Traditional spatial analysis and data mining methods fall short of extracting temporal information from data. This inability makes their use difficult to study changes and the associated mechanisms of many geographic phenomena of interest, for example, land-use. On the other hand, the growing availability of land-change data over multiple time intervals and longer time frames, often based on satellite imagery, presents to land-change study a great opportunity, given that this information can be effectively utilized. This methodological gap highlights the need to better understand the analytical challenges brought by temporal complexities, and to investigate alternative analytical frameworks that could handle those challenges. This dissertation attempted to achieve three goals: 1) finding metrics to capture temporal trends, 2) dealing with temporally imprecise data due to constraints of frequency, duration, and starting time of data collection, and 3) handling variables with time-changing values. A simulated land-change dataset based on an agent-based model of residential development and an empirical dataset from two case study sites in San Diego and Tijuana were used for this investigation. Results from the simulation dataset indicated that the survival function and the hazard function are important metrics to reveal temporal trends. In general the results of land-change analysis are sensitive to time frequency, in particular when time-dependent variables are also present. Longer duration benefits land-change analysis since longer durations contains more information. However, time-dependent variables with measures over a long period are more difficult for detection, which may pose a challenge. Starting time also affects the analytical results because the level of process uncertainty varies at different starting times. Findings from real world data mostly agree with those from computational data. Time dependent variables present a major challenge in land-change analysis, and survival analysis can better handle time-independent variables and thus better forecast urban growth.