The impacts of agricultural expansion on wetlands are diverse and complex. Land surface temperature (LST) has a great potential to act as a global indicator of the status of wetlands and changes in their hydrological and evapotranspiration regimes, which are often linked to land use and cover changes. We use the whole MODIS LST archive (2000–2017) to perform time series analysis in the Kilombero catchment, Tanzania; a large wetland that has experienced major land conversions to agriculture during the last two decades. We estimated pixel based trends using three models: a seasonal trend model, and aggregated time series using annual means and percentile 90. We characterized the trends found by using land cover change maps derived from Landsat imagery and a post-classification comparison. The relation between Normalized Difference Vegetation Index (NDVI) and LST trends was also studied (r =−0.56). The results given by the seasonal trend model and annual means were similar (r = 0.81). Fewer significant trends were found using the percentile 90, and these had larger magnitudes. Positive LST trends (i.e. increasing) corresponded to deforestation and farmland expansion into the floodplain, while forestation processes resulted in negative LST trends. Moderate increases of LST in natural wetlands suggest that the impacts of human activities extend also into non-cultivated areas. We provide evidence of how time series analysis of LST data can be successfully used to monitor and study changes in wetland ecosystems at regional and local scales.
|Number of pages||10|
|Journal||International Journal of Applied Earth Observation and Geoinformation (JAG)|
|Early online date||25 Apr 2018|
|Publication status||Published - Aug 2018|
- Wise use
- time series