资源环境科技发展态势分析平台

The interseasonal variability of two basic climatic parameters (precipitation and temperature) is assessed, over vulnerable and data sparse region of Pakistan (23 degrees to 37 degrees N and 60 degrees to 75 degrees E), for two Coupled Model Intercomparison Project 3 (CMIP3) based Atmospheric-Oceanic General Circulation Model (AOGCM) versions: CM2.0 and CM2.1 by Geophysical Fluid Dynamics Laboratory (GFDL), and two CMIP5 based AOGCM versions: CM2p1 and CM3.0. A recent historical 50-year period (1951-2000) is analyzed and compared with APH-RODITE for precipitation and National Centers for Environmental Prediction/National Center for Atmospheric Research (NCEP/NCAR) re-analysis based gridded data-sets for temperature for the following four seasons: DJF, MAM, JJA, and SON. The study area is divided into three regions: all Pakistan, northern Pakistan, and southern Pakistan. The interseasonal variability of the precipitation and the temperature are derived from all three (five) runs of CM2.0 (CM2.1) and from all ten (five) runs of CM 2p1 (CM3.0). The bias, root mean square error (RMSE), one-sigma standard deviation (SD) and correlation coefficient (CC) are used as assessing metrics. The following individual runs have positive CC with respect to APH-RODITE at <= 1% Confidence Level (CL). On seasonal basis for CMIP5 based GFDL models during DJF: CM2p1R5 (for all Pakistan), CM2p1R5 (for northern Pakistan), and during MAM: CM2p1R5 (for southern Pakistan; this run has the lowest centered RMSE of 0.11 mm/day), whereas on annual basis: CM3.0R3 (for all Pakistan). However, out of these four runs, only CM2p1 (for southern Pakistan) has SD < SDobs (0.08 < 0.12 mm/day). There are 13 other runs for which the positive CC is at <= 5% CL, both relative to observed precipitation or temperature. Out of these 13 runs, only the average of runs of GFDL-CM2.1 in CMIP3 in JJA in southern Pakistan has SD < SDobs (0.56 < 0.59 degrees C) with a centered RMSE value of 0.65 degrees C. These characteristics of GFDL-CM2p1 runs are supported by their relatively better simulation of spatial distribution of 1000-850 hPa averaged layer wind patterns, relative to NCEP/NCAR 1000-850 hPa averaged wind patterns, over Pakistan, in respective seasons. A variance based bias adjustment when applied displays considerable interseasonal bias reduction both in precipitation and temperature in long term mean with no change in trend.