Comparison of Multiple Datasets with Gridded Precipitation Observations over the Tibetan Plateau

Qinglong You,Jinzhong Min,Wei Zhang,Nick Pepin,Shichang Kang

DOI: https://doi.org/10.1007/s00382-014-2310-6

IF: 4.901

2014-01-01

Climate Dynamics

Abstract:Precipitation is a critical component of the water balance, and hence its variability is critical for cryospheric and climate change in the Tibetan Plateau (TP). Mean annual and seasonal precipitation totals are compared between gridded observations interpolated to a high resolution (0.5° × 0.5°) and multiple reanalysis type-datasets during 1979–2001. The latter include two NCEP reanalyses (NCEP1 and NCEP2), two European Centre for Medium-Range Weather Forecasts (ECMWF) reanalyses (ERA-40 and ERA-Interim), three modern reanalyses [the twentieth century reanalysis (20century), MERRA and CFSR] and three merged analysis datasets (CMAP1, CMAP2 and GPCP). Observations show an increase in mean precipitation from the northwestern to the southeastern (SE) regions of the TP which are divided by an isohyet of 400 mm, and overall trends during the studied period are positive. Compared with observations, most of the datasets (NCEP1, NCEP2, CMAP1, CMAP2, ERA-Interim, ERA-40, GPCP, 20century, MERRA and CFSR) can both broadly capture the spatial distributions and identify temporal patterns and variabilities of mean precipitation. However, most multi-datasets overestimate precipitation especially in the SE where summer convection is dominant. There remain substantial disagreements and large discrepancies in precipitation trends due to differences in assimilation systems between datasets. Taylor diagrams are used to show the correlation coefficients, standard deviation, and root-mean-square difference of precipitation totals between interpolated observations and assimilated values on an annual and seasonal basis. Merged analysis data (CMAP1 and CMAP2) agree with observations more closely than reanalyses. Thus not all datasets are equally biased and choice of dataset is important.