3.1 Spatiotemporal distribution of LULC
Fig. 2 (a c) illustrates LULC categories, and they are useful to identify state of forest cover in the study area. Inspection of individual land use/cover categories, derived from Sentinel images, revealed pre and post influx situations. For example, loss of forested land was largely distributed along the Inani National Park and TWS before influx of Rohingya (Fig. 2a), however, the distribution of degraded forest, defined by poor vegetation health, increased substantially in the subsequent years (e.g., 2018 and 2019) (Fig. 2b c).
Temporal information of LULC data over the study area showed that four land covers, viz., homestead vegetation, shrubs, mixed and canopy forests, experienced a significant decline from 2017 to 2019, and a manifold increase in two human dominated land covers, viz., camps and degraded forest (Table 4). For instance, mixed forest cover declined from 10,593 ha in 2017 to 9,645 ha (2018) and 9,303 ha (2019), respectively. In contrast, Rohingya camps increased from 78 ha (2017) to 1,968 ha in 2019. Likewise, degraded forest increased from 1,862 ha (2017) to 3,792 ha in 2019. Further, field works demonstrated that areas nearby the camps are completely cleared, and therefore, topsoil is severely exposed (Field Survey, 2018) (Fig. 3a). Moreover, due to high demand of fuelwood, soil is dugged to uncover and pull out the plant’s remnants, especially the root of trees (Fig. 3b).
The gains and losses are analyzed, which demonstrated that agricultural land decreased to about 267 ha in 2017 2018 and 190 ha in 2018 2019. The maximum reduction is observed in two land covers categories, shrubs and mixed forest. During 2017 2019, shrubs reduced to about 1,495 ha and mixed forest experienced a decrease of 1,289 ha. Homestead vegetation, planted young forest and canopy forest covers also reduced (208 ha, 160 ha and 184 ha, respectively) but not in the same magnitude of shrubs and mixed forest land covers. On the other hand, degraded forest had a maximum increase of 1,929 ha followed by Rohingya camps (1,889 ha) during 2017 2019. Because forest covers such as shrubs, mixed forest, plantation forest and canopy forest are vital components of a forest ecosystem, loss of these covers is indicative of the deterioration of the ecosystem. Hence, this finding is crucial to understand loss of individual forest covers as well as specific decline of the respective ecosystem functions.
LULC changes between years is presented in Fig. 4, which shows changes of one land cover to another. It also shows spatial trend map of 2017 2019, suggesting that the impact of Rohingya on forested lands was higher in and around the refugee camps than locations further away. This clearly features the effect of Rohingya on the forest covers in the peninsula.
A ‘from to’ analysis with a GIS function was performed (Table 5), which shows contribution of major LULC categories to camps and degraded vegetation class during 2017 to 2019. Agriculture, homestead vegetation, shrubs, mixed forest, plantation forest and canopy forest land covers contributed to the establishment of refugee camps between 2017 and 2018, when massive influx started for the first time in August 2017, however shrubs, mixed forest, plantation forest and canopy forest covers contribution reduced substantially during 2018 2019. This possibly reflects host country’s measures to protect important forested lands in the later period. On the other hand, shrubs and mixed forest contributed largest to degraded forest cover in 2017 2018 relative to other categories. During 2018 2019, four important forest covers (viz., shrubs, mixed, plantation forest and canopy forest) experienced greatest degradation (Table 5). For instance, loss of shrubs cover was 701 ha (i.e., converted to Rohingya camps and degraded forest covers) in 2017 2018 which increased to a loss of 905 ha during 2018 2019, suggesting amplified pressure of Rohingya refugee on the forest systems of the peninsula.
The analysis of forest degradation as a function of population pressure was conducted using total population of the camps and six land use/covers variables (e.g., canopy forest, mixed forest, shrubs, planted trees, degraded forest and camps) and the result is presented in Table 6. The relationship indicates that an increase of population resulted in a decrease of forested lands and positively related with Rohingya camps and degraded forest covers. In other words, degraded forest cover and camps substantially increases with an increase of human populations in the study area (Table 6). However, p value of the correlation matrix was statistically insignificant, ranging from 0.022 0.361 at the 95% confidence interval.
The accuracy assessment showed that overall accuracy of 2017, 2018 and 2019 LULC maps is 86.85%, 89.12% and 91.45% with corresponding kappa of 0.86%, 0.88% and 0.91%, respectively. This signifies that derived LULC information have an acceptable level of accuracy (Abdullah et al. 2019; Zhang et al. 2013). Hence, these maps are inputted to predict spatiotemporal changes in future LULC by assuming business as usual scenario, i.e., if the Rohingya communities continues to live in current locations.