Search

This study inter-compares the spatiotemporal features of precipitation extremes at global and regional scales within six daily precipitation datasets, i.e., gauge-based (Global Precipitation Climatology Center dataset, GPCC), satellite-retrieval (Precipitation Estimation From Remotely Sensed Information Using Artificial Neural Networks-Climate Data Record, PERSIANN-CDR), three reanalysis datasets (ERA-Interim, ERAI; the National Center for Atmospheric Research Reanalysis 2, NCEP2; and the WATCH-Forcing-Data-ERA-Interim, WFDEI), and products merged from the three type datasets (the Multi-Source Weighted-Ensemble Precipitation, MSWEP). All datasets reproduce similar spatial patterns of both annual and seasonal precipitation extremes over the period from 1979 to 2017. Compared to the reference dataset gauge-based GPCC, the reanalysis WFDEI outperforms among six products with spatial correlation coefficients of 0.89 and 0.80 for the annual extreme indices (i.e., annual total amount of 95th precipitation and maximum 1-day precipitation), respectively. The satellite-based product PERSIANN-CDR performs better than reanalyses and merged datasets in capturing the temporal variability of the intensity and amount of precipitation extremes with similar changing tendencies and magnitudes of about 45 mm day−1 and 230 mm at the global scale, respectively. The reanalyses and merged products underestimate the intensity of precipitation extremes. The selected six datasets behave differently in various regions. For the percentile-based frequency of precipitation extremes, NCEP2 performs well in regions of the Southeast Asian (SEA) and Amazon (AMZ), while WFDEI better matches GPCC over East North America (ENA) and North Australia (NAU) in both spatial patterns and temporal changes with correlation coefficients of 0.84 and 0.90, respectively. For the intensity features of annual precipitation extremes, NCEP2 performs better than other four datasets over regions of SEA, AMZ and West Africa (WAF). ERAI and WFDEI are consistent with GPCC in ENA and NAU with correlations coefficients of the intensity between ERAI (WFDEI) and GPCC are 0.82 (0.77) and 0.78 (0.64) for ENA and NAU, respectively. For the intensity of seasonal precipitation extremes, GPCC shows the highest estimates in regions of SEA, AMZ, ENA and WAF. ERAI and WFDEI perform better in reproducing the spatial patterns of seasonal precipitation extremes in all regions. NCEP2 (ERAI and WFDEI) show(s) consistent temporal variability of seasonal precipitation extremes with GPCC in regions of AMZ and WAF (ENA and NAU). Overall, there are large discrepancies in the absolute values of daily precipitation among datasets, and performances of non-gauged-based precipitation datasets in capturing the spatiotemporal variability of precipitation extremes are dependent on seasons, regions, and time periods. [ABSTRACT FROM AUTHOR]

Climate change and large-scale climate patterns may result in changes in probability distributions of climate variables that are associated with changes in the mean and variability, and severity of extreme climate events. In this paper, we applied a flexible framework based on the Bayesian spatiotemporal quantile (BSTQR) model to identify climate changes at different quantile levels and their teleconnections to large-scale climate patterns such as El Niño-Southern Oscillation (ENSO), Pacific Decadal Oscillation (PDO), North Atlantic Oscillation (NAO) and Pacific-North American (PNA). Using the BSTQR model with time (year) as a covariate, we estimated changes in Canadian winter precipitation and their uncertainties at different quantile levels. There were some stations in eastern Canada showing distributional changes in winter precipitation such as an increase in low quantiles but a decrease in high quantiles. Because quantile functions in the BSTQR model vary with space and time and assimilate spatiotemporal precipitation data, the BSTQR model produced much spatially smoother and less uncertain quantile changes than the classic regression without considering spatiotemporal correlations. Using the BSTQR model with five teleconnection indices (i.e., SOI, PDO, PNA, NP and NAO) as covariates, we investigated effects of large-scale climate patterns on Canadian winter precipitation at different quantile levels. Winter precipitation responses to these five teleconnections were found to occur differently at different quantile levels. Effects of five teleconnections on Canadian winter precipitation were stronger at low and high than at medium quantile levels. [ABSTRACT FROM AUTHOR]