Evaluation of Reanalysis Precipitation Data and Potential Bias Correction Methods for Use in Data-Scarce Areas

Daily precipitation is a crucial input for water resources modeling and an important basis for storm-related planning. In East Africa, the scarcity of good quality, continuous rainfall and other climate data is a challenge that climate and hydrology researchers have frequently faced. A comparison of Climate Forecast System Reanalysis (CFSR) precipitation data to available observed data from stations in the East African countries Kenya, Uganda, and Tanzania showed notable differences between the two datasets, particularly with respect to precipitation totals and number of days receiving rainfall. The CFSR dataset covers the time period 1979-2010, and the CFSR version 2 (V2) dataset is a similar product covering the period from 2011 to present day, with the effective cutoff being 2019. Bias correction techniques generally fall under two categories: scaling and distribution adjustment. Examples of scaling techniques include linear scaling, local intensity scaling, and power transformation, while methods such as empirical quantile mapping and daily translation involve modifications to existing distributions. For this study, power transformation, local intensity scaling, and empirical quantile mapping were determined to be the best prospects based on the completeness of the raw datasets and the data needs and distribution-related limitations of the different methods. The aim of this study was to evaluate CFSR data as a substitute for on-site measurements in data-scarce areas and identify bias correction methods that would be appropriate to improve the accuracy of the substitute dataset. Specifically, to determine the extent of discrepancies between the observed data and corresponding CFSR data, identify the most suitable bias correction method using available data in the study region, and develop bias-corrected datasets for select stations within the study region. The study focused on the East African countries Kenya, Uganda, and Tanzania, where limited locations have reasonably detailed weather records. The region's geomorphology has a strong influence on the climate conditions experienced at a local level, with the mountain and valley region (Great Rift Valley) cutting through Kenya and Tanzania, the Lake Victoria region, and distinct plateaus, all of which comprise roughly half of the study area. Kenya is fairly temperate in the southeast, but to the northeast, the climate is more arid. Bordering the Indian Ocean, coastal areas in Kenya and Tanzania are characterized by heat and humidity. The remaining majority of Tanzania can be characterized as a tropical or subtropical plateau with mild weather, with altitude as the main driver of temperature variation. Uganda also consists mainly of temperate tropical plateaus but is warmer, particularly during dry periods. Average annual precipitation in Kenya ranges from less than 500 mm in the north to over 2,000 mm in the southeast. The study used the R environment for all statistical calculations, bias corrections, and comparisons, selected because it is a reliable software that provides free access to powerful tools. The study targeted observed datasets that were readily available and easily accessible to anyone with internet access. The results of the study indicate that bias-corrected CFSR precipitation data provides an improved basis for water resources applications in the study region. Methodologies and approaches are extendable to other data-scarce regions or areas where complete and consistent data are not easily accessible.