We are using a multi-staged approach aimed at combining physical climate indicators (Phase 1) to develop a global composite drought indicator (GCDI) centered on physical climate indicators of drought in Phase 1, which will then be combined with socio-economic elements of risk and vulnerability in Phase 2. We will use a combination of subject matter expertise and machine learning to develop and evaluate a global drought hot spot indicator. This platform will be capable of integrating an array of historical, gridded data inputs allowing multiple environmental parameters related to drought to be integrated into a more holistic view of drought conditions.
The proposed GCDI concept is built on a flexible, percentile-based and well-established analytical framework (i.e., the U.S. Drought Monitor and the existing Middle East/North Africa and South African regional composite drought indicators) that allows for a convergence of evidence approach and provides decision-makers and resource managers with historic context as to how rare a given drought is at a given severity level.
During Phase 1, the GCDI proof of concept project will initially utilize operational, publicly-available global data sets related to precipitation, soil moisture, evapotranspiration and vegetation health to produce a beta version of the tool for global, drought hot spot detection. It is our intention to:
- Build historical archives for each of the four parameters listed above;
- Combine the parameters above for the initial GCDI utilizing machine learning techniques coupled with subject matter expertise and impacts where applicable;
- Build a historical archive for the GCDI;
- Develop a user interface to access the GCDI, individual GCDI inputs and change map archives;
- Conduct an inventory of available socio-economic, drought risk/vulnerability data.