Monitoring and prediction of LULCC through a combined MLPNN and Markov chain model in a mediterranean catchment
(2026) International Journal of Environmental Science and Technology — Vol. 23, n° 9 (2026)
Files
s13762-026-07387-8.pdf
Open Access - Adobe PDF
- 3 MB
Details
- Authors
Chniba, S 
- Abstract
- Mediterranean land use and land cover (LULC) systems are highly dynamic, reflecting continuous transformations driven by anthropogenic pressures and environmental gradients. Changes in LULC represent a key indicator of environmental conditions across spatial and temporal scales. This study aims to simulate future LULC dynamics in the Mediterranean Rmel watershed located in northeastern Tunisia by integrating a multi-layer perceptron neural network (MLPNN) with a Markov chain (MC) model within the TerrSet environment. LULC maps for 2001, 2014, and 2022 were generated from Landsat imagery using supervised Maximum Likelihood Classification, yielding high Kappa coefficients (0.81, 0.85, and 0.87 respectively). Model validation, based on the comparison between the predicted 2022 map and the observed reference map, confirmed the robustness of the modeling framework (Kstandard = 0.85, Klocation = 0.93, Kno = 0.87, and quantity disagreement = 0.06), supported by class level ROC/AUC values. Future projections were developed under three scenarios: (i) a business-as-usual (BAU) scenario reflecting historical trends, (ii) a conservation scenario emphasizing cereal expansion and stabilization of olive groves, and (iii) a reforestation scenario prioritizing forest and scrubland expansion. Although subject to increasing uncertainty over time, simulated trajectories for the 2040-2070 horizon suggest that changes will affect all classes, with significant variability observed mainly in olive groves, cereal crops, forest cover, and built-up areas. While these long-term simulations represent exploratory scenarios rather than deterministic forecasts, they provide valuable decision support insights for land use planning and sustainable watershed management in the Rmel catchment.
- Affiliations
Citations
Chniba, S., Mosbahi, M., Benabdallah, S., & Vanclooster, M. (2026). Monitoring and prediction of LULCC through a combined MLPNN and Markov chain model in a mediterranean catchment. International Journal of Environmental Science and Technology, 23(9). https://doi.org/10.1007/s13762-026-07387-8 (Original work published 2026)