Key points are not available for this paper at this time.
The application of global datasets in climate and hydrological studies is growing, yet uncertainty regarding their use across catchments remains a concern. Malawi, a landlocked country in southeastern Africa, has sparse climate-station coverage, hindering the monitoring of climate variability, extremes, and related hydrological assessment. This study explores the added value of utilizing existing global climate datasets when station observations are limited or unavailable. Three high-resolution gridded datasets are assessed in the study: CHIRPS, PERSIANN-CDR, and ERA5-Land. The comparison encompasses climatological means (annual, seasonal, and monthly) and assessment of the temporal variability in daily times series of temperature and rainfall over Malawi using well-known performance metrics: Kling-Gupta Efficiency (KGE) and Pearson correlation coefficient. The analysis is conducted for Malawi as a whole and for the four climate regions, during the period 1985 to 2015. Results indicate that all gridded datasets exhibit a good correlation with station observations, with KGE values reaching 0.55 or higher. Notably, ERA5-Land outperformed the others, achieving a median KGE of 0.59 for rainfall and 0.77 for temperature on a daily scale. Performance varied regionally and with season: ERA5-Land outperformed the two other datasets during both the wet and dry seasons, while CHIRPS showed less bias and better correlation on an annual scale. PERSIANN-CDR showed the lowest performance overall, especially annually with slightly better performance at the monthly timescale. This study enhances our understanding of the strengths and limitations of global climate datasets for climate impact research, and provides valuable insights for water management studies, particularly hydrological modelling. • Three global climate datasets (ERA5-Land, CHIRPS, PERSIANN-CDR) were evaluated against station observations in Malawi. • ERA5-Land had the best daily rainfall performance and PERSIANN-CDR lowest; no dataset was best across regions/timescales. • ERA5-Land further demonstrated high temperature skills and is thus recommended for further studies.
Nkhoma et al. (Wed,) studied this question.