LAUSR

A major challenge to climate-related research is the lack of a well-documented historical climate dataset. Unfortunately, most of the climate databases in developing countries have a critical data quality problem (missing data in time series and inhomogeneity). This study aimed to prepare a complete and homogenized climate dataset for the Tekeze river basin in Ethiopia and to assess the impact of detected inhomogeneity in major climate characteristics. Satellite and reanalysis climate products were used to fill in the missing data in the time series after a thorough evaluation and bias correction to the latter dataset. Together with other quality control approaches, the Multiple Analysis of Series for Homogenization (MASH) software was used for climate data homogenization. Results show that both minimum temperature (Tmin) and maximum temperature (Tmax) have a large number of breakpoints compared to rainfall. After homogenization, the original raw climate data was compared with the homogenized data. As the results show, the detected inhomogeneity causes a significant error to the region’s climate characteristics. For instance, following homogenization process, some stations have shown a change in trend magnitude from – 122 mm/decade to 0.2 mm/decade, from 1.2 °C/decade to 0.3 °C/decade, and from 1.5 °C/decade to 0.6 °C/decade for rainfall, Tmin, and Tmax respectively. The temperature trend resulted from the homogenized dataset is by far in agreement with trend results provided at the country level (nearly 0.3 °C to 0.4 °C/decade). Therefore, there is an urgent need of documenting a complete and homogenized historical climate dataset as a prerequisite for reliable climate-related studies.