Dynamic land use classifications and hydrological simulation in African cultivated areas

It is important to represent seasonal dynamics in agro-hydrological models to properly represent the spatial land dynamics of hydrological and agricultural processes. Agro-hydrological models can represent these spatial dynamics, e.g. through crop rotations, but this is rarely implemented especially in Africa, even though often up to three cropping seasons may happen within a year. In this paper, we produced three seasonal land use maps from remote sensing to derive dynamic and static trajectories for the Kikuletwa basin that is a predominantly cultivated catchment with clear distinct seasons (dry and wet) and up to three cropping seasons. Together with information obtained from crop calendars, we implemented decision tables to represent the crop phenology and its management schedule for the dominant trajectory of each land-use class. A comparison between the default SWAT+ set up and a seasonal dynamic SWAT+ model is done for the spatial mapping of Evapotranspiration (ET) and Leaf Area Index (LAI). The results indicated an improved vegetation simulation when dynamic land use is implemented. The LAI and ET dynamics of the trajectory implementations showed more realistic temporal advancement patterns that corresponded to the rainfall pattern within the basin and in agreement with the remote sensing products as compared to the default model. The results demonstrate the usefulness of seasonal dynamic land-use maps in the improvement of hydrological simulation and further lead to a better calibrated model.