Necessity for postprocessing dynamically downscal

Authors: Qunying Luo

Publish Date: 2016/03/04

Volume: 30, Issue: 7, Pages: 1835-1850

PDF Link

Abstract

This work aims to answer if postprocessing climate model outputs will improve the accuracy of climate change impact assessment and adaptation evaluation To achieve this aim the daily outputs of CSIRO Conformal Cubic Atmospheric Model for periods 1960–1979 1980–1999 and 2046–2065 and observed daily climate data 1960–1979 1980–1999 were used by a stochastic weather generator the Long Ashton Research StationWeather Generator to construct long time series of local climate scenarios CSs The direct outputs of climate models DOCM and CSs were then fed into the Agricultural Production System sIMulator—Wheat model to calculate seasonal climate variables and production components at three locations spanning northern central and southern wheat production areas in New South Wales NSW Australia This study firstly compared the differences in climate variables and production components derived from DOCM and CSs against those from observed climate for period 1960–1979 The impact difference arising from the use of DOCM and CSs for the future period 2046–2065 was then quantified Simulation results show that 1 both the median/mean and distribution/variation of climate variables and production components associated with CSs were closer to those derived from observed climate when compared to those from DOCM in most of the cases median/mean distribution/variation climate variables production components and locations 2 the difference in the mean and distribution of climate variables and production components derived from DOCM and observed climate was significant in most of the cases 3 longer dry spells in both winter and spring were found from CSs across the three locations considered in comparison with those from DOCM 4 the median growing season GS rainfall total GS average maximum temperature GS average solar radiation GS length and final wheat yield were lower from DOCM than those from CSs and vice versa for GS rainfall frequency and GS average minimum temperature in 2055 5 the mean and distribution of these climate variables and production components arising from the use of DOCM and CSs are significantly different in most of the cases This implied that using the direct outputs of spatially downscaled general circulation model without implementing postprocessing procedures may lead to significant errors in projected climate impact and the identified effort in tackling climate change risk It is therefore highly recommended that postprocessing procedures be used in developing robust CSs for climate change impact assessment and adaptation evaluationThe author would like to thank Dr JL McGregor CSIRO Marine and Atmospheric Research for providing the outputs of the CCAM Dr M A Semenov Rothamsted Research UK for providing the LARSWG Dr Anne Colville University of Technology Sydney proof read this manuscript

Keywords: