Characterization of variability and trends in daily precipitation and temperature extremes in the Horn of Africa

Abstract

Modeling of extreme weather conditions is an important issue in environmental risk assessment, management and protection. In this paper, Annual and monthly maxima of daily temperature and rainfall data dating from 1901–2016 and 1950–2018 are characterized for nine countries within the greater horn of Africa. The generalized extreme value (GEV) distribution is fitted to the data sets from each of the nine countries using the method of maximum likelihood. Both the location and scale parameters of the GEV is formulated as a function of time to account for variability and trends in the extremes of temperature and rainfall so that their future behavior can be predicted. Based on our results, We provide return levels for the years 2, 10, 50 100, and 200 which could be used as measures of flood protection. To understand the spatial cross-correlation patterns on different time scales in each country and how rainfall and temperature are related to agricultural variables, we utilize the detrended cross-correlation coefficient(ρDCCA) and its generalization, (DMCx2) to account for that, respectively. Given the results of the (DMCx2) and the fact that climate extremes pose a huge threat to agriculture and food security, we employed copula based models to describe the structure of dependence between climatic variables and the crop related variables such as yield and production quantity. The results from the copula analysis show that irrespective of the country, climatic factors and the agricultural products(production/yield) the strongest dependence is demonstrated by the pairs involving cereal crops, while the weakest dependence is characterized by the pairs involving regular potato.