Abstract: This study aims to improve the quality of long-term hourly observations at regional automatic observation stations and enhance their application value in coping with and adapting to climate change. To this end, homogeneous hourly temperature data from regional automatic weather stations during 2008–2023 are developed, taking the Tianjin area as the study site. This dataset objectively eliminates erroneous values from the observational time series and effectively mitigates and evaluates the systematic bias caused by non-climatic factors such as station relocation, instrument upgrades, and collector renewal on the time series. First, in the data quality control process, 148 errors at 15 stations were detected through internal consistency checks for three temperature variables and were set to missing values or corrected through manual verification. In addition, 19 errors at 3 stations, 21 errors at 18 stations, and 66 errors at 31 stations were set to missing values based on climate outlier detection for hourly mean temperature, maximum temperature, and minimum temperature, respectively. Correspondingly, from the spatial consistency results, 5 errors at 3 stations, 2 errors at 2 stations, and 1 error at 1 station were set to missing values. Second, in the homogenization analysis, 6 stations with statistically significant breakpoints supported by accurate metadata were eliminated using the penalized maximal F test (PMFT) combined with station metadata. Finally, the developed hourly temperature data from regional automatic observation stations were verified as relatively reliable by comparison with corresponding observations from national stations across 11 administrative regions in Tianjin. In addition, analysis of hourly data from 106 regional automatic observation stations showed that the increasing trend and amplitude of the maximum temperature in Tianjin during the recent 10 years were relatively the most significant, followed by the mean temperature, while the minimum temperature showed the least increase. In particularly, during autumn, 100% of the regional automatic observation stations showed an increasing trend in maximum temperature, among which 63.2% were statistically at 0.05 significance level, with amplitudes ranging from 1.316°C to 3.760°C over the last 10 years.