| dc.description.abstract | Concerns aboutthe scarcity of our natural resources and the widespread effects of climatechange are part of the most common debates in the world nowadays. The Land Useand Climate Change interactions (LUCCi) project in the Vu Gia- Thu Bon (VGTB)river basin (Central Vietnam), aims at developing strategies for sustainableland use.
Land use andland cover changes involve complex interconnected processes. The objective ofthis research is to understand the drivers and processes involved indeforestation and forest degradation and to model these processes in order toprovide insights in future deforestation risk areas.
The research isfounded on the complex adaptive systems (CAS) theory. Socio-ecological systemscan be considered CAS, as they involve many variables, are highly dynamic and itsdifferent components adapt or learn as they interact.
The methodologyof this research is characterised by two main phases. In the first phase, landcover data from Landsat TM satellite imagery from 2001, 2005 and 2010 was usedin combination with spatial data on eight potential correlating factors todeforestation. The eight factors considered are elevation, slope, and sixdistance factors (distance to cropland, grassland, small settlements, largesettlements, all roads and paved roads). Statistical tests compared forestchange cells with unchanged forest cells (i.e. the control group). It was foundthat all tested factors showed a significant correlation with deforestation. Themost important factors were found to be distance to cropland and distance tosmall settlements.
In the secondphase of the research, the insights from the previous phase were used as inputfor the model design of an agent-based model, called SoDRA LUCCi. The modelsimulates future deforestation risk areas under a business-as-usual scenarioand the effects of REDD measures on the projected deforestation for 2010-2020.The agents represent rural households, who are considered to be the keydecision-making entities regarding land use and land use change.
The model wascalibrated and parameterised by using the correlation and other statisticalresults from the previous phase. The best-fit calibration method resulted in agood quantitative representation of modelled deforestation when compared to themeasured deforestation of 2001-2010. A qualitative pattern check betweenmodelled and measured deforestation showed that the scattered deforestation inremote areas is underrepresented in the model, causing false negatives in themodel results. Still, modelled deforestation cells were mostly within anacceptable distance from measured deforestation. According to the localsensitivity analysis, the most sensitive parameters are the one related to thedistance to cropland and the parameter that defines the threshold below whichthe location factors of a forest patch are unsuitable for deforestation, calleddeforestation-potential-point.
For the2010-2020 era, the SoDRA LUCCi model predicts a scattered pattern ofdeforestation in the VGTB area, with the highest concentrations in the northwest and centre of the region. With regards to the REDD scenarios, it can be concludedthat measures that implement a quota defining the maximum deforestation perhousehold have the largest impact compared to the modelled business-as-usualscenario. Measures that enforce prohibition or reduction of deforestation inexisting protected areas or in areas with high carbon stocks are expected tohave only limited influence on reducing deforestation. In order to achieve theprojected effects as modelled in SoDRA LUCCi, particular REDD measures mayfocus on (financial) incentives, capacity building and technology transfer forstimulating (alternative) sustainable livelihood activities and strategies.
The firstversion of the SoDRA LUCCi model is still relatively simple. The model can beimproved by distinguishing between agent types. Therefore, it is necessary tohave proper socio-economic data on agent behaviour.
The researchconcludes that agent-based modelling provides a tool for revealing large-scalepatterns that are induced by micro-level actions. Rather than getting lost in aforest of details, it offers an instrument for greater understanding of thebigger picture while acknowledging that those details form the backbone of thesystem. To see the forest for the trees'¦ | |
