Assessing Effects Of Climate Variability And Change On The Onset And Cessation Dates Of Seasonal Rainfall In Kenya

Abstract

Climate change has great impact on various sectors that are dependent on rainfall. As a result of increased global warming due to continuous emission of greenhouse gases into the atmosphere, the influence of climate change on human activities is great especially those dependent on rainfall for agriculture production. The IPCC fifth assessment report indicated that increase in the frequency and/or magnitude of climate related disasters such as flood and droughts could be linked to some of the impacts that climate change impose on human activities around the world. This study has therefore focused on assessing effects of climate change on the onset and cessation dates of seasonal rainfall over Kenya. To achieve this, the study aimed at determining the current climate variability and trend of onset and cessation dates based on gauged observations, evaluating the skill of CORDEX RCA4 models in simulating onset and cessation dates and determining the future climate variability and trends of onset and cessation dates from CORDEX model projections for the RCP4.5 and RCP8.5. Onset date was defined as 1st day from 1st March and 1st October for long and short rains respectively with at least a total rainfall of 20 mm while cessation was defined as the date after 1st May for MAM and 1st December for OND when the supply of water to the soil becomes null and after which there is no rainfall for the next ten days. The study used CHIRPS and CORDEX data sets to determine the past and future variability and trends of onset and cessation dates for seasonal rainfall over Kenya. The CHIRPS data period was from 1981-2010 and CORDEX data was from 1976-2005 for baseline and 2021-2050 for projections. The CORDEX data used were from CNRM, ICHEC, MPI and their ensemble. Future variability and trends onset and cessation dates were determined based on RCP4.5 and RCP8.5 scenarios for the period 2021-2050. Mean and standard deviation were calculated to determine both current and future variability of onset and cessation dates. Mann Kendall was used to determine the trends of onset and cessation dates. Root Mean Square Error (RMSE) and Pearson correlation coefficient methods were used to assess the skills of the models. The results showed high variability of onset dates (ranging from 5-17 days) compared to cessation dates (ranging from 4-15 days) for both MAM and OND seasons for the period 1981-2010. The results further showed high variability in MAM season compared to OND season for both onset and cessation dates for the period 1981-2010. Skill assessment results using RMSE method showed low RMSE values for ensemble model compared to individual CNRM, ICHEC and MPI model in simulating onset and cessation dates of seasonal rainfall. The performance of ensemble model based on Pearson correlation showed varying correlation coefficients values compared to other three GCM models. Going into the future, the results showed that the onset dates have higher variability than the cessation dates for both MAM and OND seasonal rainfall based on both RCP4.5 and RCP8.5. Results also showed that based on the RCP4.5 projections, more regions will experience insignificant negative trends for both MAM and OND onset dates compared to RCP8.5 projects that showed that more regions will experience insignificant positive trend for the MAM and OND onset dates. Therefore, with these kind of results, there is need for climate users like farmers to work closely with the climate producers through consultation in order to have a clear information on when seasonal rainfall will begin and end.