Toward the Estimation of High-Resolution Daily Precipitation in Complex Regions - The Study of Intertwined Physiographic, Vegetative, and Climatologic Factors for PRISM Enhancement

Precipitation data with accurate, high spatial-temporal resolutions are crucial for modeling occurrences of extreme precipitation events and for improving our understanding of basin-scale hydrology. Currently, with the various deficiencies inherited from the precipitation data sources of rain gauge measurements, radar estimates, and satellite estimates, high quality precipitation estimation is very difficult to derive. The PRISM (Parameter-elevation Relationships on Independence Slopes Model) interpolation method, for example, is based on the relationship of physiographic effects and rain gauge data. While PRISM can account for phenomena such as terrain-induced climate transitions, cold air drainage and inversions and coastal effects, it cannot adequately capture spatial variations in the orographic precipitation gradient during years when more than 70% of the precipitation occurred on days with extreme events, especially for areas with sparse networks of climate stations. To improve the model, we are developing a non-linear approach involving the applications of remote sensing and data mining techniques. This study focuses on analyzing the influence of the factors such as vegetation feedback, water vapor path, air temperature, wind direction, and wind speed on precipitation at various scales in addition to the geomorphologic factors used in PRISM. These predictor variables are measured with the accumulated downstream watershed covering downstream variability and seasonal fluctuation for their respective rain gauge stations. Preliminary analysis of this data reveals that the integrated factors of spring temperature, summer rainfall, and Leaf Area Index (LAI obtained from MODIS) in May seem to have moderate to weak relationships with the precipitation of the following winter and spring depending on resolution and location. More testing of these relationships at different scales are underway. To resolve the problem of unevenly distributed rain gauge stations, the Tropical Rainfall Measurement Mission (TRMM) 3B-42 data is downscaled based on MODIS Enhanced Vegetation Index and on water vapor data from the GOES satellite. The downscaled data is then bias-corrected based on the rain gauge data and converted to point data to fill the gaps in the areas where rain gauge stations are sparsely installed. This study is executed for two "atmospheric river" days, two summer days, and two spring days. The comparison of the spatial distribution pattern of the daily precipitation generated using local regression method for "atmospheric river" days and "non-atmospheric river" days aims not only to show the improvement of PRISM but also aims to discovering physical principles that can be used to develop a high-resolution daily precipitation model in the second phase.