Spatial Simulation Modelling of Future Forest Cover Change Scenarios in Luangprabang Province, Lao PDR

Abstract

Taking Luangprabang province in Lao Peoples's Democratic Republic (PDR) as an example, we simulated future forest cover changes under the business-as-usual (BAU), pessimistic and optimistic scenarios based on the Markov-cellular automata (MCA) model. We computed transition probabilities from satellite-derived forest cover maps (1993 and 2000) using the Markov chains, while the "weights of evidence" technique was used to generate transition potential maps. The initial forest cover map (1993), the transition potential maps and the 1993-2000 transition probabilities were used to calibrate the model. 708projected that current forest areas would decrease, whereas unstocked forest areas would increase in the future. Conversely, the optimistic scenario projected that current forest areas would increase in the future if strict forestry laws enforcing conservation in protected forest areas are implemented. The three simulation scenarios provide a very good case study for simulating future forest cover changes at the subnational level (Luangprabang province). Thus, the future simulated forest cover changes can possibly be used as a guideline to set reference scenarios as well as undertake REDD/REDD+ preparedness activities within the study area.

Prioritizing Areas for Rehabilitation by Monitoring Change in Barangay-Based Vegetation Cover

“…In County Centre in Pennsylvania, the rate of forest loss at county level from 1993 to 2000 was 0.07% per year (140 ha/year), whilst the rate of forest gain was around 0.12% per year (248 ha/year), giving a net conversion rate to forest of 0.05% per year (109 ha/year) [23]. In Lao PDR, the rate of "current forest" loss from 1993 to 2007 was 2.4% per year (437,400 ha/year), whilst the rate of "current forest" gain was 0.7% per year (129,800 ha/year), giving a net conversion rate from "current forest" of 1.7% per year (307,600 ha/year) [61]. Based on the net change, the rate of primary forest loss in Tam Dao National Park in Vietnam from 1993 to 2007 was 2.4% per year (912 ha/year) [20], whilst the rate of forest loss in Kathmandu valley in Nepal from 1978 to 2000 was around 1% per year (137 ha/year) [24].…”

“…The number of land cover classes ranges from three to nine with an average of six. Those that used an unsupervised classification method [27,31,61] have an overall accuracy ranging from 85% to 90% with an average overall accuracy of 88%, and the number of land cover classes ranges from three to eight with an average of six. In this study, the number of land cover classes (4) influenced the accuracy of the land cover maps, rendering relatively high classification accuracy.…”

Deforestation trends and spatial modelling of its drivers in the dry temperate forests of northern Pakistan — A case study of Chitral

“…Wolfram (1983) refers to nearly 50 papers related to applications of cellular automata, indicating the model's accuracy and reliabi lity. Soares-Filho et al (2006) used a similar approach for modelling deforestation trends in the Amazon basin, as Geri et al (2011), Kamusoko et al (2011), Maeda et al (2011) and Farfán (2012 did in their studies of land cover change including deforestation.…”

REDD+: Quick Assessment of Deforestation Risk Based on Available Data

“…Several tools have been developed and used since the 1990s to simulate forest-clearing dynamics and to predict which areas are subject to the risk of losing carbon due to deforestation [14][15][16][17][18][19]. However, limited data availability can hamper their use in some developing countries [20][21][22].…”