This paper addresses the problem of missing latent time series information caused by the differences in the analysis of time series data and non-time series data. A time series trend structure model (TSTM) was established using the analysis of time series patterns and rules, the trends of patterns and rules, and trends in confidence and support. Shandong Province was selected as the study area. Rainfall and evaporation time series data from this area were input into the TSTM. The results show that: (1) the structure of multi-year precipitation and evaporation trends of the meteorological stations in the study area have continuously increasing or decreasing characteristics. The TSTM can excavate the different trend structure characteristics of different meteorological elements and enables diversity in time series data analysis; (2) the evaporation trend structure tends to change synchronously with increases and decreases in precipitation and evaporation. The synchronous change frequency is essentially the same as that of the rainfall trend structure. This indicates that the TSTM has spatial and temporal characteristics for time series data analysis; and (3) from the maximal non-descending and non-ascending subsequence in the TSTM, it can be concluded that there exists continuity in the years when the trend structure of precipitation and evaporation increases and decreases synchronously. In addition, the degree of similarity in the model is well reflected in the spatial distribution characteristics of time series data, and the model provides clustering characteristics for time series data analysis. The TSTM proposed in this paper can effectively obtain the potential hydrological information contained in time series data, and provides a scientific and reliable basis for rules for the spatial optimization of watershed data and for the calibration of hydrological models.