A comparative study on 10 and 30-year simulation of CMIP5 decadal hindcast precipitation at catchment level

Abstract

Early prediction of precipitation has many positive benefits as it enables longer time for proper planning and decision making especially for the water managers, agricultural stakeholders, and policy and decision-makers. However, due to ongoing climate change along with the chaotic nature of precipitation, a too early prediction may lead to inefficient planning and decision making due to higher uncertainty and poor skills of the predicted data as the climate models are imperfect replicas that needs continuous improvement to predict future change. To investigate the difference between the short (a decade) and near-term (30 years) time simulation, this study aimed to compare the performance of 10 and 30-year simulation of CMIP5 decadal hindcast data of 0.05 degree spatial resolution at catchment level. For this, monthly hindcast precipitation of five general circulation models (GCMs); MIROC4h, MRI-CGCM3, MPI-ESM-LR, MIROC5 and CMCC-CM were downloaded from the CMIP5 data portal. Firstly the model data were cut for the Australian region and then the unit of the GCMs data was converted to the millimetre. In the next step, the GCMs data were spatially interpolated onto 0.05-degree spatial resolution using the second-order conservative method by Climate Data Operator (CDO) tool. Monthly observed gridded data of 0.05-degree spatial resolution were collected from the Australian Bureau of Meteorology (BoM). In the last step, both the observed and GCMs data were cut for the Brisbane River catchment in Queensland, Australia. Models’ performances are assessed comparing with the corresponding observed values through four skill tests; mean bias, mean absolute error, anomaly correlation coefficient and index of agreement. The results show that, 30-year simulations have comparatively higher mean bias and lower skills than 10-year simulated data that seems relevant to ensemble numbers and the external forcing from increasing GHGs due longer simulation period.