Spatial Resolution Improvement of Precipitation Measurements for Explaining Rainfall Related Landslide Hazards

Document Type


Publication Date



Department of Geological and Mining Engineering and Sciences


Precipitation is a key triggering factor for landslides. Often landslide hazard characterization at regional or statewide scales use sparse rain gauge measurements or assume constant precipitation. This study explores the need for improved precipitation measurements for landslide hazard characterization. Identifying spatial and temporal precipitation patterns allow for improved understanding of their influence on geohazards, like landslides. Rain gauges collect point data in situ, and satellites gather areal precipitation data remotely at a coarse spatial resolution. Integration of point rain gauge measurements with downscaled areal precipitation data improves the overall spatial resolution of precipitation maps. The areal precipitation product being downscaled is the monthly Integrated Multi-satellitE Retrievals for Global Precipitation Measurement data (IMERG) for Colorado for several months in 2018. This project also examines the relationship between precipitation with vegetation and terrain variables. Correlation coefficients for April 2018 between Colorado’s areal precipitation with Normalized Difference Vegetation Index (NDVI), elevation, and aspect are 0.95, 0.41, and 0.11, respectively. Correlation coefficient results indicate vegetation and areal precipitation are highly correlated, but terrain variables are poorly correlated with precipitation. Spatial resolution of the product is improved tenfold from 0.1° x 0.1° to 0.01° x 0.01°. Areal precipitation maps are downscaled using two methods: (1) area-to-point kriging (ATPK), using areal variogram deconvolution and (2) machine learning regression algorithms relying on NDVI as the explanatory variable. The rain gauge station measurements for monthly precipitation are interpolated with the downscaled satellite precipitation products using several simple cokriging algorithms, and the results are compared. Finally, landslide locations are correlated to areas experiencing high precipitation throughout the year.

Publication Title

AGU 100 Fall Meeting