Variability in land-surface precipitation estimates over 100-plus years with emphasis on mountainous regions

Understanding the spatial and temporal variability of land-surface precipitation is prerequisite to assessing the potential impacts of climate change on water resources. Our current knowledge of land-surface precipitation variability and change, however, is uncertain (IPCC, 2007; Nickl et al, 2010). An important factor contributing to this uncertainty is the low spatial density of weather stations in regions with complex terrain (e.g. mountainous regions) which makes it difficult to estimate the spatial distributions of precipitation and precipitation change. This problem calls for the development of a spatial-interpolation method that better takes into account topography.;Three sets of available estimates of the spatial and temporal patterns of annual land-surface precipitation (produced by the Climate Research Unit, CRU; the University of Delaware, UDel; the Global Precipitation Climate Center, GPCC; and resolved at 0.5° grid resolution) are evaluated, analyzed and compared for the period 1901-2008. Large errors within and substantial differences among precipitation variability estimates contained in the three sets of estimates are apparent, especially within mountainous regions. A new spatial interpolation method for precipitation is developed and evaluated. My approach to interpolation makes use of horizontal-distance-and-direction influences (from nearby raingage-station observations) as well as topographic influences, including estimates of optimal orographic scales and orographic regions. An evaluation of this new method is performed for a region of high topographic and precipitation variability (the San Joaquin Valley in Western United States). Results are encouraging and show more realistic representations of precipitation variability and relatively low errors over areas with higher topographic variability.;Using this new spatial interpolation procedure, monthly and annual fields of land-surface precipitation are re-estimated for the period 1901-2008 in order to improve our estimates and understanding of the spatial and temporal variability and changes in land-surface precipitation since 1900. The representation of the spatial variability of land-surface precipitation is enhanced, especially over mountainous regions. Finally, the new annual land-surface precipitation estimates are used to reanalyze changes in annual land-surface precipitation for the period 1901-2008 and are compared with the changes estimated from the CRU, UDel and GPCC datasets. The re-estimated land-surface mean is higher than the others and the estimated year of major change is 1975. Within the new dataset, only a slight change during the period 1950-1975 is apparent, but a dramatic decrease over the period 1975-2008 stands out, especially within mountainous regions.