|
Abstract:
|
Trend changes in vegetation give valuable information toward effective land use and development. In this research, vegetation trends are studied in the East African region based on the NDVI series from satellite remote sensing data collected between 1982 and 2006 over 8-kilometer grid points. In previous research, multiple testing procedures controlling the mixed directional false discovery rate (mdFDR) were used to detect areas with significant monotonic vegetation changes based on arbitrarily chosen square regions of land. This paper improves the detection procedures by first formulating as a temporal assignment problem. Due to the complexity of the formulation, a heuristic approach using dynamic programming and pixel reassignment was applied. Pixels were assigned to adjacent clusters based on similar behavior before applying the multiple testing methodologies. The results of this analysis find a larger number of detected regions than the previous research, while still maintaining control of the mdFDR under certain assumptions.
|
