Research project SR/00/300 (Research action SR)
CONTEXT AND OBJECTIVES
The advent of a new generation of high spatial and temporal resolution satellite sensors (i.e. Sentinel-1, Sentinel-2 and PROBA-V) opens the doors to develop new prototype applications much more connected to the so called end-users. In parallel, mobile communication technology is shaping a different world where interactions and near real time information exchange become very easy and flexible. Building on the most recent remote sensing and Information and Communication Technology penetration, the The management strategies of agroecosystems have evolved during the last 50 years consisting of intensifying the use of agrochemicals and agricultural machinery, expanding irrigated areas, and incorporating new crop varieties. But as excessive use of input also plays a significant role in the degradation of our environment, for the optimization of the cultivation of some crops (winter wheat, maize and potato), rates of nitrogen fertilizers application are balanced at field scale according to the crop’s needs.
Good practices in management of nitrogen inputs have several economic and environmental (run-off, leaching…) benefits and increase yield and quality of the crop. Optimized fertilization practices as well as winter cover crop reducing erosion and leaching are in line with the evolution of cropping systems toward a sustainable agriculture, sometimes called conservation agriculture. research challenge is to completely change the interactions between information users and producers and to markedly speed up the critical learning process for the remote sensing providers thanks to input and near real time feedbacks from the users.
This project, working on a JECAM site and taking advantage of the BELAIR campaigns, will contribute to GEOGLAM, an initiative supported by the G8. Indeed, in the era of Sentinels and PROBA missions, Belgium is a very interesting experimental site to design and test innovative methods for a collaborative information system because of its crop diversity, landscape fragmentation, atmospheric variability, and all the already available data sets (digital soil map, annual field delineation, etc.). The development of this BELCAM prototype is built on already existing in situ monitoring practices (wheat, potato) and on interactions with the pilot centers providing demonstration fields distributed all over the country.
METHODOLOGY
Building on the complementarity of local (professional crowdsourcing) and satellite remote sensing technologies, the research aims to develop a new collaborative pathway for operational remote sensing monitoring at the local level for one of the most dynamic and most widely distributed applications: crop monitoring. The cornerstone for such a field level monitoring will be the use of on farm information in near real time to support remote sensing services. Emphasis is given to the shortening of the information loop linking the producer and the users.
The main objective of the project is to develop remote sensing methods and processing chains able to ingest crowd sourcing data provided by farmers or associated partners either on voluntary basis or through information service exchange to deliver relevant and up-to-date information at the field and district level. The first research objective is to conceptualize and experiment collectively an open collaborative agriculture monitoring system supported by local and satellite remote sensing. On one hand, the classical product cycle made of requirements - product design - users feedback will be much speed up to iterate several full cycles during the course project. (source of imagery : ©ESA) The second collaborative component corresponds to professional crowdsourcing based on the near real time transmission of information locally collected by professionals to enrich along the season the processing of satellite time series through machine learning techniques or algorithm tuning. The outcome of this objective is the concept of the Collaborative Agriculture Monitoring (CAM).
The second research question investigates the sources of variability of the remote sensing retrieval performances for several biophysical variables such as the chlorophyll concentration, the Nitrogen content, the fAPAR, the LAI, the biomass as well as for major stress and pest damages. The outcome of this research question will be a set of crop information products relevant for practical use at field level and at administrative level.
The third research question aims to address the change of paradigm in crop modeling. Indeed, most of the existing crop growth models have been developed in data poor contexts, referring to the ‘80s and ‘90s. The emphasis was on a mechanistic approach to reproduce the physiological processes leading to the potential yield only. This third research question is to identify the mechanistic relationships between crop yield and the available EO observations and to design a new (semi-)mechanistic model ingesting the remotely sensed variables or metrics.
These three scientific questions described above will be addressed in parallel as they will interact progressively along the course of the project.
The overall outcome of the project is a prototype of a Belgian Collaborative Agriculture Monitoring system (BELCAM) allowing continuous improvement of remote sensing outputs on one hand to support farming advices at the field level to be delivered by the CRA-W, and on the other hand, to contribute to operational agriculture services for their a posteriori growing season analysis (stress, diseases and impact assessment according to regions). The working assumption of the BELCAM relies on the fact that the professionals providing information in a crowdsourcing mode (possibly anonymously but always geolocated at the field level) will obtain a more reliable information from satellite remote sensing than the ones not involved.
The research strategy will proceed by increasing progressively the number of professional partners involved in the collaborative dynamic, the sources of the satellite data along with their actual availability and the development of different methods and information products.
Therefore, the 4 growing seasons will be covered by Sentinel 1 while Sentinel 2 will cover at least 3 growing seasons. Meanwhile, the preparation of Sentinel 2 exploitation will rely on the BELAIR campaigns as well as on some RapidEye data, mainly for its spectral resolution in the red-edge.
EXPECTED SCIENTIFIC RESULTS
The overall outcome of the project is a prototype of a Belgian Collaborative Agriculture Monitoring system (BELCAM) allowing continuous improvement of remote sensing outputs on one hand to support farming advices at the field level to be delivered by the CRA-W, and on the other hand, to contribute to operational agriculture services for their a posteriori growing season analysis (stress, diseases and impact assessment according to regions). The working assumption of the BELCAM relies on the fact that the professionals providing information in a crowdsourcing mode (possibly anonymously but always geolocated at the field level) will obtain a more reliable information from satellite remote sensing than the ones not involved.
The research strategy will proceed by increasing progressively the number of professional partners involved in the collaborative dynamic, the sources of the satellite data along with their actual availability and the development of different methods and information products. The BELCAM dynamic will lead to the deliverables corresponding to processing methods and collaborative strategy, retrieval algorithms, interpretation scheme and models leading to the six products e to be delivered to the professionals along the course of the three seasons progressively.
The above deliverables will be progressively elaborated with the first year relying mainly on the BELAIR campaign and possibly SPOT 5 Take 5, the second year on S1 and most probably S2 while PROBA at 100m is expected to last the duration of the project. The Sentinel 1 systematic acquisition plan over Europe will be fully operational for the 2015 growing season while the Sentinel 2 is expected to be fully operational at the best during 2015 and at the latest for the 2016 growing season.
Therefore, the 4 growing seasons will be covered by Sentinel 1 while Sentinel 2 will cover at least 3 growing seasons. Meanwhile, the preparation of Sentinel 2 exploitation will rely on the BELAIR campaigns as well as on some RapidEye data, mainly for its spectral resolution in the red-edge.
EXPECTED PRODUCTS AND SERVICES
The BELCAM dynamic should lead to the following deliverables relevant for farmers and technical centers. If successful from a science point of view and relevant from crop management point of view, the following items could be turned into operational services as the basis of a public farming advice system :
1. Annual Nitrogen balance-sheet forecast at field level, based on previous crops rotation, farming practices (i.e. crop residues, green manure, tillage…) and field conditions (slope, soil texture…);
2. Annual field zoning, to help adjusting farming practices according to field heterogeneity compiled from previous years;
3. Potato/winter wheat Nitrogen status at the parcel level, deciding respectively whether a secondary N application would be required or not or on the level of the 3rd application of N fertilizer, based on the Chlorophyll a,b retrieval and the plant N estimate;
4. Overall crop status (LAI, fAPAR, biomass) for potato/winter/maize along the season and aggregated at the district level;
5. Major pest, N, water stress or disease damages assessment, aggregated at the district level from SAR and optical time series analysis;
6. Yield estimate, derived from the newly developed model aggregated at the district level.
The deliverables will be gradually elaborated with the first year relying mainly on the BELAIR campaign and SPOT 5 Take 5.
The second year on S1 and most probably S2 while PROBA at 100m is expected to last the duration of the project.
The Sentinel 1 systematic acquisition plan over Europe will be fully operational for the 2015 growing season while the Sentinel 2 is expected to be fully operational at the best during 2015 and at the latest for the 2016 growing season. POTENTIAL USERS
The Belgian farmers are the potential end-users as well as any entities supporting the crop production in Belgium. The BELCAM experience is expected to be promoted in the JECAM network for internationalization of the potential users.
