- English
Flood-dominated floodplains are undergoing rapid land-use and land-cover (LULC)
change, with consequences for exposure, ecosystem condition, and local development
planning. This study examines historical LULC change, transition intensity, trend-consistent
land-use scenarios, and event-based flood exposure in Bongaigaon District, located
in the lower Brahmaputra floodplain of Assam, India. Landsat imagery for 1990, 2000,
2010, and 2022 was classified in Google Earth Engine using a Random Forest classifier.
Three-level intensity analysis was used to examine interval, category, and transition-level
change. CA-Markov scenarios were simulated for 2032, 2042, and 2052, and the 2022
LULC map was overlaid with the major 2022 inundation layer to assess event-based exposure.
Classification accuracy ranged from 85.2% to 90.4%, with class-wise F1 scores
exceeding 80%. Between 1990 and 2022, built-up land increased markedly, while vegetation
declined substantially. Intensity analysis showed faster landscape reorganisation
during 1990–2000 and 2000–2010, followed by more selective change during 2010–2022.
The 2022 overlay showed that built-up and agricultural land coincided with the very high
inundation class, but this result should be interpreted strictly as event-based exposure for a
single observed flood event. The study provides a district-scale evidence base for interpreting
land transition, event-based exposure, and selected SDG-relevant proxy indicators. It
supports screening-level land-use discussion, but does not constitute a long-term flood-risk
assessment or an official measurement of SDG progress.
- English
