Comparisons of Rainfall Estimation from Different Sources in Hawaiʻi

Accuracy of rainfall data is critical to describe hydrological responses of a watershed with high rainfall gradients, such as those in mountainous coastal areas. Watersheds in Hawaii are particularly responsive to the received rainfall given the small size of watersheds and volcanic geology. However, rainfall data often come from different sources with different sensing methods and resolutions, and therefore might not always be concordance with their estimate on receiving volume and timing on the ground. We obtained rain records between 2006 and 2013 from four methods including ground rain gauges, reanalyzed PRISM with selected rain gauges, Hydro-Estimator satellite, and Next Generation Weather Radar level III. Then, we examined and compared the temporal and spatial resolutions among these methods. We calculated the hourly rate and examined the rain vs. no-rain periods and the continuity of the hourly raining rate. Within a watershed, the rainfall-runoff ratio with rainfall from the above methods was summarized and calculated for a few selected flood events. We found and summarized the characteristics of rainfall data from different methods: Rain gages were expected as a true observation but has challenges of spatial interpolation to describe the spatial variability. The reanalyzed PRISM rainfall incorporated both spatially modeled and observed rainfall yet missed localized patterns and extreme heavy events. Satellite offered observed spatial rainfall patterns but seemed to miss certain events. Radar provided the finest spatial and temporal resolutions but overestimated the rainfall. This study gained foundation knowledge on the advantages and short-comings of rainfall data from different sources. With this understanding, we aim to propose a better blending process of rainfall data by considering their characteristics.