A High-resolution Precipitation 2-step mapping Procedure (HiP2P): Development and application to a tropical mountainous area

Understanding the spatial and temporal variability of precipitation in tropical high mountain areas remains a key challenge. Point measurements are often not sufficient to capture the strong spatial variability particularly in mountain regions. Satellite remote sensing allows capturing the spatial heterogeneity of precipitation, yet it is generally characterized by significant bias. Rainfall satellite products such as those coming from the Tropical Rainfall Measuring Mission (TRMM) are being continuously improved and an increasing amount of high- and medium-resolution remote sensing data on biophysical surface properties is becoming available. Here we present a methodology that blends two TRMM products with remote sensing data on vegetation and topography to quantify the spatial distribution of precipitation in areas where direct observations are lacking. The approach assumes that vegetation cover, the topography and satellite-derived estimates of rainfall are reasonable indirect measures of ground-based precipitation. The methodology is evaluated for an area in the Andes of Ecuador. The results show that around 40% of the variance in weekly precipitation is explained by these proxies. During the drier periods of the year, vegetation is the strongest proxy. In the very wet areas and during the wet periods vegetation is usually in a climax development phase with no development trends to correlate with rain, and the other proxies dominate precipitation estimation. A cross-validation procedure in which each one of the weather stations is sequentially excluded from the analysis, was applied to test the performance of the methodology. The performance was satisfactory, and as expected it is related to the density of the weather station network and temporal rainfall variability. Overall we conclude that the methodology is useful for areas with very high variable conditions, where sufficient ground-data is available to establish the relationships with the different remote sensing datasets. (C) 2013 Elsevier Inc All rights reserved.