Global Analysis of Daily and Monthly GRACE data for Flood Prediction

Abstract:

Floods resulted in an average of 100 deaths annually and annual property damages of ~3 billion US dollars, with likely similar impacts globally. Global warming is projected to intensify the magnitude and frequency of flood events. Increasing prediction lead times of floods could reduce flood-related losses and damages. The objective of this study was to assess the value of monthly and daily Gravity Recovery and Climate Experiment (GRACE) and its Follow-On (GRACE-FO) Total Water Storage data to extend flood lead time predictions. We evaluated spatiotemporal coincidence rates between flood events, defined by significant precipitation (≥90^th percentile) based on daily GPCP data and changes in GRACE Total Water Storage based on two monthly mascon solutions (UTCSR and JPL) and three daily solutions (UTCSR-RSWM, GFZ-RBF, and ITSG-2018). Preliminary results show that monthly GRACE data significantly improve prediction rates of annual flooding with lead times of ~ two months across regions with higher median annual precipitation (e.g., high northern latitudes and tropical zones).

In contrast, daily GRACE Total Water Storage significantly improve flood prediction rates during seasons with higher precipitation and lower for the annual flooding. However, the three daily solutions show notable differences in the magnitude, significance, and locations of flood predictions. Further investigation is needed to attribute significant prediction rates to states of the climate, topography, and soil types and quantify differences among the solutions within the context of their varying solution approaches.

This preliminary analysis suggests the high potential for GRACE data to extend flood forecast lead times and potentially improve management strategies.