Development of a Global Agricultural Hotspot Detection and Early Warning System

Abstract:

The number of web based platforms for crop monitoring has grown rapidly over the last years and anomaly maps and time profiles of remote sensing derived indicators can be accessed online thanks to a number of web based portals. However, while these systems make available a large amount of crop monitoring data to the agriculture and food security analysts, there is no global platform which provides agricultural production hotspot warning in a highly automatic and timely manner. Therefore a web based system providing timely warning evidence as maps and short narratives is currently under development by the Joint Research Centre. The system (called “HotSpot Detection System of Agriculture Production Anomalies”, HSDS) will focus on water limited agricultural systems worldwide. The automatic analysis of relevant meteorological and vegetation indicators at selected administrative units (Gaul 1 level) will trigger warning messages for the areas where anomalous conditions are observed. The level of warning (ranging from “watch” to “alert”) will depend on the nature and number of indicators for which an anomaly is detected. Information regarding the extent of the agricultural areas concerned by the anomaly and the progress of the agricultural season will complement the warning label. In addition, we are testing supplementary detailed information from other sources for the areas triggering a warning. These regard the automatic web-based and food security-tailored analysis of media (using the JRC Media Monitor semantic search engine) and the automatic detection of active crop area using Sentinel 1, upcoming Sentinel-2 and Landsat 8 imagery processed in Google Earth Engine. The basic processing will be fully automated and updated every 10 days exploiting low resolution rainfall estimates and satellite vegetation indices. Maps, trend graphs and statistics accompanied by short narratives edited by a team of crop monitoring experts, will be made available on the website on a monthly basis.