Work

Historical and projected land-cover dynamics in the Toba Highland Peatlands, Indonesia: An ANN–CA modeling approach

Year: 2026

Type: article

Abstract: ABSTRACT Tropical mountain peatlands have received far less attention than lowland peat systems, and future land-cover trajectories have not previously been projected for the mapped Toba Highland Peatlands of North Sumatra, Indonesia. Using previously classified Landsat land-cover maps for 2014, 2019 and 2023, this study quantified land-cover transitions and projected land cover for 2028–2038 with an Artificial Neural Network–Cellular Automata (ANN–CA) model in MOLUSCE. The analysis used a nested design: a 65,603-ha landscape comprising Lintong Nihuta, Dolok Sanggul and Pollung districts (Humbang Hasundutan Regency), and the c. 6,180-ha mapped peatland within it. Near-surface peat bulk density was measured at four cultivated sites (eight ring samples) as complementary field information. Cultivated land expanded from 24.8% to 44.0% of the landscape between 2014 and 2019 and remained at 43.2% in 2023, mainly at the expense of natural forest and open land. A 2023 hindcast agreed well with the classified reference map (Correctness 93.32%; Kappa Overall 0.896). Under the modelled conditions, cultivated land is projected to occupy 65.05% of the landscape and 94.75% of the mapped peatland by 2038, up from 70.61% of the peatland in 2023, while natural forest within the peatland declines to 0.64%. Dry bulk density ranged from 0.21 to 0.47 g cm⁻³; these site-level values do not represent peatland-wide conditions. The projections are conditional scenarios rather than forecasts, but they show that land-use pressure is concentrated within the peatland, where peat hydrology and the remaining non-agricultural cover should be priorities for monitoring and management.

Source: Environmental Challenges

DOI: 10.1016/j.envc.2026.101702