High-Resolution Precipitation Gridded Dataset on the South-Central Zone (34° S–41° S) of Chile

Abstract

Having a distributed gridded precipitation product responds to a multitude of scientific field’s needs: climatology, hydrology, biology, and socioeconomic, among many others. When the insufficient spatial distribution of the rainfall network does not allow (for either lack of quality, spatial or temporal resolution) the correct study of hydro climatic variability, grilled data resulting from different sources (satellite, in situ, models, et al.) are used. Although there are many precipitation grids worldwide they do not respond to high spatial-temporal variability in regions with complex topography, and even more so, along western South America, and central Chile, where the hydroclimate and topography is modulated by Coastal and Andes cordillera. To provide and improve the accessibility of meteorological data on the area, we developed a dynamic geostatistical method to build a high resolution (~800 m) monthly gridded precipitation product for the central-southern zone of Chile for the period 2000-2011. The product involves data from high resolution regional atmospheric modeling, global precipitation grid sets, and 136 in-situ observation stations. The central-southern zone of Chile ($34\cc S-41\cc S$) encompasses from O'Higgins to Los Rios regions, including the Rapel, Mataquito, Maule, Itata, BioBío, Imperial, Toltén and Valdivia hydrologic basins which have different topographies and distinct climates. Thus, for its development, varying spatio-temporal multiple linear regression techniques are applied over different geographical domains where topographic descriptors variables are used as independent variables (elevation, slope, exposure and continentality). Results show that most of the precipitation spatial-temporal variability is well-captured by the model in the north and central regions, from O'Higgins to Biobío, with goodness-of-fit (R^2) fluctuating around 0.86 and 0.82 respectively. Toward the South, Araucanía and Los Ríos, goodness-of-fit (R^2) decrease to values around 0.74 and 0.72 respectively. Both the modified Willmott coefficient (d) and the nse indicated a good model skill, with values over 0.8 and 0.7 respectively. Meanwhile, the $\sigma_e$, $nrmse$ and $pbias$ changed between 0.04-0.2, 0.35-0.52, and 12\%-22\%, respectively. This database is freely available to different regional or national institutions and will help the development of better understanding and management of local and regional hydrology.