layers with the likelihood of their occurrence within each pixel and a layer of WRB reference soil groups (RSG). This new approach may provide better input for digital modellers and predict the spatial continuum of the soil cover in a much better resolution than the traditional polygon based approaches. At the same time the diagnostic elements, as building blocks of the classification systems help the correlation of the national soil classes into integrated databases and maps.
FLOWERED: Crowdsourcing and Copernicus Data for Soil Contamination from Fluoride in African Rift Valley
La Mantia, Claudio (1);
Melis, Maria Teresa (2);
Deflorio, Anna Maria (1);
Drimaco, Daniela (1) - 1: Planetek Italia s.r.l., Italy;
2: University of Cagliari
This study is part of the EU H2020 research Project FLOWERED (de-FLuoridation technologies for imprOving quality of WatEr and agRo-animal products along the East African Rift Valley in the context of aDaptation to climate change). FLOWERED project, currently ongoing, aims to develop technologies and methodologies at cross-boundary catchment scales to manage the risks associated with high Fluoride water supply in Africa, focusing on three representative test areas along the African Rift Valley (i.e. Ethiopia, Kenya and Tanzania), characterized by high fluoride contents in waters and soils, water scarcity, overexploitation of groundwater and high vulnerability to risks arising from climate change, as drought and desertification.
It also is empowering local communities to take responsibility for the integrated-sustainability of the natural resources, growing national and international environmental priorities, enhancing transboundary cooperation and promoting local ownership based on a scientific and technological approach.
Fluoride is one of the hazardous geogenic contaminant. Fluoride-related health disease are a common issue in the Eastern Rift Valley where, the high concentration of F in soils and drinking waters is related to the alkaline volcanic activity of the Rift Valley. Tests of innovative mitigation options for Fluoride contamination in agriculture have been carried out through trials in greenhouse, lab and on the field during FLOWERED project. A strategy for a proper monitoring and better interpreting the fluoride contamination phenomenon implies the collection of multiple local geo-information on land use, water uses, irrigation systems, household features and the other information needful for the specific knowledge of water supply.
The approach proposed in FLOWERED is based on a detailed knowledge of the hydrogeological setting, with the identification and mapping of the specific geological conditions of water contamination and its relation with the different land uses (Da Pelo et al., 2017). In the framework of the project, the development of a land cover mapping system has represented a primary key for understanding and assessing of the fluoride pollution. A land cover mapping system, based on Copernicus Sentinel-2 data, has be implemented with a focus on the ground effects caused by pollution due to the high concentration of Fluoride in water.
Within the FLOWERED project, a mobile application (FLOWERED-GeoDBapp) has been also developed with the aim to collect multiple geo-information through an action of crowd-generating data by local communities (students and people involved mainly by NGOs). These data have been used to enhance and strength the information on land cover extracted from satellite data and field activities mapping. The SHAREGEODBapp is proposed as an innovative tool for water management and agriculture institutions at regional and local level.
Soil Moisture Monitoring with Spire’s Planned Constellation of GNSS-R CubeSats
Freeman, Vahid;
Masters, Dallas;
Mano, Fabio - Spire Global Inc., Luxembourg
Soil moisture is an essential climate variable influencing the physical, chemical and biological properties of the soil. Monitoring of soil moisture is extremely valuable for understanding the complex processes in the soil and Earth system dynamics. Traditional undistributed field-sampling fails to obtain the accurate status of the soil water content at large scale as the soil moisture rapidly changes due to high spatial variability of water distribution and its direct relationship with air and soil temperature and precipitation.
In the last years, several active and passive remote sensing techniques have been developed and used successfully in space-borne missions for monitoring of soil moisture dynamics. The soil moisture data obtained by microwave remote sensing methods revealed to be very useful monitoring tools at a global scale, although the measurements may differ not only by methods and sensor limitations but also due to high spatio-temporal variability of the soil surface water content. Still, there is a need to improve data quality and spatial and temporal resolution of the soil moisture measurements to be used more efficiently in hydrological and meteorological models and applications. One of the recently developed techniques for soil moisture measurement is using the reflected signal from Global Navigation Satellite Systems (GNSS). The GNSS‑R (GNSS-Reflectometry) technique, as opposed to conventional backscatter measuring radars and radiometers, makes use of the forward scattering of the transmitted energy, and therefore it is less sensitive to surface roughness and vegetation. This has been demonstrated in several studies using the data from recent space-borne GNSS-R missions, such as TechDemoSat-1 and NASA’s CYGNSS (Cyclone Global Navigation Satellite System).
Beginning in 2019, Spire Global is planning to launch a large constellation of GNSS-R Earth observation (EO) satellites for soil moisture monitoring. The data acquired by Spire’s GNSS-R satellites will provide unprecedented sub-daily global coverage at sub-kilometer spatial resolution. Such intensive data acquisition of soil moisture is of great importance to many hydro-meteorological and agricultural applications, and will be initially used to improve the Spire’s weather forecasting model. We will present the latest developments of GNSS‑R technique for measuring of soil moisture and introduce the Spire EO mission for land applications.
Spatial Models for Topsoil Clay Content Predictions at National Scale – Evaluating Different Predictor Datasets
Gebauer, Anika (1);
Sakhaee, Ali (2);
Don, Axel (2);
Ließ, Mareike (1) - 1: Helmholtz-Centre for Environmental Research - UFZ, Germany;
2: Thünen Institute of Climate Smart Agriculture, Germany
High precision spatial soil information is important for the application of landscape-scale process models. The German Agricultural Soil Inventory provides new opportunities for soil data regionalization by offering a harmonized, reliable database including soil pedon information on soil organic carbon, pH, bulk density and soil texture obtained by systematic sampling on agricultural soils throughout Germany. Machine learning algorithms can be used for spatial predictions by relating the sampled soil properties to predictors approximating different soil forming factors. These models are able to operate on predictor interaction as well as nonlinear predictor-response relations. Model performance highly depends on the predictor variables. These are derived from different data sources and differ concerning data processing as well as quality assessment.
In order to develop models for high precision clay content predictions at landscape scale we evaluated predictors representing six different soil forming factors. Six boosted regression tree models were built by iteratively adding predictor datasets to the model input: (1) TanDEM-x based topography predictors (2) information on parent material and (3) soil systematic units, both based on classical mapping (4) remote sensing based land cover classification, (5) regionalized climate data and (6) geographic position. A nested k-fold cross-validation approach was applied for model evaluation and tuning, tuning parameters were selected by differential evolution optimization.
Results show that the more predictors are integrated the better the model performance. Models based on all available predictors explained up to 56 % of the clay content’s variance. Performance was improved the most after adding the parent material as predictor to the model input: on average R² was increased by 27 %, while the mean RMSE value was decreased by 10 %. For all models, elevation was identified as the most important predictor. Further remote sensing data products shall be incorporated to improve the predictive performance of future spatial models.
Spectroscopy, Spectral Imaging and Colour Features for Soil Organic Carbon Estimation under Visible Spectrum
Gholizadeh, Asa (1);
Saberioon, Mohammadmehdi (2);
Viscarra Rossel, Raphael A. (3);
Boruvka, Lubos (1);
Klement, Ales (1) - 1: Czech University of Life Sciences Prague;
2: University of South Bohemia in Ceske Budejovice;
3: Curtin University
Effective measurement and management of soil organic carbon (SOC) are essential for management of carbon cycle, ecosystem function and food production. SOC has an important influence on soil properties and soil quality. Conventional SOC analysis is expensive, time-consuming and difficult. The development of spectral imaging sensors enables the acquisition of larger amount of data using a cheaper and faster method. In addition, satellite remote sensing offers the potential for performing surveys more frequently and over larger areas. This research aims to measure SOC content with colour as an indirect proxy. The measurements of soil colour were made at an agricultural site of the Czech Republic with an inexpensive digital camera and the Sentinel-2 remote sensor. Various soil colour spaces and colour indices derived from the (i) reflectance spectroscopy in the selected wavelengths of the visible (VIS) range (400˗700 nm), (ii) RGB digital camera and (iii) Sentinel-2 visible bands were used to train models for prediction of SOC. For the modelling, we used the machine learning method, random forest (RF), and the models were validated with repeated 5-fold cross-validation. For prediction of SOC, the digital camera produced R2 = 0.85 and RMSEp = 0.11, which were higher R2 and almost similar RMSEp compared to those obtained from the spectroscopy technique (R2 = 0.78 and RMSEp = 0.09). Sentinel-2 predicted SOC with lower accuracy than other techniques; however, the results were still fair (R2 = 0.67 and RMSEp = 0.12) and comparable with other methods. Colour measured with a digital camera enabled accurate and reliable predictions of SOC, overcoming limitations of more traditional laboratory methods of SOC analysis.
Healthy Soils and Land Management in a Changing Climate
Gobin, Anne;
Delalieux, Stephanie;
Deronde, Bart;
Dong, Qinghan;
Eerens, Herman;
Gilliams, Sven;
Haesen, Dominique;
Piccard, Isabelle;
Raymaekers, Dries;
Smets, Bruno;
Tits, Laurent;
Van Hoolst, Roel;
Van Tricht, Kristof - Flemish Institute for Technological Research (VITO), Belgium
Soils matter! Soils deliver ecosystem services that enable life on earth. Soil functions are inherent soil capabilities that include biomass and food production, maintaining soil biodiversity, carbon and nutrient sequestration, water filtration and transformation, landscape and heritage, and source of raw materials. Soils are an important carbon stock: more than twice as much carbon is held in soils as compared to the storage in vegetation or the atmosphere. These multiple functions benefit society, influence the management of other natural resources and are vital to the performance of the earth system. Therefore soils are key to achieving many of the sustainable development goals through for example SDG2-sustainable soil productivity; SDG3-closing nutrient cycles; SDG6-water filtering; SDG7-bio-energy; SDG11-soil sealing; SDG13-GHG sequestration and buffering; SDG12-soil quality; and, SDG15-degradation neutrality. A global soil partnership to achieve these goals (SDG16) is of paramount importance.
Soil health is a pressing global issue that sits at the heart of three UN conventions on biodiversity, desertification and climate change (UNCBD, UNCCD, UNFCCC). The Action Programmes defined by the UN conventions identify the decline of soil health worldwide as an environmental risk that undermines not only soil fertility and productivity and hence food security, but also the progressive stabilisation and subsequent reduction of atmospheric GHG concentration levels. Sustainable land management and changes therein in agriculture, forestry, grassland and peatlands greatly influence soil health.
Soils are a mixture of minerals, organic matter, gases, liquids and a myriad of micro- and macro- organisms. Temperature, moisture and evapotranspiration influence different soil properties through their effects on the balance between plant growth and microbial decomposition. Many of the soil properties, influencing processes and forming factors can be monitored by different sensors. The Sentinel satellites of the European Copernicus Programme offer advances in monitoring the earth surface at unprecedented spatial, temporal and spectral resolutions. Data from proximal sensing provide rapid and inexpensive data collection at fine spatial and temporal resolutions for a better understanding of the spatio-temporal variation of soil properties. Remote sensing-derived information and in particular the combination of Sentinel-1 and Sentinel-2 offers possibilities to derive objective and harmonized indicators across the globe. Examples of monitoring soil properties, processes and functions will be drawn from different on-going projects at VITO’s Remote Sensing Unit:
In conclusion, the business case for investing in soils is diverse ranging from increasing direct revenues to avoiding costs, and serves multiple purposes. Proximal and remote sensing offers valuable opportunities for monitoring soil (-related) information.
Soil Classification Techniques in Transylvania Area Based on Satellite Data
Gorgan, Dorian (1);
Rusu, Teodor (2);
Bacu, Victor (1);
Stefanut, Teodor (1);
Nandra, Constantin (1) - 1: Technical University of Cluj-Napoca, Romania;
2: University of Agricultural Sciences and Veterinary Medicine Cluj-Napoca, Romania
An efficient technical management in agriculture is based on a good, reach and accurate information on the environment and especially on the soil. The need for up-to-date and high-resolution soil information and direct access to this information in a flexible and simple manner is imperative for pedology and agriculture specialists, farmers, soil monitoring and land management organizations, pedology maps developers, Earth Observation and Earth Science oriented software development companies, sustainable development, and universities. The satellite data open the possibility of developing solutions for high resolution (e.g. 10m by Sentinel-2) and large geographical areas. The satellite data processing and interpretation is very particular to each area, time, season, and context, and they need to be calibrated by real field measurements that are collected periodically, though quite rarely and by high costs.
This presentation concerns on soil classification approach based on spectral signature from satellite data as a footprint of the soil, and machine learning techniques. The spectroscopic analysis can provide a different perspective to understand the soil genesis and the soil classification as a supplement to the traditional viewpoint of soil properties.
The HorusApp application supports the development of pedological maps by combining field studies and measurements with Sentinel-2 data. The application is layered on the HORUS platform which enables specialists to scale the processing over a cloud infrastructure. The platform integrates ESA’s SNAP software tool in order to process remotely the sensing data. The development of the HORUS platform and HorrusApp application have been supported and funded by the Romanian Space Agency (ROSA). Some examples from Transylvania county are highlighted by analysing the results, issues and perspectives of such an approach.
LUCAS - The Truth On Soil?
Jones, Arwyn;
Fernandez Ugalde, Oihane;
Orgiazzi, Alberto - European Commission Joint Research Centre
Soil is having increasing policy relevance. The reformed Common Agricultural Policy has soil condition at it's core. Many of the targets associated with the Sustainable Development Goals are directly or indirectly linked to soil functions and associated ecosystem services. In addition, soil condition is critical to successful bio and circular economies. However, soil is under pressure from competeing demands on land and unsustainable or inappropriate management practices. Combatting pressures on soil to achive land degradation neutrality and combat desertification is a key political target for the coming decade (for example, the Soil Health Mission of Horizon Europe).
Presentation will focus on how repeated soil data collected through the European Commission's LUCAS monitoring programme from over 20,000 locations across the European Union can be used to improve the quality of earth observation products that aim to assess soil condition and key policy parameters such as soil organic carbon content, and other ecosystem services. Presentation will highlight the need for the development of soil-focused use remote sensing techniques (including proximal sensing), which to date have not been widely used due to the inherent nature of soil and technological limitations of sensors. The use of the LUCAS land cover and land use components will also be highlighted.
Visible Near-infrared Spectroscopy and Water Vapor Sorption for Soil Specific Surface Area Estimation
Knadel, Maria Augusta (1);
de Jonge, Lis Wollesen (1);
Tuller, Markus (2);
Rehman, Hafeez Ur (1);
Jensen, Peter Weber (1);
Moldrup, Per (3);
Greve, Mogens Humlekrog (1);
Arthur, Emmanuel (1) - 1: Aarhus University, Dept. of Agroecology, Denmark;
2: The University of Arizona, Dept. of Soil, Water and Environmental Sciences, USA;
3: Aalborg University, Dept. of Civil Engineering, Denmark
The soil specific surface area (SSA) is a fundamental property that governs a wide range of soil processes and behaviours relevant for numerous engineering, environmental, and agricultural applications. Capitalizing on the excellent reproducibility and rapidity of spectroscopic and vapor sorption measurements, we propose a method for SSA determination based on a combination of proximal visible near-infrared spectroscopy (vis–NIRS) and vapor sorption isotherm measurements. Two models for water vapor sorption isotherms (WSI) were used: the Tuller–Or; TO, model, and the Guggenheim–Anderson–de Boer; GAB, model. They were parameterized with sorption isotherm measurements and applied for SSA estimation for a wide range of soil types (N=270) from 27 countries. The generated vis–NIRS models were further compared with models where the SSA was determined with the ethylene glycol monoethyl ether (EGME) method. Moreover, three types of regression techniques including machine-learning methods (partial least squares; PLS, support vector machines; SVM and artificial neural networks; ANN) were tested and compared. The performance of all SSA models was mainly dependent on the range and variation in SSA values, however, an independent validation indicated very good and nearly identical estimation capabilities for SSATO, SSAGAB, and SSAEGME, with an average standardised root mean square error (SRMSE= RMSE/range) of 0.05, 0.06 and 0.05, respectively. In general, the machine-learning techniques (especially SVM) performed better than PLS regression. The results of this study indicate that the combination of vis–NIRS with the WSI as a reference technique for training vis–NIRS models can provide SSA estimations similar to the EGME method. Further, the application of hyperspectral images to determine SSA should be investigated to easily facilitate a source of data supporting bigger scale engineering, environmental, and agricultural applications.
Steps Toward Implementing INSPIRE Rules on Soil Data Specification, some issues
L'Abate, Giovanni;
Marchetti, Alessandro;
Barbetti, Roberto;
Pepe, Antonio Gerardo - Consiglio per la ricerca in agricoltura e l’analisi dell’economia agraria, Italy
Since 1997 Consiglio per la ricerca in agricoltura e l’analisi dell’economia agrarian (CREA-AA) is storing soil data and, on 2007, published ”Linee guida dei metodi di rilevamento e informatizzazione dei dati pedologici” (Guidelines of the methods for soil survey and data informatization) a volume with an attached CD-rom software (CNCP 3.0, Database for soil observations and pedological units storing, correlating and geoexploring). The software was adopted as soil information system by several regional soil services in Italy and could be considered the Italian ”de facto” standard for soil database architecture. After the software publication, standards to describe soil properties have been deeply defined, with several ISO specifications and international thesauri available for specific applications. The European directive on “Infrastructure for Spatial Information in the European Community (INSPIRE)” has brought together existing standards into a definite model. Since INSPIRE Data Specification on Soil – Draft Technical Guidelines (D2.8.III.3 ) publication, on 2013, few works have been published facing the aim of implementing rules in a physical geodatabase. Up to date, according the INSPIRE portal results no real implementations for the INSPIRE Theme Soil, made by and for implementers. In this document, both conceptual UML schema and XML mapping have been widely defined by the Thematic Working Group SOIL (TWG-SO) but structured relational databases developed until 2013 by different institutions in Europe often differ from the INSPIRE soil schema both for granulation and relates among different features. The published Enterprise Architect project allow to export the schema to an ArcGIS workspace environment . Accordingly, ESRI published in 2014 (and updated in 2016) several Geodatabase Templates to be used with the ArcGIS for INSPIRE tool. Themes that have been implemented by ESRI were, INSPIRE Annex I; INSPIRE Annex II: Land Cover and Geology; INSPIRE Annex III: Land use, Mineral Resources, Statistical units. It was not published a template for the Soil theme. After TWG-SO guidelines publication, there have been actions in mapping existent soil database to the INSPIRE schema and few structured databases have been developed according such a schema. Most of this works shown an incorrect comprehension of the UML conceptual model and Schema definition. Some authors seem not to comprehend the difference between some concepts or how storing analytical parameters accordingly the ISO19156 Observations and Measurements standard.
Both CREA volume and software require now a critical revision enlighten by the TWG-SO work. Besides the need to migrate to a new, INSPIRE based soil database structure, expressed by several pedologists collecting soil data within this institution.
The present contribution has the aim to clarify some issues on mapping a soil database based on the previously cited guidelines. Authors present a critical review of the few papers that cited the TWO-SO guidelines and present the actual state of the art work to map the database architecture to the INSPIRE schema. The mapping is presented suggesting the followed steps. The aim of implementing guidelines and schema into a structured geodatabase template was not jet reached by the working group, but critical reasons are commented.
Distributed Ledger Technology in the Analysis of Earth Data and Its Use in Decision-Making for Sustainable Food Systems
Chunggaze, Mohammed (1);
Leveille, Genevieve (1,2);
Risi, Amanda (1) - 1: AgriLedger, United Kingdom;
2: techUK, United Kingdom
Idea Description:
Adapting to the Earth becoming warmer every year is required. The continued emissions of greenhouse gases and the effects of global warming cannot be forecasted with accuracy or perfectly mitigated[1]. As such, figuring out how localised climates and the economies linked to them are impacted is an enormous challenge.
Emerging digital ledger technologies (DLT), demonstrated in projects taking place in micro economies, are now able to inform decision-making and planning as the provision of access to relevant and localised information is improved.
With a focus on solving the problem of monitoring rapidly developing events and providing accurate earth surface information such as crop data crucial to farmers and policy makers, the Copernicus Sentinel-3 now has two sensors in orbit[2]. The Copernicus Sentinel 3, now with 3A and 3B in tandem, has in addition ensured the recording of quality measurements over inland waters, with 32,500 virtual stations that define lakes, reservoirs, rivers and glaciers worldwide[3]. Sentinel 3A is also the only sensor in space able to detect cyanobacteria and map in near real-time the presence of this serious threat to human and animal health [4].
By presenting the changes reflected in this type of data map in near real-time in specific localities directly to local farmers and policy makers based on DLT, decisions on farming techniques and adaptation to the changing conditions can be made. These contributions, together with data gathered and recorded by farmers, citizen scientists, distributors, logistics providers and customers, make it possible to find smarter ways of managing our food systems, test and implement far sooner[5].
Objectives:
Role of the ESA and AgriLedger:
Through the implementation of AgriLedger’s solutions starting with Haiti, making use of citizens mobilised and connected to the ESA network of earth observers, there is an opportunity to deliver insights based on high-quality and analysis-ready data, making better decisions about local and global food systems.
The long-term opportunity is the ability to respond to climate change challenges as these arise, with data delivered through AgriLedger’s applications in near real-time. This will enable the support of global supply chain stakeholders in embedding the knowledge system required for effective ongoing deployment of the AgriLedger solutions.
Impact and benefits of this research:
Establishing a reliable, distributed system that directly supports a diversified community of problem solvers[5] including food producers and supply chain stakeholders with useful data sets from ESA satellites and localised ground-truth in near real-time while linking this to better decision-making systems (human, AI & ML) can both transform the limitations of current food traceability and monitoring systems and provide a forum for ongoing R&D that improves response times in our adaptation to changing climate conditions.
Supported by an immutable and trustworthy distributed-ledger-based record, each item is traceable and can be connected to specific micro-climate data influencing its journey from farm to table, along with production data. End buyers are thus able to make better decisions regarding purchase within commodities markets.
TerraVisie: An Operational Soil Displacement Monitoring Service Based On Satellite Data
Loenen, Edo (1);
van Haver, Sven (2);
van Zwieten, Jeroen (3);
van der Auweraert, Jasper (3) - 1: Science [&] Technology, The Netherlands;
2: Orbital Eye, The Netherlands;
3: Sobolt, The Netherlands
TerraVisie is a monitoring service for the detection of digging operations, large-scale soil displacement and / or soil improvement for specific areas of interest such as: construction sites, soil deposits, waste processors, etc. Because such activities can involve contaminated soil and / or secondary building materials, government supervision is needed to prevent improper storage, incorrect processing, incorrect application or malpractice.
Visual inspection is time consuming and given the large number of activities that take place daily the chance of detecting environmental infractions is small. For example, in the Netherlands there are 500,000 digging movements per year, of which approximately 100,000 involve contaminated soil.
Earth observation satellites can be used to increase inspection efficiency, and allow authorities to get a more complete and time-accurate picture of the soil displacements around the areas of interest (AOIs). For efficient monitoring and inspecting a large number of AOIs using earth observation data (EO data) an automated detection system is required.
TerraVisie includes an automated detection system that uses radar satellite imagery and filtering algorithms to recognize relevant changes. Detected changes are visualized on a map and supplemented with additional information. Image interpretation techniques are used to determine the severity of the excavation work. For each notification a recent optical image is automatically retrieved, which is used for further classification of the detected event (urgency), and an estimate can be made of the amount and type of soil that is displaced.
In our presentation we will present the overall TerraVisie system, discribe the techniques used for the detection and classification of events, and showcase a number of illustrative results.
Estimation of Different Carbon and Nitrogen Cycles Parameters at the Valencia Anchor Station
Lopez-Baeza, Ernesto (1);
Lidon Cerezuela, Antonio (2);
Bautista Carrascosa, Inmaculada (2);
Lull Noguera, Cristina (2);
Albero Peralta, Erika (1) - 1: University of Valencia, Faculty of Physics, Spain;
2: Polytechnic University of Valencia, Soil Science Research Lab, Spain
Carbon and nitrogen cycles are the two most significant biogeochemical cycles in the terrestrial system, particularly in agricultural environments, since they represent a relevant positive contribution to different environmental aspects. The carbon footprint concept is today a significant tool that is being incorporated in many European countries in the agricultural and environmental management policies. On the one hand, the accurate estimates of different carbon fluxes at plot level permits to compute the carbon balance and an estimation of the agroecosystem carbon footprint in a reliable way. On the other hand, nitrogen is a major nutrient, which directly affects plant performance with a strong link between nitrogen content and photosynthetic activity. In agricultural systems, the purpose of fertilisation is to provide enough nitrogen to the plant to be able to obtain an adequate production. Nitrogen excess may cause different problems to the plant and affect the environment with undesirable issues such as water pollution or gaseous emissions. Sustainable agricultural production with reduced emissions and other sustainable environmental impacts is a priority nowadays. Therefore, optimum nitrogen fertilisation requires a correct assessment of soil or crop nitrogen status. Since vegetation plays a key role in the global energy balance, carbon and nitrogen cycles, and water budget of the Earth, it is crucial to finely characterise vegetation with biophysical parameters such as FAPAR (fraction of absorbed photosynthetically active radiation), LAI (leaf area index) and chlorophyll content. In this sense, the combination of the physiological approaches and the remote-sensing observations allows to extrapolate these results to other large crop conditions and other land uses. In this work, as an example, we present the results of the validation of FAPAR and chlorophyll content over an OLCI pixel (300 m x 300 m) on a vineyard field at the Valencia Anchor Station, well known Earth Observation validation site in the Valencia region, Spain and the proposal for estimating the different components of the carbon and nitrogen cycles. This approach includes: soil respiration rate, net assimilation CO2rates, water vapour and CO2fluxes between the canopy cover and the atmosphere, soil organic matter, soil nitrogen content, leaf nitrogen content and nitrate concentration in the sap, parameters that are certainly significant in the framework of the development of space-based EO tools for mapping and monitoring soils.
Remote Sensing and Sustainable Management of Soc in the Sahelian Area
Loum, Macoumba - National institute of pedology, senegal
Spatial characterization of soil variability at a fine scale is a real need in the developing country where details information on chemical and physical soil properties are not often available.
The objective of this study was to evaluate the SOC mapping by testing the effectiveness to include the Sentinel 2 remote sensed data in the characterization of the variability of the soil properties. Ordinary kriging applied under ArcGIS is compared with multiple linear regression calibrated under R software. The results of the study, carried out in a Sahelian region of Senegal, showed a slight decrease of the root mean square error ranging from 0.18 with kriging to 0.16 for multiple linear regression. Carbon variability was also detailed scale with multiple linear regression at the pixel scale from 10 to 20 m. Spectral bands situated in the visible wavelength, NDWI and NDVI were the most discriminating explanatory variables in the spatial modeling of organic carbon by multiple linear regression. Specific locations that require inputs of manure or compost were also geo-localized with multiple linear regression in order to ensure sustainable management of soil organic carbon. The use of remote sensed data also puts into perspective the possibility of spatializing the physical and chemical properties of the soil on larger areas and correcting the lack of information on soil mapping in the Sahelian regions of Senegal.
Hyperspectral Remote Sensing Soil Heavy Metals Inversion based on Collected Regressions on Encoded Land-Use Features
Ma, Lei (1);
Zhu, Yu (2);
Li, Feimo (1) - 1: Chinese academy of science, institute of automation, China, People's Republic of;
2: Beijing Institute of Spacecraft System Engineering， China Academy of Space Technology
As an important carrier of human survival, soil provides nutrients and water for crops, and is another important medium for root extension and retention of crops. With the intensification of mining and mining enterprises, the increase of mining waste gas and waste-water irrigation, heavy metal pollution in the surface soil of mining areas has become an environmental hot spot, so rapid and efficient soil heavy metal pollution range monitoring becomes a critical prerequisite for mining risks assessment. Hyperspectral imaging based soil heavy metal contamination range monitoring is based on the utilization of spectral response of electronic transitions and molecular vibrations of heavy metals in soil, where every typical kind of heavy metal element has its own unique pattern in spectral reflection, which can be captured by the hyperspectral imaging instrument with its high spectral definition. Besides, as its unlimited observation spatial scope, safe and reliable, satellite based soil composition monitoring has grown to be a principal technical means for inversion of heavy metal pollution in mining areas.
Compared with multi-spectral images, hyperspectral images have higher spectral resolution and higher detection sensitivities for detailed soil elements analysis. Yet the additional information also amounts to the complexity of hyperspectral data analysis, alongside the introduction of large quantity of noise and interference, which adds up to the difficulties in algorithm development, makes it hard to capture the clues of the possible excession of poisonous heavy metal contamination analytically.
To address these issues, we propose a novel soil heavy metal detection algorithm based on the improvement of classical logistic regression method, with enhanced feature encoding by the help of state of art land use classification principle. Specifically, the raw primitively preprocessed hyperspectral data is fed to the encoding module of a land-use classifier. As to facilitate the extraction of spectral patterns on the target heavy metal elements, the land-use classifier is weakly supervised with a limited collection of prepared hyperspectral dataset which contains land types in weak association with the target soil region, so that better decomposing outcomes of the spectral data can be expected. After that, the decomposed features are fed to a collection of regressors built with appointed connection with the target heavy metal pollutants. With the help of deep network module encoding, experimental results turn out to be more accurate in the inversion of soil heavy metals, which stands as an important reference value for mining area soil heavy metals monitoring.
“Crop Sustainability System: The Preservation of Soil in Agriculture”
Mc Donnell, David - Anuland Limited, Ireland
Ireland is comprised of an area of 6.9 million hectares, of which 4.5 million hectares of soil is used for agriculture. There are approximately 126,000 farms in Ireland, the vast majority of which are family owned livestock farms, with an average size of 32.4 hectares.
The Intergovernmental Panel on Climate Change (IPCC) report, released in 2018, highlighted the need to push agriculture towards a more sustainable and tech-oriented future, with the inferred responsibility resting on the farmer.
Ireland is particularly suitable as a test country in grassland pasture-based trails due to its climatic conditions. It is the seventh largest milk producer in the EU, with production representing almost 5% of EU total production amounting to 170.1 million tonnes of raw milk. Ireland is also responsible for exporting dairy products to 140 countries around the globe with a value exceeding €4.6 billion.
Despite the robustness of the dairy sector, land and soil quality are limiting factors. The complexity of this problem coupled with limited available technologies, has led to growing concerns about the need to increase crop productivity without causing detrimental environmental harm.
In light of this Anuland, as an Ag-Tech start-up company, focused its primary research on best available techniques, in sensor technologies for enhanced crop management decisions. We implemented a patent pending system, it focused its research on the integration of in-situ above and below ground, physical property sensors and imagery devices. The research that we conducted allowed us to send data from infield sensors to our cloud server, using Artificial Intelligent techniques and deep neural networking, to extrapolate crop management decisions for the farmer.
The focal point of which is to maximise nutrient efficiency while optimising crop growth rates and limiting any negative environmental impact. Our aim is to empower farmers to take this responsibility upon themselves through the education and management capability that our system provides them.
Sampling Design Optimization Using DEM And Sentinel 2 Spectral Data For Precision Agriculture
Minařík, Robert;
Žížala, Daniel - Research Institute for Soil and Water Conservation, Czech Republic
Current geoinformation technologies, remote sensing and sensor measurements provides a lot of useful data such as digital terrain models and satellite images from which the variability of soil properties can be indirectly derived. This spatial variability can be used for a sampling design optimization. The optimization is most often performed for purpose of minimization of the sampling costs without losing part of the variability of the parameter being monitored.
This contribution presents the optimization workflow applying fuzzy c-means clustering algorithm on available supporting data (digital terrain model and its derivatives, Sentinel 2 satellite images) to build soil clusters with similar soil properties. The sampling points are then selected by conditioned Latin Hypercube sampling method in the feature space of membership values of pixels. It also includes automatic determination of optimal number of classes using cluster validity indices (Partition Entropy, Partition Coefficient, Modified Partition Coefficient and Fuzzy Silhouette Index). The whole process is automated in R. This workflow can be applied to previous sampling networks or to new unsampled areas. Users define only the number of sampling points. The effectiveness of the proposed method is validated on pilot farmland area (290 ha) in South Moravia, Czechia. The study area is covered by chernozems at different stages of degradation due to the variability of the parent substrate and the long-term effects of the erosion of accumulation processes.
The results of the case study showed that the optimized sampling network described the variability of the population with half the number of samples compared to random sampling. The method is also applicable to reducing optimization of existing networks. In the case study, the original number of samples was reduced by 50% without significant loss of information on the variability of soil properties.
Supported by the Ministry of Agriculture of the Czech Republic, institutional support MZE-RO2018. Supported by the Ministry of Agriculture of the Czech Republic, project no. QJ1610289 “Efficient use of soil productivity by site specific crop management.”
Earth Observation for Soil Mapping and Monitoring in The Netherlands
Mulder, Vera Leatitia (1);
van Orsouw, Tijn L. (1);
van Egmond, Fenny (2);
Lips, Frans (3);
Okx, Joop (2) - 1: Soil Geography and Landscape group, Wageningen University, The Netherlands;
2: Wageningen Environmental Research, Wageningen, The Netherlands;
3: Ministerie van Landbouw, Natuur en Voedselkwaliteit, The Netherlands
Smart and sustainable land and water management is crucial to secure the well-being of future generations, especially in highly populated areas like the Netherlands, where many different functions are combined in a small area. Accurate and up-to-date soil data at high resolution are required to allow decision makers to make the right choices. Consequently, the demand for accurate, up to date and high-resolution soil information increased. This includes reliable baseline soil information that can subsequently be used for deriving thematic soil information and monitoring of (changes in) soil properties. This is used for reporting on and supporting initiatives at the national (Climate Envelop), European (Common Agricultural Policy) and global level (IPCC, Sustainable Development Goals, UNCCD and the Global Soil Partnership).
At the moment, we have several research projects ongoing that aim to better satisfy the user needs. They focus on developing and applying methods 1) allowing continuous and affordable updating of soil information and the soil condition, i.e. generating the baseline and thematic information and 2) monitoring of soil properties and soil conditions. Examples include soil carbon, soil compaction and soil degradation for evaluating the implementation of e.g. different agricultural soil management strategies, at the farm/regional/national scales.
At the national scale, the generation of high-resolution soil information heavily relies on available earth observation data. That is, the prediction models typically implemented are machine learning techniques using optical and radar satellite data or derived products as complementing auxiliary data. Copernicus has been very useful for obtaining this data. At the farm and regional level, airborne and proximal soil sensing with geophysical sensors and VNIR spectroscopy demonstrated to be relevant for developing methods for soil monitoring. Therefore, focus is now on the integration of the latter with current and upcoming satellite products for upscaling to the regional/national level (i.e. multi-scale, multi-platform monitoring strategy). Hereby, the development of national Vis-NIR-MIR spectral libraries and use in combination with (hyperspectral) composite imagery may provide a cost-effective long-term monitoring strategy.
This research is embedded within the strategic research plan of the Netherlands Space Office 2019-2025. We expect that aligning our national research activities with the European Space Agency will greatly benefit the prospect of developing a national improved soil mapping and monitoring system. Moreover, a potential pathway to further improve the alignment of research activities supporting both national and international priorities is to align our activities with the European Space Agency, GEO and the Global Earth Observation System of Systems (GEOSS).
Building the Italian National soil spectral library
Lorenzetti, Romina (1,2);
Priori, Simone (1);
L'Abate, Giovanni (1);
Fantappiè, Maria (1);
Barbetti, Roberto (1) - 1: CREA, Research Centre for Agriculture and Environment, Florence, Italy;
2: CREA, Research Centre for Agricultural Policies and Bioeconomy, Rome, Italy
Successful soil property predictions based on diffuse reflectance spectroscopy (DRS) in the Visible and near infrared (VisNIR) domain were found by several authors, and many studies were carried out in several areas of the world, from local to continental scale. However only few countries have currently published their own national soil spectral library, among them France, Denmark, Florida, Czech Republic, Brasil, Columbia. The two main objectives of wide soil spectral libraries are: i) development of predictive models for rapid assessment, in laboratory or in the field, of some soil properties; ii) supporting calibration of predictive models based on hyperspectral satellite images. This study provides a first version of the Italian Soil Spectral Library, and some examples of calibration of predictive models at national scale. The overall objective was to build a representative soil spectral library of the Italian soils for statistical inference models to allow the exploitation of rapid quantitative predictions of soil properties, classifications for digital soil mapping, and soil monitoring at national scale. Spectra were acquired over soil samples currently stored in the soil archive of the CREA, Research Center for Agriculture and Environment (Florence) which currently groups about 16,800 samples. Location, site properties, and soil parameters of collected samples were stored in the national soil database (ISIS – Italian Soil Information System). The Spectral library inherited not only soil physico-chemical parameters data, but also parent material, morphology, soil region, soil systems and soil classification. Spectra were acquired by FieldSpec 3 Hi-res (ASD, U.S.A.), with a spectral range of 350-2500 nm. The spectroradiometer was equipped with a dedicated contact probe (halogen bulb light source 6.5W), designed for direct contact with solid material. The actual version of the Italian spectral library contains over 1179 collected spectra. It includes the basic soil spectra variation of the Italian soils, since 19 reference soil group (WRB) are represented. The pedological variability covered by the collected samples resulted to be higher than the variability covered by the free European library (LUCAS) for the Italian territory. The spectral library was designed to be compatible with the Global Soil Spectral Library (Soil Spectroscopy Group). Principal component analysis (PCA) was carried out as explorative investigation of the spectral variability and to individuate eventual sub-groups of samples. PCA score plot showed two sub-groups, slightly differentiated from the rest of the samples. The first one was characterized by high content of clay (>40%) and the second one by high content of organic carbon (SOC>10%).
Partial least square regression (PLSR) was used for modelling over the whole spectral dataset. Due to the very high variability of those variables, national models were not expected to be very accurate. However, results showed a fair predictability for clay (R2=0.77, RMSEP=7.69%, RPD=2.10), SOC (R2=0.75, RMSEP=0.84%, RPD=1.98) and calcium carbonate (R2=0.87, RMSEP=4.76%, RPD=2.79). RPD showed very good values (around 2) because of high standard deviation of the data. Specific models of spectral data subgroups (e.g. forest soils, high clay soils) should be tested to improve the prediction accuracy.
Assessment of Climatic Variability on Optimal N in Long-term Rice Cropping System
Regmi, Sabina - Nepal Agricultural research Council, Philippines
Climatic variability is one of the most significant factors influencing year-to-year crop production, even in high yielding and high-technology agricultural areas. Many studies have attributed variation in yield and crop response to N fertilizer in general terms to differences in varietal characteristics, but few attempts have been made to systematically disentangle the contributions of the genotype from other factors as climatic conditions. In this study, we used ORYZA V3 rice crop model to evaluate impact of climatic variability on optimum nitrogen application rate in rice cropping system. The results shows that, solar radiation and N management practices play important roles in the response of N in grain yield. Maximum and minimum temperature have less effect on the grain yield compared to the solar radiation. Optimum N was higher in the dry season compared with the early wet season. Optimum N rate for the grain yield was around 200, 150 and 100. Nutrient use efficiency (NUE) was higher in early wet season(EWS) and late set season(LWS) in higher rate of nitrogen compared to the dry season( DS). Observed grain yield and simulated grain yield was almost similar in both seasons. The ORYZA simulation model performs well for estimating optimum N application.
Key words: ORYZA v3, climatic variability, grain yield,
A Semi-arid Mediterranean Soil Spectral Library of Degraded Soils to Support Ground Truthing and Validation of Space-based EO Data
Schmid, Thomas Fritz (1);
Chabrillat, Sabine (2);
Milewski, Robert (2);
Pelayo, Marta (1);
Sierra, María José (1);
Millán, Rocio (1);
Ben Dor, Eyal (3) - 1: CIEMAT - Research Center for Energy, Environment and Technology, Spain;
2: GFZ - German Research Centre for Geosciences, Potsdam, Germany;
3: Department of Geography and Human Environment, Tel Aviv University, Israel
Mediterranean soils are in general fragile and erosion, salinization, compaction, desertification and pollution are some of the main processes causing their degradation, which occur over space and time. Soil erosion is a land degradation process which is often found in cultivated environments due to natural causes (e.g., climate events) and accelerated by human activities (extensive tillage). However, degraded lands have been suggested as a possible solution to land scarcity for agricultural purposes and are needed to meet mounting global demands for agricultural goods. There is a tendency to use degraded or marginal lands with the aim of avoiding the environmental consequences of agricultural expansion into high-value ecosystems. However, the current status and use of these degraded lands is not well characterized. Inadequate land management cause changes to the abiotic and biotic soil surface properties. The result is a transformation of the soil as different surface or subsurface horizons will be exposed.
It has been shown that advanced hyperspectral VNIR/SWIR spectroscopy to be a promising tool to obtain detailed information on different topsoil properties. The objective of this work was to: 1) compile site-specific spectral libraries to characterize and monitor different soil surface properties that are associated to soil conditions and erosion processes within the semi-arid Mediterranean region of Central Spain, and 2) use the spectral information to validate current and future satellite-borne multispectral and hyperspectral sensors. In this context, work was carried out using an integrated approach using multiple source data which included field spectroradiometry, hyperspectral airborne data, high spatial resolution multispectral data taken with a drone and physical and chemical analyses of soil samples from reference sites as well as existing cartographic data. The study areas are in the Autonomous Region of Castilla La Mancha within the provinces of Toledo and Guadalajara. The first example was carried out with data that was obtained within the EU-FP7 EUFAR SedMedHy 2011 and Masomed 2017 projects in Camarena. The latter example is within the framework of the ERANeT FACCE_SURPLUS INTENSE project "Intensify production, transform biomass to energy and novel goods and protect soils in Europe" in Casasana. The developed site specific data base of semi-arid Mediterranean soils contains a total of 740 field spectra representing sandy, calcareous and gypsiferous soils and their corresponding physical and chemical soil analyses. The spectral data were acquired over different top and emerging underlying soil horizons and associated to slightly, moderately and strongly eroded soils and accumulation zones. The development of these spectral library databases form an essential part of the integrated approach to support: 1) the methodological development for the calibration/ validation of soil spectral models for soil properties prediction, and 2) developing the test sites of Camarena and Casasana for calibration/validation of L3 soil products from airborne and space-borne hyperspectral and multi-spectral sensors. Furthermore, with available space-based EO time series data this will serve to monitor the sustainable management of marginal lands and degraded soils to support future stakeholders and policy makers decisions.
Assessment of Crop integrating Vegetation and Soil Indices and soil properties. Case study: NE of Romania
Stoleriu, Alexandra Petronela (1,2);
Breaban, Iuliana Gabriela (1,2) - 1: Doctoral School of Geosciences, Faculty of Geography and Geology, "Al. I. Cuza" University of Iasi, Romania;
2: Integrated Center of Environmental Science Studies in the North Eastern Region - CERNESIM, Iasi, Romania
In the last decades, the development of human society has been based on the rapid progress of techniques and technologies that are oriented to a qualitative and quantitative knowledge of environmental components, as well as the design of efficient systems for processing, organizing and storing the information obtained thru remote sensing.
Remote sensing is a powerful tool for monitoring and identification of crop types from certain area. In order to have yield statistics close to reality, classifying the different types of crops it is very important. Crop mapping plays an important role in sustainable and precision agricultural practices, dealing with the environmental challenges concern and climate change. Classification of crops provides essential information that is useful in a various decision making process for managing agricultural resources. Products based on satellite image processing and analyzing can provide timely and accurate information on progress of vegetation, crop phenology, crop type development, reliable estimation of crop production using advance classification techniques.
The analyses was carried out in a cropland area located in NE of Romania with coordinates 47°21ʹ0.86ʺ N and 26°49ʹ37.07ʺ E, characterized by continental climate (hot dry summers and cool winter). Rainfall ranges from 500 mm to 700 mm, 80 % from the entire area being predominantly from 500 mm to 600 mm assuring the irrigation of crops growing in the summer. Soils are mostly chernozems (60 %) and phaeozems (10 %) followed by anthrosols (12 %) and aluviosols (6 %). The Valea Oii catchment has 9700 hafrom which the total agricultural surface is 6000 ha plus 3700 ha occupied by other land uses. The study focused on two main crops growing in the summer season maize (35 %) and wheat (10 %).
The crop type mapping based on both sattelite data (SENTINEL 2, 3) and crop data represent the main objective of the current study. With this aim, 15 images from Sentinel 2 between April and November 2018 were accquired in order to studying changes of vegetation, especially in farmland with wheat and maize, based on spectral information obtained from SENTINEL 2 and the influence of the soil properties: texture (silt, clay, sand), humus, total N (TN); available P (AP); available K (AK); pH. To classify the different crop types Random Forest classification algorithm was applied using SNAP software. Several vegetation and soil indices were generated; NDVI, GNDVI, NDWI are the foremost indices for studying the crop characteristics, whereas SAVI, MSAVI gives the information about soil and vegetation reflectance. An overall accuracy of 97.84 % and 0.98 RMSE were estimated.
The Combination Of EO And None EO Data To Quanitify And Monitor Chemical And Physical Properties Of Soils In Real World Agronomy Applications.
Travers, Marcus (1);
Wilson, Jim (1);
Woods, Doug (1);
Petit, David (2) - 1: SoilEssentials Ltd, United Kingdom;
2: Deimos Space UK Ltd, United Kingdom
Through government funded innovation research and development, SoilEssentials has worked in collaboration with agronomy companies, EO data companies, Universities and food producers. They have created applications that bring real benefits to farmers, food processors and retailors. Understanding the physical makeup and chemical properties of soil is an essential part of the successful creation and running of applications.
The information provided by the applications reduces waste, improves carbon use efficiency and promotes soil health and quality. These enhance food security and benefit the whole supply chain. Worldwide, this leads to sustainable economic development, improvements in health and social welfare. Understanding and quantifying the status and changes in soil chemical and physical properties are essential in all parts of the process from plough to plate.
The vast areas used in food production mean that EO data is an essential part of the data mining and analysis processes. These data then need to be combined and analysed alongside other forms of data to create information that is useful to farmers, suppliers, manufacturers and retailers.
Despite the success so far, challenges still exist in EO data provision. Data quality, availability and cost are all barriers to widescale uptake of EO driven applications. The raw EO data requires interpretation and processing to create derived data sets that are useful in applications. This can often lead to misinterpretation, error and unsuitable products that result in derived data that is unusable in agronomy applications.
The lack of suitable EO derived data products often comes from a lack of appreciation of the product requirement from data suppliers or a misunderstanding of what can be supplied from application developers. In other words, a simple lack of communication and understanding. Often the derived data is created before the application has been identified or understood. This results in unnecessary further development or data interpretation.
To drive the industry forward and make best use of the workshop, the authors will demonstrate with fully commercial examples, what currently works well in EO derived soil data and what barriers still exist. The objective is to build on the communication between EO data services and applications development in the food production sector. This will accelerate the provision of actionable data form EO data sources and benefit farming, food production and the environment as a whole.
Sentinel-2 time series to map Topsoil Organic Carbon content over Temperate Croplands: an Overview of recent results from the Versailles plain (221 km², France)
Vaudour, Emmanuelle (1);
Gomez, Cécile (2);
Loiseau, Thomas (3);
Baghdadi, Nicolas (4);
Arrouays, Dominique (3);
Lagacherie, Philippe (2) - 1: UMR ECOSYS, AgroParisTech, INRA, Université Paris-Saclay, France;
2: Lisah, Univ of Montpellier, INRA-IRD-Supagro, Montpellier France;
3: INRA, InfoSol Unit, US 1106, 45075 Orléans, France;
4: Irstea, University of Montpellier, UMR TETIS, Montpellier France
Recent works have shown the capability of satellite Sentinel-2 (S2) to update topsoil organic carbon (SOC) content prediction over temperate agroecosystems characterized with annual crops. As S2 sensors acquire data with regular and high temporal (weekly) frequency, they enable to investigate the specific influence that date acquisition may have on such performance. Moreover, because such spectral models rely on the availability of bare soils, the predicted area is limited. A possible solution for increasing the bare soil area consists in creating mosaics composed of several dates.
For the purpose of SOC content prediction over the croplands of the Versailles Plain (221 km², France), this study addresses first, the influence of acquisition date; and subsequently, mosaicking strategies.
The specific influence of date is analysed through soil moisture measurements and S1 derived maps of soil surface roughness. A new S2 soil moisture index is proposed for characterizing the time variation in soil moisture across the considered dates. Best single performances (RPD ≥1.4) were scored for Spring dates with low roughness and low moisture content.
For mosaicking strategies, four variants of composite image were constructed each on the following criteria amongst several available acquisition dates: i) least NDVI value; ii) least soil moisture index; iii) best single prediction performance by decreasing order amongst dates; iv) least average soil moisture index of the soilscape unit monitored for soil moisture amongst several available dates. Approach iv based on a reduced number of dates scored the best performance (RPD 1.5), and its areal gain shall be improved by incorporating more Spring dates from more years.
Acknowledgment: This work was supported by CNES, France and was carried out in the framework of the TOSCA “Cartographie Numérique des Sols (CNS)” program (grant number 3261-3264 CES Theia CartoSols) of the CNES and also benefited support from the TOSCA-PLEIADES-CO program (grant number 3249 SENTINEL_PLEIADES-CO) of the CNES.
Can We Retrieve Soil Property Profiles from Spaced Based EO, In-Situ Data and Modeling?
Zeng, Yijian;
Zhao, Hong;
Su, Bob - Department of Water Resources, ITC Faculty of Geo-Information and Earth Observation, University of Twente, Netherlands
Soil properties used in the global land surface models (LSM) are mostly from existing global soil databases, which is not necessarily representing the in-situ reality. This simplification of soil property will lead to biased LSM results. For example, if the soil property in LSM has a smaller saturated hydraulic conductivity (Ks) than the in-situ one, it will be expected that the LSM will hold much more water than the in-situ reality (e.g., a smaller Ks leads to a slow soil water drainage).
Furthermore, most of the global soil database only provides soil properties for the topsoil layer, and there is no profile information available. On the other hand, the simulation of soil processes, capturing the soil moisture and soil temperature dynamics, needs soil properties at different soil layers. The realistic simulation of soil processes will help provide the close-to-reality land surface states and processes, which is fundamentally essential and needed for Numerical Weather Prediction models to increase their forecast skills.
In this study, we compared soil properties, as collected in-situ over Tibetan Plateau across three different climate zones, with the existing global soil databases. The comparison results show the bias (/uncertainty range) in global databases vary with climate zones. Furthermore, the approach for retrieving the soil property profiles from in-situ radiometer’s TB observation (and other in-situ remote sensing measurements) was discussed.
Effect Of Sandy Amendment Combined With different Doses Of ManureOn Water Retention In Sandy soil Of Arid Areas: Soil Negga In Southeastern Tunisia
Zriba, Zied (1);
Karbout, Nissaf (2);
Bousnina, Habib (3);
Moussa, Mohamed (4) - 1: Institut des régions arides, Tunisia;
2: Institut des régions arides, Tunisia;
3: institut national agronomiqueTunisia;
4: Institut des régions arides, Tunisia
This work aimed to study the effect of the sandy amendment on the water retention of sandy soils in arid zones. It was driven on two plots located in Negga in southern Tunisia, the first plot is amended by 20 cm sand sandy soil and with three different amounts of sheep manure (5 tons / ha, 10 tons / ha and 15 tons / ha) . the sand and manure used comes from the Negga region (southwestern Tunisia). The results show that adding manure increases the available water supply for plants; This effect is evident that the amount of manure is greater than 10 tonnes / ha in a control soil. From this threshold, the saving of useful water relative to unamended soil is all the more important as the rate of amendment is high. The gain from the change in the amount of 15 tonnes / ha is slightly greater than that of the quantity of 5 tonnes / ha and 10 tonnes / ha.
GROW Observatory as a Soil Moisture Data Validation Service
Konstantakopoulos, Georgios (1);
Hemment, Dr Drew (2);
Gonzalez, Dr Raquel Ajates (3);
Xaver, Angelika (4) - 1: Future Everything;
2: University of Edinburgh;
3: University of Dundee;
4: TU Wien
Soil moisture is an essential climate variable recognized by the Global Climate Observing System (GCOS). Measuring soil moisture at a global scale is a major challenge for satellite science. The need for spatially distributed ground observations providing long-term reference observations is evident.
The GROW Observatory (GROW) is a citizens' observatory (COs) funded by the European Commission (H2020) with a purpose to demonstrate a complete CO system for sensing soil, in particular, soil moisture, and its changing state. It is the first attempt to deliver an operational CO at a continental scale and with a long term, sustained commitment.
To this end, GROW is generating an unprecedented crowdsourced soil moisture dataset with the purpose of validating Sentinel-1 satellites of the European Earth observation program Copernicus. In doing so to:
Improve our understanding of the climate changes
Inform and improve the way we grow food
Maintain and improve our soil quality
The availability of observations on the ground is vital, not only for validating but also for further developing the emerging high-resolution products based on the Sentinel-1 mission.
GROW adds a novel in-situ component to the Global Earth Observation System of Systems (GEOSS). By implementing open data standards, GROW integrates with traditional GEOSS data sets bringing together policymakers, small and medium-sized enterprises, academics, grassroots communities, farmers, designers and artists into the GEOSS community, creating an unparallel wealth of knowledge and experience in understanding and analysing soil moisture data.
Technically, GROW, as a system, is using state-of-the-art technologies (IoT, Mobile Apps, APIs, M2M) both in platforms and components. The system has been designed to be fast and reliable using ground sensors in collecting and analysing data. Trust in data quality is addressed by a strategy validating data at two stages: before and after data collection. Citizen-focused assurances include training, robust design of protocols and submissions forms, sensor calibration, trials, and data quality checks.
Unlike other citizen science projects, the distinctive dimension of GROW is its core focus on closing the loop between new scientific data, policy and practice, creating useful data products and applications, and overcoming barriers to uptake. Among GROW’s core objectives is, therefore, to create a collaborative environment capable of designing, developing and delivering better-informed solutions on an unprecedented scale.
GROW is a scalable and replicable model able to extend the soil moisture in-situ dataset spatially and temporally, to address new and emerging data needs and in doing so can act as a policy engagement platform (observatory <-> policy <-> interface) and is proposed to be distributive, with a commercial model providing data-as-a-service and observation-as-a-service applications.
We believe GROW is among the target users for the World Soils 2019 user consultation meeting. GROW is capable of providing a wealth of knowledge from our experiences in developing the project and in doing so a plan about the Observatory's future sustainability as a unique-of-a-kind solution to esa in regards to soil moisture data generation and validation for space programs.
09:00 - 10:00
09:00 -
Peat Soil Condition Monitored With Copernicus Sentinel-1 Data
Bradley, Andrew V (1);
Sowter, Andrew (2);
Large, David J (1);
Marshall, Chris (1);
van Leeuwen, Hans (3);
Marsh, Stuart (1);
Andersen, Roxane (4) - 1: Department of Chemical and Environmental Engineering, Faculty of Engineering, University of Nottingham, UK.;
2: Geomatic Ventures Limited, Innovation Park, Nottingham, UK.;
3: Gaiavision, hesselingen 12, meppel, the Netherlands;
4: Environmental Resources Institute, University of Highlands and Islands, North Highland College, Thurso, Scotland, UK.
Peat soils are around 3% of the land area, contain roughly a third of soil carbon, whilst functionally sequestering around half of the annual fossil fuel emissions. Damage to many peat soil areas through resource extraction, persistent pastoral and arable use, poor management and drainage, has reduced their capacity to swell ‘healthily’ and maintain effective peat soil functioning. We demonstrate how Sentinel-1 data detects peat motion and can characterise the range of different conditions of peatland to assist in the monitoring and management of peat soils. Using the ISBAS technique, surface deformation can be detected in areas normally rejected by conventional satellite interferometry techniques where there is low coherence (for example in agricultural fields and blanket bogs). We show capability to monitor peat soil motion at the national level for both blanket bogs and fenland type agricultural areas, progress in monitoring peat soil recovery in peatland restoration programmes and the challenges for in situ monitoring. From these results we propose that this method to monitor and characterise peat soils with Sentinel-1 radar data will be a valuable contribution to an EO based soil monitoring system.
09:20 -
A 4-Year Experiment On The Estimate Of Soil Moisture By Spaceborne InSAR Data: Comparison With In-Situ Measurement, Soil Moisture Models And Future Perspectives
Conde, Vasco (1);
Mira, Nuno (1);
Nico, Giovanni (2,3);
Catalao, Joao (1) - 1: Instituto Dom Luiz (IDL), Faculdade de Ciências, Universidade de Lisboa, Lisboa, Portugal;
2: Istituto per le Applicazioni del Calcolo (IAC), Consiglio Nazionale delle Ricerche (CNR), Bari, Italy;
3: Institute of Earth Sciences, Saint Petersburg State University (SPSU), Saint Petersburg, Russia
In this work we present the results of a 4-year experiment carried out in a farm located approximately 20 km east of Lisbon, Portugal, close to the Tagus River estuary (N 38º 48.291’, W 8º 51.122’). Two test sites were selected to study the effects of both bare soil and vegetation. The first test site is characterized by low vegetation, a few scattered trees and a soil composed mainly of sand while the second one consists of bare soil, mainly composed by clay. In each site, a set of five soil moisture sensors were deployed and set to record soil moisture in an hourly basis. Soil moisture maps were generated by processing Sentinel-1 images acquired over the two test sites using different interferometric techniques [1].
In this work we present the results of the experiment and discuss them in terms of the impact of vegetation on the capability of spaceborne C-band SAR to provide accurate estimates of soil moisture and the assumption made to model the interferometric phase and coherence in terms of soil moisture. We also provide examples of soil moisture maps obtained by inverting the interferometric phase images.
Finally, we discuss the perspective use of Sentinel-1 images to operationally provide high resolution moisture maps, updated every six days, for both ecological and precision farming applications. The topic of merging Sentinel-1 measurements of soil moisture with estimates provided by other spaceborne sensors is also discussed.
References:
[1]. Conde, J. Catalão, G. Nico, P. Benevides, “High resolution mapping of soil moisture in agriculture based on Sentinel-1 interferometric data”, Proc. SPIE 10783, Remote Sensing for Agriculture, Ecosystems, and Hydrology XX, 107831U, 2018.
09:40 -
Soil Properties Zoning for Precision Farming Based on Climate-driven Clustering of Landsat and Sentinel-2 Time Series
Reyes, Francesco (1);
Casa, Raffaele (1);
Mzid, Nada (1);
Pascucci, Simone (2);
Pignatti, Stefano (2) - 1: DAFNE, University of Tuscia, Italy;
2: CNR-IMAA, Rome, Italy
The recognition of spatial patterns within agricultural fields, presenting temporally stable similar plant response, is very important for a range of applications, for example in precision management of sowing, fertilization and irrigation practices. This can be currently performed at low-cost by using open access, high spatial resolution multi temporal imagery, such as that provided by Landsat and Sentinel-2. For this purpose, a number of clustering algorithms have been developed, mainly based on the analysis of vegetation indices related to plant vigor, possibly integrated by yield maps and geophysical surveys, when available. The response of rainfed crops growth to soil and eviromental factors, observed from remote imagery, is likely to provide some specific indication on the within-field variability in soil properties, such as the water holding capacity. It is expected that the spatial pattern of crop vigour will differ between dry and rainy years, revealing areas of different soil texture and thus different available water content.
In order to assess this hypotesis, a novel clustering methodology was developed, that takes into account the seasonal dynamic patterns of crops, weather and soils. A dataset of Landsat and Sentinel-2 images acquired over the last five cropping seasons in agricultural plots in Central Italy was assembled. Meteorological data obtained from nearby weather stations were used to calculate the Standardized Precipitation Index (SPI) for individual years and sites. Accordingly, years were classified in three rainfall regimes (dry, median and wet years).
A Normalized Difference Vegetation Index (NDVI) time series was calculated for every field based on Sentinel-2 and Landsat imagery. Crops were classified as potentially-irrigated (Summer) and non-irrigated (Winter/Spring) based on the timing of the vegetation vigour peak. Only non-irrigated crops were then considered for successive analysis, as their soil water conditions were only influenced by weather variability, and images corresponding to vegetated periods were retained for clustering. For every site, images were grouped based on three rainfall regimes, resulting in up to three groups of images per field. A field based k-means clustering was performed for vegetation indices corresponding to the same rainfall regimes in order to assess the intra-plot spatial yield patterns. Clustering inputs also included a digital elevation model (DEM), obtained from LiDAR scans at 1m resolution, for those fields having variable topograpy.
The spatial classifications obtained by the clusterings for the different years were compared with maps of soil properties such as texture and available water content obtained from intensive ground sampling and soil resistivity maps. The methodology provided interesting results revealing the usefulness of considering the seasonal climate for the indirect assessment of within-field variablity of available water content.
10:00 - 10:20
10:20 - 10:40
10:40 - 12:20
10:40 -
The Copernicus Space Component and possible way forward
Bamps, Catharina - European Commission, DG GROW - CLMS, Belgium
N/A
11:00 -
The Copernicus High Candidate Expansion Mission LSTM
Koetz, Benjamin - ESA/ESRIN, Italy
N/A
11:20 -
The Copernicus High Candidate Expansion Mission CHIME
Rast, Michael - ESA/ESRIN, Italy
N/A
11:40 -
The ASI PRISMA Mission, Status and Perspectives
Lopinto, Ettore - ASI - Agenzia Spaziale Italiana, Italy
N/A
12:00 -
EnMAP - Mission Status and Soil Related Mission Science Preparation Programme
Chabrillat, Sabine (1);
Foerster, Saskia (1);
Guillaso, Stephane (1);
Kuester, Theres (1);
Ward, Kathrin (1);
Safanelli, Jose Lucas (2);
Milewski, Robert (1) - 1: Helmholtz Center Potsdam GFZ German Research Center for Geosciences, Germany;
2: University of Sao Paulo, Brazil
Reflectance spectroscopy of soils in the visible-near infrared have been a largely demonstrated method for the quantitative determination and spatial mapping of several key soil surface properties based on laboratory, airborne and spaceborne data at different scales and resolution. The predictions were successful in local areas, when bare soils are exposed at the surface, surface conditions are appropriate, and ground data are available. With the recent and upcoming launches of hyperspectral missions, more advances are necessary to ensure the capabilities for the delivery of high quality soil properties maps at regional and global scale. For example, limitations in data and software availability, in soil spectral modeling applicable at larger scale, and in availability of global ground databases, are major areas of research nowadays with extensive developments.
In particular, in the frame of the science preparation program of the EnMAP hyperspectral mission in Germany, areas of intensive research in soil mapping are linked to the demonstration of the potential of spaceborne applications of hyperspectral imagery and focus on: a) methodological developments toward improved algorithms and operational tools for bare soil mapping based on large-scale soil spectral libraries, b) development and harmonization of soil spectral databases for model calibration, and c) influence of disturbing factors on the extraction and modeling of soil properties. For this, digital soil mapping studies using airborne and simulated EnMAP imagery from several test sites are used to demonstrate potential and limitations for the quantitative prediction of key top-soil properties such as soil organic carbon, clay and iron oxide content. In this presentation, we will show developments related to the extension of current toolboxes and software for operational soil mapping (HYSOMA/ EnSoMap), and to the use of the EU-wide LUCAS soil spectral library for model calibration. Further, methodological advances related to the analyses of the influence of green and dry vegetation cover on the soil properties prediction, and development of correcting factors will be shown. Then laboratory developments related to the assessment of the influence of different set-ups and instrumentation on the merging of soil spectral libraries will be demonstrated, along with the use of Google Earth Engine for the development of global soil maps that can be used in synergy with more detailed hyperspectral characterization in regional and dynamic areas. Overall, this paper demonstrates the high potential of upcoming hyperspectral satellite missions to support international efforts in soil mapping and monitoring, and brings the discussion on the post-processing of this soil information to users and stakeholders and how it can be translated into understandable messages for policy makers.
12:20 - 12:30
12:30 - 13:30
13:30 - 15:10
13:30 -
Digital Soil Mapping and Assessment: Why and How Earth Observation is a Key Component
Poggio, Laura (1);
Mulder, Vera Laetitia (2);
Samuel-Rosa, Alessandro (3);
Padarian, Jose (4);
Arrouays, Dominique (5) - 1: ISRIC, Netherlands, The;
2: Soil Geography and Landscape group, Wageningen University, the Netherlands;
3: Department of Agronomy, Universidade Tecnologica Federal do Parana, Brazil;
4: University of Sydney, Australia;
5: INRA – Orleans, France
Soil is an important component of the environment and it plays a fundamental role in many ecosystems and their functioning. Spatial information about soil is fundamental to understand its variability in the landscape. Soil maps have long been produced using base information coming from other environmental sources, such as vegetation or topography. In the early 2000s the concept of digital soil mapping (DSM) was formalised as the production of spatial soil information based on a statistical relationship between soil observations and environmental information, such as vegetation, geo-morphology, land cover/use and climate, as proxies for the soil forming factors. Earth observation (EO) data are often used to derive environmental information. In recent years the abundance and availability of EO data and products made them a fundamental tool for DSM. The main challenges are:
1) the selection of the most relevant EO data and products, given that often EO data are reflecting the current cover on the soil (e.g. vegetation) more than the soil itself or its past conditions. It is therefore important to select products that represent key factors for soil properties variability in space and time. Examples include soil organic carbon, pH, texture, cation exchange capacity, electrical conductivity, and soil hydrological properties;
2) the limited coverage of the electromagnetic spectrum relevant for soils and the coarse band widths for the available sensors. Most sensors are optimised for vegetation, water or human activities monitoring. Consequently, retrieval of useful soil information from existing EO data is often sub-optimal. Therefore hyper-spectral sensors would be a key tool to detect more soil-related components;
3) EO data mainly represents the surface of the soil or only a very thin layer. However soil is a 3D body and further relationships have to rely on statistical models between what is observed on the surface and what is measured at depth.
So far the main application of EO data in DSM focussed on soil properties (primary or derived) and soil taxonomy (i.e. soil types). However EO could and should also be used to monitor the quality and quantity of soils and to create time series of soil degradation. Research is also needed on the harmonization of existing EO platforms, allowing to directly assess soil and soil condition with EO data having sufficient spatial, spectral and temporal resolutions. This is key for developing a robust global soil monitoring system.
This work will present the general methodology of how EO is integrated with data from field soil observations and present examples from different regions and sensors, with a special focus on Copernicus products.
13:50 -
Soil Composite Mapping Processor (SCMaP) Product Suite - Potential and Challenges of an EO-based Soil Monitoring Data Base
Heiden, Uta;
Zepp, Simone;
Marconcini, Mattia;
Metz-Marconcini, Annekatrin;
Jilge, Marianne - DLR Oberpfaffenhofen, Germany
Due to the joint effort of environmental institutes, there are a number of large-scale regional soil information systems such as the European Soil Database (Panagos, 2006), the Africa Soil Information Service (Hengl et al. 2015), the Soil Map of China (Shi et al., 2004) but also global maps such as the Soil Map of the World (FAO/UNESCO) and the Harmonized World Soil Database (FAO/IIASA/ISRIC/ISS-CAS/JRC, 2009). Most of the recent maps are generated using digital soil mapping approaches that combine medium resolution EO-based information about vegetation presence and dynamics, terrain metrics and derived land cover / land use information as covariates with soil profile information.
With the ability to access the archives of high-resolution multispectral data such as Landsat 4-8 and Sentinel-2, new opportunities using high resolution EO data arise that allow direct information retrieval from exposed soils. Thus, EO-based soil monitoring systems can be improved due to higher spatial detail of top soil formation, higher thematic detail (continuous soil parameter) as well as the ability to investigate the potential of such systems for soil monitoring. High spatial and temporal soil information is crucial to analyse soil developments and for monitoring long term changes to assess soil degradation. This information is essential to achieve sustainable food security, health and high productivity of soils.
The Soil Composite Mapping Processor (SCMaP) is a fully automated approach to make use of per-pixel based bare-soil compositing (Rogge et al. 2018). It results in a Landsat-based (L4-8) product suite consisting of multispectral soil reflectance composites, albedo, distribution of exposed soils, statistical information related to soil use and intensity as well as information about the used number of pixels for evaluating the reliability of the generated data sets. The products will be presented for several European regions and different time periods. Major technical studies such as the threshold analysis across Europe to differentiate between exposed soils and covered soils will be presented. This is important since different climatic conditions, land use and agricultural practices can have an influence on the extraction of the bare soil composites. Thus, several European countries have been processed to test the transferability of the method to different climatic conditions and agricultural practices. Validation with German statistical data and with CORINE layer shows the high accuracy of the derivation of exposed soils from multispectral data archives. The presentation also reveals the potential of a combined analysis of the various existing soil profile archives with the SCMaP exposed soil composites. In future, SCMaP will be adapted to Sentinel-2 and the harmonized L8/S2 data sets to continue the provision of SCMaP data products. Further, the combination of top-soil profile data bases with SCMaP exposed soil reflectance composites to continuous and high-resolution information about soil parameter such as soil units, SOC and texture and mineralogy will be explored.
14:10 -
Multi-Scalar Global Soil Erosion Propensity Analysis, a Problem-Solving Approach
Modugno, Sirio (1);
Borrelli, Pasquale (2) - 1: UN-World Food Programme,Emergency Preparedness Response Division, Italy;
2: Environmental Geosciences, University of Basel, Switzerland
Land cover/ use change and associated land degradation are significant phenomena in global change. A better understanding of global land degradation and soil erosion remains a primarily goal of current societies. Detecting land cover/ use change at global scale and where lands are becoming more exposed to erosion processes, is crucial to manage food insecurity and to contextualize environmental risks. Here, we present a two main steps-aims work to assess the effects of global land cover/ use change on soil erosion:
i) Define a simplified methodology to obtain an overview on global land degradation phenomena;
ii) Perform sub-national erosion propensity assessments by the means of a multi-scalar approach.
A modified and simplified version of the Revised Universal Soil Loss Equation (RUSLE) based on open data was used to globally describe potential erosion hotspots. Land cover, rainfall, topography, and soil property are the main factors considered. Global land cover/ use dynamics were obtained from MODIS-MCD12q1 product. A specific cover factor was added to each MODIS-MCD12q1 land cover class based on the presence of vegetation. Before doing that, a specific comparison with other global land cover data has also been performed. The rainfall has been considered by the WorldClim data base elaborating using the Fournier Index. Topographical factors such as the slope length has been calculated from SRTM Digital Elevation Model (DEM) data. Finally, the soil property has been calculated by the Wischermann index.
The final product has been translated in a friendly indicator to support the map production and divulgate the erosion information to support decision makers and non-technical stakeholders. At the same time, changes in the vegetation patterns were used to highlight erosion patterns and to define local trend.
In the second step, we applied a regional approach by changing data inputs such as the land cover and the DEM data. In the sub-national analysis, the landcover implemented the Sentinel-2 African land cover and the 30m SRTM DEM. In this sub-national analysis, we considered Ethiopia being one of the most affected country by erosion, and Sierra Leone being affected by recent landslide events.
The results of the second step have been compared with the global analysis highlighting how it could be useful to have a global database updated with recent remote-sensing techniques. A product such as the Sentinel 2 African land cover or a detailed DEM have a positive effect on environmental studies especially in organizations which are ‘non-research oriented’ such as the humanitarian sector.
The humanitarian daily work needs a standardized and ready to use database. In our daily work the land degradation information is used as one layer among others to contextualize the food insecurity and the environmental risk. Defining a viable and open data methodology that can be easily replicated by local offices could represent a key to get updated land degradation information from multiple countries.
This work needs further evidence on the importance of up-to-date global open database. This work would be also considered to better address future database production for humanitarian support.
14:30 -
Object‐oriented Soil Erosion Modelling: A Possible Paradigm Shift From Potential to Actual Risk Assessments in Agricultural Environments
Borrelli, Pasquale (1);
Alewell, Christine (1);
Panagos, Panos (2) - 1: Environmental Geosciences, University of Basel, Basel CH-4056, Switzerland.;
2: European Commission, JoinResearch Centre, Directorate for Sustainable Resources, Ispra I-21027, Italy.
Over the last two decades, geospatial technologies such as Geographic Information System, Remote Sensing and spatial interpolation methods have facilitated the development of increasingly accurate spatially explicit assessments of soil erosion. Despite these advances, current modelling approaches rest on (i) an insufficient definition of the proportion of arable land that is exploited for crop production and (ii) a neglect of the intra‐annual variability of soil cover conditions in arable land. To overcome these inaccuracies, we are targeting novel spatio‐temporal approaches to compute enhanced land cover‐management modules for large-scale soil erosion modelling. Current research combines highly accurate agricultural parcel information contained in the Land Parcel Identification System with an object‐oriented Landsat/ Sentinel imagery classification technique to assess spatial conditions and interannual variability of soil cover conditions at field scale. We believe that this line of research may open the door for the transition from the currently used potential soil erosion risk assessments towards the assessment of the actual soil erosion risk. Testing was performed in a medium‐size catchment located in the Swiss Plateau. The next challenge is to improve our remote sensing components, integrate Sentinel-2 data, and run the model at larger spatial scale. We seek community collaboration to develop join research on this matter.
14:50 -
GROW Observatory as a Soil Moisture Data Validation Service
Konstantakopoulos, George (1);
Hemment, Dr Drew (2);
Gonzalez, Dr Raquel Ajates (3);
Xaver, Angelika (4) - 1: Future Everything;
2: University of Edinburgh;
3: University of Dundee;
4: TU Wien (Vienna University of Technology)
Soil moisture is an essential climate variable recognized by the Global Climate Observing System (GCOS). Measuring soil moisture at a global scale is a major challenge for satellite science. The need for spatially distributed ground observations providing long-term reference observations is evident.
The GROW Observatory (GROW) is a citizens' observatory (COs) funded by the European Commission (H2020) with a purpose to demonstrate a complete CO system for sensing soil, in particular, soil moisture, and its changing state. It is the first attempt to deliver an operational CO at a continental scale and with a long term, sustained commitment.
To this end, GROW is generating an unprecedented crowdsourced soil moisture dataset with the purpose of validating Sentinel-1 satellites of the European Earth observation program Copernicus. In doing so to:
Improve our understanding of the climate changes
Inform and improve the way we grow food
Maintain and improve our soil quality
The availability of observations on the ground is vital, not only for validating but also for further developing the emerging high-resolution products based on the Sentinel-1 mission.
GROW adds a novel in-situ component to the Global Earth Observation System of Systems (GEOSS). By implementing open data standards, GROW integrates with traditional GEOSS data sets bringing together policymakers, small and medium-sized enterprises, academics, grassroots communities, farmers, designers and artists into the GEOSS community, creating an unparallel wealth of knowledge and experience in understanding and analysing soil moisture data.
Technically, GROW, as a system, is using state-of-the-art technologies (IoT, Mobile Apps, APIs, M2M) both in platforms and components. The system has been designed to be fast and reliable using ground sensors in collecting and analysing data. Trust in data quality is addressed by a strategy validating data at two stages: before and after data collection. Citizen-focused assurances include training, robust design of protocols and submissions forms, sensor calibration, trials, and data quality checks.
Unlike other citizen science projects, the distinctive dimension of GROW is its core focus on closing the loop between new scientific data, policy and practice, creating useful data products and applications, and overcoming barriers to uptake. Among GROW’s core objectives is, therefore, to create a collaborative environment capable of designing, developing and delivering better-informed solutions on an unprecedented scale.
GROW is a scalable and replicable model able to extend the soil moisture in-situ dataset spatially and temporally, to address new and emerging data needs and in doing so can act as a policy engagement platform (observatory <-> policy <-> interface) and is proposed to be distributive, with a commercial model providing data-as-a-service and observation-as-a-service applications.
We believe GROW is among the target users for the World Soils 2019 user consultation meeting. GROW is capable of providing a wealth of knowledge from our experiences in developing the project and in doing so a plan about the Observatory's future sustainability as a unique-of-a-kind solution to esa in regards to soil moisture data generation and validation for space programs.
15:10 - 15:20
15:20 - 15:30
15:30 - 16:30
