The Development of a Web-service-based On-demand Global Agriculture Drought Information System

The growing demand on detailed and accurate assessments of agriculture drought from local to global scales has made drought monitoring and forecasting a hot research topic in recent years. However, many challenges in this area still remain. One of such challenges is to how to let world-wide decision makers obtain accurate and timely drought information. Current agriculture drought information systems in the world are limited in many aspects, such as only regional or country level coverage, very coarse spatial and temporal resolutions, no on-demand drought information product generation and download services, no online analysis tools, no interoperability with other systems, and ineffective agriculture drought monitoring and forecasting. Leveraging the latest advances in geospatial Web service, interoperability and cyber-infrastructure technologies and the availability of near real-time global remote sensing data, we aims at providing a solution to those problems by building an open, interoperable, standard-compliant, and Web-service-based global agriculture drought monitoring and forecasting system (GADMFS) (http://gis.csiss.gmu.edu/GADMFS/). GADMFS will provide world-wide users with timely, on-demand, and ready-to-use agricultural drought data and information products as well as improved global agriculture drought monitoring, prediction and analysis services. For the monitoring purpose, the system lively links to near real-time satellite remote sensing data sources from NASA and NOAA and relies on drought related remotely sensed physical and biophysical parameters, such as soil moisture and drought-related vegetation indices (VIs, e.g., NDVI) to provide the current conditions of global agricultural drought at high resolutions (up to 500m spatial and daily temporal) to world-wide users on demand. For drought prediction, the system utilizes a neural network based modeling algorithm, trained with current and historic vegetation-based and climate-based drought index data, biophysical characteristics of the environment, and time-series weather data. The trained algorithm establishes a per-pixel model to produce on-demand drought prediction at ~1km or higher spatial resolution. The Phase-I system implementation is concentrated on NDVI-based drought monitoring to demonstrate the concept and feasibility. In phase I, 30-year calibrated global weekly NDVI composites from AVHRR and MODIS are used to establish the baseline and dynamics of vegetation conditions for each co-registered pixel. Multiple NDVI-based agricultural drought indices, such as vegetation condition index (VCI), have been computed from the baseline and dynamics for drought monitoring. GADMFS is a contributing component of Global Earth Observation System of Systems (GEOSS) to serve the GEOSS societal benefit area of agriculture and water. The phase-I implementation shows that open and interoperable drought related data services and processing services from this system have significantly increased the accessibility of remote sensing based agriculture drought information to the world-wide users. Such a system will also increase the utilization of drought indices related applications and researches in the GEOSS community.