Multiscale assessment of spatial precipitation variability over complex mountain terrain using a high-resolution spatiotemporal wavelet reconstruction method

Studying precipitation variability in the Peruvian Andes is a challenge given the high topographic variability and the scarcity of weather stations. Yet previous research has shown that a near-linear relationship exists between precipitation and vegetation in the semiarid central Andes. We exploit this relationship by developing a new, spatially highly resolved spatiotemporal precipitation reconstruction method, using daily precipitation time series from in situ weather stations, and dekadal (10 calendar days) normalized difference vegetation index (NDVI) fields. The two data sets are combined through a wavelet decomposition method. A 4° × 4° region around Quelccaya ice cap (QIC), the world’s largest tropical ice cap located in the central Peruvian Andes, was selected as study area, due to its importance for climatic, glaciologic, and paleoclimatic research. The reconstructed end product, a ~1 km² gridded precipitation data set at dekadal temporal resolution, was validated against independent rain gauge data and compared with the Tropical Rainfall Measuring Mission (TRMM) 3B42 version 7 product. This validation showed a better overall performance of our own reconstruction than the TRMM data. Additionally, a comparison of our precipitation product with snowfall measurements at the QIC summit (5670 m) shows a regionally coherent signal at the dekadal scale, suggesting that the precipitation falling at QIC is driven by regional- rather than local-scale convective activity. We anticipate that this methodology and the type of data generated in this study will be useful for hydrological and glaciological studies, as well as for validation of high-resolution downscaling products in mountain regions.