Bruneis forest carbon dynamics by land cover change: An object-based image analysis using Landsat time-series data

Approximately 70% of Bruneis forests are covered by tropical rainforest, containing high carbon storage in forest land. It is necessary for Brunei to develop system for measurement, reporting and verficifation (MRV) for keeping track of forests. This study attempted to construct historical set of high-resolution wall-to-wall land-cover mapping and develop map for forest carbon dynamic by land-use change Tree Loss & Change Map. We used spectral exponent characteristic for land-use classification: forest, wetland and settlement. We disaggregated forest land by considering predominant forest type in Brunei including Carangas mixed forest, Peat swamp, Mangrove forest, Heath forest. We further constructed an error matrix as per-pixel confidence for determining the accuracy from satellite image on classification using Brunei field survey data and Google Earth. The average overall accuracy of Bruneis land-cover classification was ~75%, which considering as high accuracy. The accuracy assessment was scored 88% (kappa value: 0.80) for 1998, 87% (kappa value: 0.78) for 2007 and 93% (kappa value: 0.87) in 2017. We developed time-series land representation matrix to analyze land-use and land-cover reporting as following: forest land converted to other land-use category, reafforestation area, forest land remaining forest land(or the land with forest-type change). As a result, the forest degradation area was estimated as 31,456ha from 1998 to 2017, while forest land remaing forest land was 441,259ha, and reafforestation area was estimated as 47,761ha. Of total area (576,533ha), these areas are composited of 5.5%, 76.5% and 8.3% respectively