Multivariate Standardized Drought Indices to Identify Drought Events: Application in the Maipo River Basin

Abstract

Timely and reliable drought detection is the basis for sustainable drought management, which aims to keep negative impacts to a minimum. Droughts are natural and recurring events and as such cannot be avoided. Ensuring future preparedness relies on knowledge of past events. Drought indices are commonly used to detect drought events, among them standardized indices are often utilized due to their comparability across time and space. We propose a multivariate approach based on four standardized indices; the Standardized Precipitation Index (SPI), the Standardized Precipitation Evapotranspiration Index (SPEI), the Standardized Streamflow Index (SSI), and the Standardized Soil Moisture Index (SSMI), allowing us to evaluate drought patterns by each variable of interest as well as to investigate river basin specific drought propagation patterns. We apply the multivariate approach in the Maipo river basin in Chile. Detection of starting and end points identified by SPI and SPEI with three different thresholds permits us to see how and when triggering processes occur in soil moisture and streamflow. Drought propagation patterns are identified, as SSI values respond to SPI (SPEI) values with a minimum lag of three months, proving that it is not sufficient to rely only on one index for operational drought management. Further, we develop a multivariate autoregressive model (MAR1 model) for operational forecast based on SPEI, SSI, and SSMI. Model results show the applicability to obtain forecasts for each index and for jointly marginal probability density functions.