In order to solve the problem of analyzing the temporal and spatial characteristics of water yield and runoff in the absence of data, various methods, including SWAT model, multivariate statistical analysis, correlation coefficient method, and elastic analysis method, were used to analyze the spatial and temporal variations of water yield and runoff of each section in the Taohe River Basin, so as to reveal the relationship among precipitation (P), water yield (Y) and runoff (R), and the influencing factors of runoff change in regions with scarce data. The results showed that the runoff of Hongqi Station in theTaohe River Basin decreased significantly overall from 1956 to 2019, and an abrupt change occurred in 1987. The determination coefficients of SWAT model were 0.78 and 0.79, and Nash-Suttcliffe coefficients were 0.78 and 0.77 in the period of validation and calibration, respectively, and all the relative errors were less than 10%, so the simulation results were satisfactory. The middle reaches accounted for the largest proportion of runoff (50.71%), followed by the upper reaches (37.21%) and the lower reaches (12.08%), therefore, the middle reaches were the main source of the runoff. The water yield decreased from the upstream to the downstream, which was 214.18, 179.43 and 81.16 mm, respectively. The water yield in the Daisang River Basin was the largest, whereas that in the Chahe River and Dongyugou River were the smallest. The correlation order was P-Y>Y-R>P-R, and the correlation gradually decreased from the upstream to the downstream. The impacts of climate change and human activities were 20% and 80%, respectively, and human activities were the main influencing factor. The research results are helpful for the study of runoff variation and its influencing mechanism in each section of the Taohe River Basin, and provide theoretical support for spatio-temporal analysis of runoff in regions with scarce data.