Agriculture - Global

Mapping the global cropland extent is of paramount importance for food security. Indeed, accurate and reliable information on cropland and the location of major crop types is required to make future policy, investment, and logistical decisions, as well as production monitoring. Timely cropland information directly feed early warning systems such as GIEWS and, FEWS NET.  In Africa, and particularly in the arid and semi-arid region, food security is center of debate (at least 10% of the population remains undernourished) and accurate cropland estimation is a challenge. Space borne Earth Observation provides opportunities for global cropland monitoring in a spatially explicit, economic, efficient, and objective fashion. In the both agriculture monitoring and climate modelling, cropland maps serve as mask to isolate agricultural land for (i) time-series analysis for crop condition monitoring and (ii) to investigate how the cropland is respond to climatic evolution. A large diversity of mapping strategies ranging from the local to the global scale and associated with various degrees of accuracy can be found in the literature. At the global scale, despite efforts, cropland is generally one of classes with the poorest accuracy which make difficult the use for agricultural applications especially when coarse resolution is used (e.g. global map of rainfed cropland areas (GMRCA) and the global irrigated area map (GIAM) at 10km). At national scale, works has been done to realize detailed land cover maps at high resolution (e.g. Africover) but no regular updates of these products is done.  This research aims at improving the cropland delineation from the local scale to the regional and global scales as well as allowing near real time updates. To that aim, five temporal features were designed to target the specificities of crop characteristics. To ensure a high degree of automation, training data is extracted from baseline land cover maps. The method delivers cropland maps with a high accuracy over contrasted agro-systems in Ukraine, Argentina, China and Belgium. The accuracy reached are comparable to those obtained with classifiers trained with in-situ data. Besides, it was found that the cropland class is associated with a very low uncertainty. The temporal features also offer a high potential for generalization. As a result, the classifier might be used on an annual or monthly basis without retraining. Using PROBA-V 100m and 300m, the method is currently being applied to the Sahel and the entire African continent thanks to a dedicated agro-ecological stratification.

Achieving sustainable food security for all people is a priority highlighted by the new Sustainable Development Goals of the United Nations, which defined its second goal as “End hunger, achieve food security and improved nutrition and promote sustainable agriculture". However, hunger remains an everyday challenge for almost 795 million people worldwide. In June 2012, the declaration of the G20 Mexico summit emphasized the needs for enhancing food security and addressing commodity price volatility.

In response to such growing pressure, it is critical to develop better agricultural monitoring capabilities from local to global scale. Since 2011, support for agricultural monitoring using satellite data has become substantial, with formal institutional support, objectives and timelines e.g. by the Global Agricultural Monitoring Initiative (GEOGLAM) building on GEO’s Agricultural Community of Practice (AG COP) and the Joint Experiment of Crop Assessment and Monitoring (JECAM).

The continuous availability of Sentinel-2 time series will constitute a major advancement for large scale agriculture monitoring capabilities. Its temporal revisit frequency of 5 days, its specific spectral bands and its high spatial resolution (10-20 meters) combined with its 290 km wide swath are particularly suited to operational agriculture monitoring applications.

In this context, the European Space Agency launched in 2014 the Sentinel-2 for Agriculture (Sen2-Agri) project which aims to provide to the international community validated algorithms and open source codes to process Sentinel-2 data in an operational manner into relevant EO agricultural products for major worldwide representative agriculture systems. These Sen2-Agri EO products consist of four Sentinel-2 based outputs:

Cloud free surface reflectance composites;

Dynamic cropland masks delivered along the agricultural season;

Cultivated crop type map and area indicator for main crop groups;

Vegetation status indicators describing on a 5 to 10 days basis the vegetative development of crops.

The project does not deliver the products but the open-source system to generate them. The system is made of software components (each one being an independent executable corresponding to algorithms) and of an orchestrator which manages these components to monitor the whole system and execute processing jobs.

Being part of the Data User Element programme, a user-oriented approach drives the entire project in order to address concrete needs. Throughout the project, key users are involved in several activities such as user requirements consolidation, products validation and assessment, users workshops and capacity building thanks to training and full scale demonstration.

The first phases of the project focus on the algorithms selection, system design and implementation. These processes were achieved using Sentinel-2-like time series (SPOT Take 5 as primary source) over 12 globally distributed test sites. These time series were complemented by in situ crop measurements shared by JECAM network members and other teams working in various agrosystems. This dataset served as input to a benchmarking study which helped to select the most suitable algorithm(s) for fulfilling products specifications at the best. The next phase is dedicated to the demonstration of the Sen2-Agri system to deliver the products with Sentinel-2 data in near real-time. It will start early 2016 with the first full crop season of the Sentinel-2 mission. The demonstration will be done at local scale (a full Sentinel-2 swath) for five sites and at national scale (~500.000 km²) for three countries. In both cases, the demonstration will be carried out in close interactions with national partners and teams working on the field, with the final objective to transfer the system to their operations.

Through its activity, the project provides a strong scientific contribution to the JECAM network and GEOGLAM initiative. The benchmarking study allowed completing one of the very first cross-cutting analyses for multiple, globally distributed sites and the coming demonstration phase will help filling the gap between state-of-the-art remote sensing practices and operational systems.

The global and regional potentials of Earth Observation (EO) to contribute to food security and sustainable agriculture in the 2050 time frame were analysed in the ESA study EO4Food, of which  the outcome will be presented. Emphasis was put on the global societal, economic, environmental and technological megatrends, that will create demand for food and shape the future societies. They will also constitute the background for developments in EO for food security and sustainable agriculture. The capabilities of EO in this respect were critically reviewed with three perspectives 1) the role of EO science for society, 2) observables from space and 3) development of future science missions.

It was concluded that EO can be pivotal for the further development of food security and sustainable agriculture. It has the potential to become the global source of environmental information that is assimilated into sophisticated environ­mental management models and is used to make agriculture sustainable. EO allows to support the whole economic and societal value chain from farmers through food industry to insurance and financial industry in satisfying demands and at the same time to support society in governing sustainable agriculture through verifyable rules and regulations.

The challenges for EO towards implementing its potential role in food security and sustainable agriculture can be summarized as follows:

The operational fleet of Copernicus satellites and the services they provide will serve as a backbone of continued EO coverage of agriculture on which to build. They will provide excellent time series of data from which e.g. crop types, calamities and farming practices (ploughing, seeding, etc.) can be derived with high accuracy.

Further research and development is needed in order to expand EO capabilities and services towards comprehensive sustainable farm management, which ensures efficient, water saving irrigation, fertilizer saving fertilization, robust, limited and timely plant protection and both high quantity and high quality yield. 

The tightening water-food-energy nexus makes clear that EO-based sustainable farm management is a multi-parameter task. Contrary to past ESA science missions, science as well as application will need constellations of dedicated sensors, which work in synergy and feed sophisticated land surface process models, which, like in meteorology, deliver products of value for society.

Research should therefore be conducted to augment the capabilities of the existing EO backbone. Dedicated missions should be defined and implemented, which develop the EO science needed for comprehensive EO-based sustainable farm management. It should be scientifically analysed how sensor constellations can augment current observational capabilities to fully comply with EO requirements for a  future “Global Smart Farm” either through the formation of convoys or the assimilation of asynchroneous EO data in farm management models.

The Malawian Point Frame Agricultural Survey project (MPFASU) is an initiative of the Ministry of Agriculture, Irrigation and Water Development (MoAIWD) of Malawi as a follow on of the Global Monitoring for Food Security project (GMFS) of the European Space Agency (ESA), aiming at developing and providing Crop Production Estimation for the country by means of geospatial technologies and earth observation (GMFS, 2013).  

For the main season 2015 the MPFASU resulted in estimates of the main crops national and other levels, whereas particularly for maize at national level a coefficient of variation (CV) of 8% for production and of 2% for maize acreage has been achieved.

The principal objective of this study and consultancy project was to strengthen the methodological framework of the MoAIWD through the exploration of alternative methods to acquire crop statistics by means of a geo-referenced (‘point’) sampling frame integrating satellite information with innovative ground survey technologies. A further focus was on the design of an appropriate methodology and sampling for this exercise, including the development or adaptation of necessary tools for survey executions.

The project has been executed by a Joint Venture of two partners, Consorzio ITA and EFTAS who are collaborating since GMFS sin 2005. In frame of this cooperation the most relevant actions were between 2010 and 2013 to jointly set up and develop the precursor of a point frame approach as part of GMFS.

The paper reports about the evolution of the GMFS services beyond the project lifetime in 2013 and the potentials for future applications.

REFERENCES

GMFS – GLOBAL MONITORING FOR FOOD SECURITY (2013a): Services Prospectus. – http://www.gmfs.info/uk/publications/gmfs3_docs/GMFS3_S03_v3.1.pdf (13.7.2013).

Since the early airborne experiments in late 80’s (Le Toan et al., IEEE GSRS 1989), SAR data have been found the best EO data sources for rice identification and for monitoring rice growth. In addition to the SAR all weather capability, the rice monitoring potential stems from the physical interaction between the radar waves and inundated fields of rice plants. Flooded fields can lead to significant water-plant stem double bounce scattering that increases during vegetative phase, and the vertical  structure of rice plant can cause stronger attenuation in VV compared to HH. Those properties usually hold for C, L and X-band SAR data, with temporal variation of the backscatter differs as a function of the frequency, polarization and incidence angle and the plant growth period.  

The rice fields mapping methods based on SAR data that have been developed mainly rely on these two properties of rice fields. The high backscatter increase during rice growing season has been exploited in

algorithms using the temporal change of the backscatter, or the polarization ratio (HH/VV) as rice indicators, using ERS data first by CESBIO (Le Toan et al., 1997), followed by numerous other works in various countries in Asia and elsewhere. Rice fields mapping using SAR data was subsequently fully demonstrated, but mostly based on high resolution data (< 30m) with small swathwidth such as ASAR APP, RADARSAT Fine mode, COSMO-SkyMed ping pong. Partly because of the unavailability or the cost of data to cover large areas, their effective use by countries and international organizations has never been achieved. 

To estimate rice production with the help of Earth Observation data is still a research subject. The approach is based on the use of agro-meteorological model to predict rice growth. This  leads to the prediction of rice yield, and to the early warning in case of growth anomaly.  EO data can provide the timing of key phenological stages as model inputs and biophysical parameters such as biomass which can be used to compare to model outputs, in order to readjust the model parameters to adapt to the site under study. This approach to rice production estimation and prediction by coupling ORYZA 2000 and EO retrievals  was demonstrated for the first time in 1999 (Ribbes, F., & Le Toan, T. Coupling radar data and rice growth model for yield estimation, IEEE 1999 (Vol. 4, pp. 2336-2338),  but again, no effective applications have been derived so far. 

As a summary, efficient rice mapping and monitoring methods have been developed but operational applications have been hampered by the lack of systematic (and inexpensive) SAR data. The launch of Sentinel-1 in April 2014 and the expected availability of consistent SAR acquisitions (in particular with high resolution and IW mode) represent an unprecedented opportunity for operational rice monitoring applications.

To bridge the gaps from research to application, the ‘GEORICE’ project has been proposed to the ESA DUE Innovator III call. The main purpose of the project is to develop innovative products using Sentinel-1 data for services on rice monitoring. The requirements are originated by organizations which are GEOGLAM (Group of Earth Observation-Global Agriculture Monitoring) and its component Asia-RiCE, and also by national users. Asia-RiCE objectives are focused on Rice field mapping, Rice area statistics, Early warning, Crop calendar, Damage Assessment and Yield estimation.  In GEORICE, The target products include rice crop grown area and rice phenological stage at a given time of the year, rice crop area estimates per season and per year, early warning of yield shortage, and indicators of rice production. Phase 2 of GEORICE will be focused on the generation of the products that will be fully assessed at the end of the GEORICE Phase 1, in agreement with ESA. The required information and products will be automatically computed and generated and will be provided to the  users in near real-time, shortly after access to Sentinel-1 data.

The demonstration has been performed on the Mekong Delta region, the Asia-RiCE Technical demonstration site selected as a region of special interest for rice monitoring approaches.  Sentinel-1 data have been acquired since the ramp up phase, and the data have been available nearly every 12 days since October 2015. Ground data collection has been organized during the three rice seasons per year in this region.

Using IW data in descending mode, methods based on backscatter temporal changes have been refined to account for the change in incidence angle across swath, and to account for the different crop calendar and cultural practices (continuous inundation, intermittent drainage, direct sowing, machine sowing..), across this large and complex region of 300 km x 300 km. The results show that rice maps and maps of rice growth stage at 20 m or 40 m pixels can be obtained shortly after data acquisition.  Validation of rice /non rice map using 120 independent observations in 3 provinces showed a very high percent of correct identification (93%). Statistics of rice area for the 2 first rice seasons in 2015 have been obtained and compared with national statistics of the previous years. The work is going on for the retrieval of  rice  biomass  for the rice model input.  

The GEORICE generated products will be accessible through a dedicated web application hosted by the RODGER platform, developed by Capgemini, which is in charge of automatically acquiring and processing the input data, storing the input and output data, and disseminating the ready-made products to the end-users.

In summary, with the availability of Sentinel-1 data, rice monitoring from space will be soon an operational application.