Title Climatologies at high resolution for the Earth land surface areas Authors

High resolution information of climatic conditions is essential to many application in environmental sciences. Here we present the CHELSA algorithm to downscale temperature and precipitation estimates from the European Centre for Medium-Range Weather Forecast (ECMWF) climatic reanalysis interim (ERA-Interim) to a high resolution of 30 arc sec. The algorithm for temperature is based on a statistical downscaling of atmospheric temperature from the ERA-Interim climatic reanalysis. The precipitation algorithm incorporates orographic predictors such as wind fields, valley exposition, and boundary layer height, and a bias correction using Global Precipitation Climatology Center (GPCC) gridded and Global Historical Climate Network (GHCN) station data. The resulting data consist of a monthly temperature and precipitation climatology for the years 1979-2013. We present a comparison of data derived from the CHELSA algorithm with two other high resolution gridded products with overlapping temporal resolution (Tropical Rain Measuring Mission (TRMM) for precipitation, Moderate Resolution Imaging Spectroradiometer (MODIS) for temperature) and station data from the Global Historical Climate Network (GHCN). We show that the climatological data from CHELSA has a similar accuracy to other products for temperature, but that the predictions of orographic precipitation patterns are both better and at a high spatial resolution. Background & Summary High resolution climate data are essential to many applications in environmental sciences. While many studies in these fields are conducted at a resolutions of ~1km, state of the art climate reanalyses often only represent climatic variations at spatial resolutions of 0.5° 1° (ca. 25 – 100 km at the equator) at a global scale. The gap between these spatial scales is often regionally bridged using satellite data, via statistical downscaling, or interpolation methods, but climatologies based on statistical downscaling are not currently available on a global scale due to numerous methodological challenges. While interpolated datasets are available, they often fail to accurately predict certain factors such as precipitation in highly variable terrain. To achieve a finer resolution, interpolation and regression techniques either use data from local climate observations such as the Global Historical Climate Network (GHCN) and combine them with atmospheric predictors from gridded climate reanalyses