Characterizing Location and Classification Error Patterns in Time-Series Thematic Maps

The goals of this letter are to study the spatial patterns of location error and of classification error in spatio-temporal datasets, assess the role of environmental factors in determining error rate and pattern, and test for possible correlation within and between these error types in space and time. A multiple regression was used to determine the effects of local environmental factors (topography, vegetation cover) on each error type. Topographic structure and vegetation cover had significant effects on location error, where larger error was associated with north-facing aspects, steeper slopes and woody vegetation cover. Classification error was also affected by topography, and vegetation cover. Slope was the major factor that affected classification quality. Strong correlation was found between error in different time steps, for both error types. Correlation between these two error types in the same time step was much smaller and in most cases insignificant.