Use of CLIGEN to simulate climate change in Southeastern Australia

Abstract

CLIGEN is a stochastic weather generator to reproduce, statistically, daily weather variables, and CLIGEN output has been used to simulate the impact of climate change on runoff and soil erosion. Weather observations from regions with significant climate variability and change could be used to determine how to manipulate the input to or output from CLIGEN to simulate climate change scenarios. Previous studies mostly used simplistic approaches, such as changing the average daily rainfall on wet days by a fixed percentage or multiplying the CLIGEN-generated daily rainfall by a fixed factor. The aim of this article is to develop a method based on available historical data to adjust CLIGEN parameter values when, historically, rainfall has significantly changed. In southeastern Australia, rainfall showed a significant and abrupt increase in a 30-year period since the late 1940s from the preceding three decades. However, rainfall has decreased since the late 1970s, significantly at many sites in the same region. Long-term (90 years) daily rainfall data from 30 sites in this region were used to examine decadal variations in rainfall and to evaluate the changes to CLIGEN parameter values with significant changes in annual rainfall. Average daily rainfall, standard deviations, skewness coefficients, and probabilities of a wet day following a wet day and a wet day following a dry day were analyzed for each of three 30-year periods and for each of 30sites in southeastern Australia. This article shows that rainfall data for the period from 1919 to 1978 would suggest an increase in rainfall in southeastern Australia. However, from the perspective of the period from 1949 to 2008, the conclusion of decreasing rainfall would be reached. Both these 60-year periods broadly coincide with an underlying trend of increased temperature in Australia and globally. Daily rainfall data for the 90-year period show that there are strong positive correlations between changes in mean monthly rainfall and changes in mean daily rainfall, standard deviation, and the probability of wet-following-dry sequences. There is little evidence to suggest ways of adjusting skewness coefficients or wet-following-wet probabilities to simulate changes in mean monthly rainfall for this region. A set of regression equations was developed to allow easy adjustment of CLIGEN parameter values to simulate monthly rainfall change for both increasing and decreasing rainfall change scenarios. The results show that when CLIGEN parameter values were adjusted using changes in monthly rainfalls and regional relationships for the three important parameter values, output from CLIGEN was able to reproduce the changes in rainfall when compared with historical observations. The proposed methodology for adjusting CLIGEN parameters is not site-specific and could also be used for other similar regions in the world.