ASAP - Anomaly hot Spots of Agricultural Production, a new global early warning system for food insecure countries

摘要

Real time monitoring of vegetation conditions is of paramount importance, particularly for food insecure countries, to detect possible crop and pastures production drops as early as possible. This monitoring is classically based on remote sensing indicators of vegetation conditions (typically NDVI and rainfall produced every ten days at a resolution of 1 km or more) and many web portals now offer anomaly maps and time profiles derived from these indicators; however, timely and coherent interpretation of this coarse resolution information at global scale remains challenging. With the ASAP system (Anomaly hot Spots of Agricultural Production) we propose a two-step analysis to provide every month warning of production deficits in water-limited agricultural systems. The first step is fully automated and classifies each Gaul 1 unit (i.e. first sub-national administrative level) according to a warning scale. These warnings are triggered only during the crop/pasture growing season, as derived from a remote sensing based phenology. For each Gaul 1 unit, the classification takes into consideration its fraction of agricultural area affected by a severe anomaly for two rainfall-based indicators and one biomass indicator, as well as timing in the crop cycle at which the anomaly occurs. In the second step, agricultural analysts check the automatic warnings to identify the countries which qualify as "hot spot" because of their potentially critical conditions. The system elaborates the warnings for the whole globe but the analysts focus on 80 food insecure countries in Africa, Asia and America. In their evaluation, the analysts are assisted by graphs and maps automatically generated in the first step, agriculture and food security-tailored media analysis and high resolution imagery (e.g. Landsat 8, Sentinel 1 and 2) processed with Google Earth Engine. Maps and statistics, accompanied by short narratives are then published on the website and can be used directly by food security analysts with no expertise in crop monitoring with remote sensing, or can contribute to global early warning bulletins such as the GEOGLAM Early Warning Crop Monitor, which synthesizes every month crop conditions analysis from various institutions.