Joint Modelling of Drought Severity and Duration using Copula Theory: A Case Study of Ghana

Abstract

Analysing and understanding the occurrence and development of droughts is of great significance in mitigating drought impacts. This study assessed the possible changes in the joint distribution of drought duration and severity in two major cities of Ghana, Accra and Yendi. The duration and severity of droughts, estimated using the Standardized Precipitation Index (SPI), were determined based on run theory. The best-fitted Copula models were used to combine the drought duration and severity to analyse the drought return period. The gamma, lognormal and Weibull distributions were considered to select the marginal distributions for the duration and severity, while the normal, t, Gumbel, Joe, Clayton and Frank copulas to select the best-fit Copula model. Bias corrected climate simulations of six Global Climate Models (GCMs) of the CMIP6 were used to project drought characteristics for the near and far futures. The results showed the Clayton and Frank copulas as the most suitable for fitting the joint distribution of drought duration and severity at Accra and Yendi, respectively. Lognormal and Weibull distributions were the most suitable for the marginal distributions of severity and duration, respectively. The joint return periods of droughts showed almost no change in the future compared to the historical period in Accra with a historical mean of 11.36 and a near and far future mean of 12.26 and 10.30 respectively but significantly reduced return periods in the future in Yendi with a near and far future means of 1.47 and 2.13 respectively compared to a historical mean of 17.40. The drought risks estimated for different future periods can provide useful information in planning, management, and assessing the adequacy of the water structures in the region.