3.2 Prediction of forest cover change
Based on population density, physiography, accessibility and other factors (Table 3) along with transition probabilities and spatial trend, this study predicts forest cover scenario for 2023 and 2027. Fig. 5 shows observed versus simulated LULC categories of 2019. The result clearly demonstrates performance of Markov CA approach in simulating LULC. The accuracy of the prediction showed overall accuracy and kappa of 86.21% and 0.85%, suggesting a good performance of the model. However, poor simulation was achieved for landcover of mixed forest and degraded forest categories whilst best agreement was obtained for agriculture, urban, homestead vegetation categories.
Spatial pattern of land use/covers during 2023 and 2027 is shown in Fig. 6, which indicated that the distribution of degraded forest would be widespread, if Rohingya camps exist in the peninsula at the expanse of dominant land covers (e.g., shrubs, mixed forest, plantation forest and canopy forest). Specifically, shrubs land cover is expected to decline from 7,306 ha in 2019 to 5,800 and 4,871 ha in 2023 and 2027. Other land covers such as mixed forest, planted trees and canopy forest would reduce significantly as well (Table 7). Conversely, a substantial increase in degraded forest is highly likely during two years (e.g., 2023 and 2027) though a subtle increase is seen in agriculture and camp land covers (Table 7).
Forest degradation, as a function of fuelwood collection, illegal logging and other activities, was determined based on predicted LULC of 2023 and 2027. The analysis revealed that loss of forest cover would increase dramatically, if present rate of anthropogenic activities continues in the study area. Since addition of refugee is not expected due to host country’s repeated denial, it is seen that shrubs, mixed forest, plantation forest and canopy forest would experience massive reduction of which loss of shrubs and mixed forest could be substantial (Table 8).