When simulating a daily performance of a solar thermal powered adsorption refrigeration system, daily weather data (i.e. daily maximum and minimum ambient temperatures and daily total solar radiation) are normally used to simulate a 24-h refrigeration cycle. However, the simulation obtained in this way is not that accurate, due to which could not reflect real hourly processes during the daytime. In present study, hourly weather data, especially during the isosteric and desorption processes in the sunshine duration, has been utilised to simulate hourly accumulated daily cycles. Seven possible cycles, depending on hourly weather change patterns, have been identified in this study. It has been found that the desorption process is not necessarily an isobar process as previous studies assumed, which depends on real hourly weather changes. Therefore, an improved mathematical model has been developed and validated. It has been found that the accuracy of the simulation with hourly weather data has been improved by up to 5%, compared to that of the model using daily weather data. The simulated error in one-day performance would be accumulated if a long-time performance of such a system was required to be simulated. Therefore, a case-study is conducted to demonstrate the difference between the two models when simulating a daily performance under three typical weather conditions.