Browsing by Subject "Salience-credibility-legitimacy"

Now showing 1 - 1 of 1

Salience, credibility and legitimacy in land use change modeling : model validation as product or process?
(2015) Lusiana, Betha; Cadisch, Georg
Sustainable resource management requires balancing trade-offs between land productivity and environmental integrity while maintaining equality in resource access. Scenario analysis based on a credible simulation model can help to efficiently assess the dynamic and complex interactions in between components and their trade-offs. However, despite the potential of simulation models as decision support tools, acceptance and use by decision makers and natural resource managers are still major challenges, particularly in developing countries. This study was carried out to address issues related to validation of simulation models that includes users’ perspectives on validity of simulation models, scenario-based trade-offs analysis and uncertainty assessment for designing management intervention. Firstly, the current study analyzed users’ perspectives on validity of a simulation model for natural resource management based on two activities. The first activity is based on surveys in four countries (Indonesia, Kenya, Philippines and Vietnam). It explored the perceptions and expectations of potential model users (researchers, lecturers, natural resource managers, policy makers, communicators) on a hypothetical model. The second activity was a participatory model evaluation in Aceh, Indonesia involving use of the spatially explicit FALLOW model and evaluation of its outputs. When assessing a hypothetical model, potential model users’ considered salience (relevance) as the most important attribute in a simulation model followed by credibility. Once a model was used, the ability of the model results to depict reality on the ground (credibility) became a critical and most important aspect for users. Nevertheless, even in cases where model performance was poor, users considered the scenario approach in evaluating their landscape a novelty. Potential model users’ profession, prior exposure to a simulation model and interest in using models did not significantly influence respondents’ ranking of model attributes (salience, credibility, legitimacy). In the second study, to improve salience of a FALLOW model application, a livestock module was developed and tested for a peri-urban situation in the Upper Konto catchment, East Java, Indonesia. This study aimed to explore the impact of land use zoning strategies on farmers’ welfare, fodder availability and landscape carbon stocks. Scenario analysis revealed that the current land zoning policy of establishing protected areas and allowing farmers’ access to fodder extraction in part of the protected areas is the most promising strategy in balancing the trade-offs of production (farmers’ welfare) and environment (represented by above-ground carbon sequestration). Compared to the scenario reflecting current policy, the ‘open-access’ scenario that allows opening land in protected areas, was simulated to increase farmers’ welfare by 13% at the expense of losing 23% of landscape carbon. The extended FALLOW model with its livestock module proved an effective tool to examine the interactions between livestock, cropping systems, household decision and natural resources in data poor environments. The FALLOW model was able to simulate the land cover spatial pattern in the catchment (2002-2005) with a goodness of fit of 81% while the ability of predicting land change was 34.5% at a pixel resolution of 1 ha. In the third study, to understand the effect of uncertainty in input parameters influencing model outcome, an uncertainty analysis of landscape C stock and emissions was carried out using several approaches that can cater for different situations of data availability (plot level carbon stocks and land cover maps). The analysis used data collected during a study assessing opportunities for REDD+ (Reducing Emission from Deforestation and Degradation) in a forest frontier region in Jambi, Indonesia, during 2000-2009. In a minimum data set situation (only single plot carbon estimates and a single land cover map available) the average landscape C stock estimates were 114.5 Mg.ha-1 and 81.0 Mg.ha-1 for 2000 and 2009, respectively. Based on an ‘expected-carbon-deviance’ curve, the confidence levels that the landscape C estimates were correct were 70% and 63% for 2000 and 2009, respectively. For other cases of enhanced data availability, Monte Carlo simulations were carried out to evaluate the propagation of land use classification errors and plot-level carbon stocks variation, jointly influencing landscape C stock and emission estimates. Results showed that excluding errors in land use classification resulted in biased estimates of landscape C stock and emissions. However, the bias over the whole area was estimated to be less than 7.5% (or 2.8 Mg.ha-1) with a coefficient variation of less than 0.2%. In the last study, we combined spatial aggregation analysis on the error-perturbed C emission maps (resulting from Monte Carlo analysis in the third study) with local stakeholders’ perspectives to develop an effective REDD+ scheme at the district level. The uncertainty analysis formed the basis for determining an appropriate scale for monitoring carbon emission estimates as performance measures of a REDD+ scheme. Changes in spatial resolution of C emission maps influenced the magnitude of potential area eligible for carbon payment and the uncertainty in carbon emission estimates. At 100 m resolution, 34.8% of the area would be eligible for REDD+ with an uncertainty of 82% , while at 5000 m resolution only 6.5% of the area would be eligible with a 1% error. At 1 km2 pixel size (1000 m resolution), the errors dropped below 5%, retaining most of the coarser spatial variation in the district. Hence, feasible measures for emission reduction in the district, derived from a participatory planning process, are compatible with the 1000 m spatial resolution of the C emission map. Overall, the research elucidates the importance of involving model users in evaluating a simulation model, including scenario development and subsequent results analysis and interpretation. The study also indicates the importance of making efforts to improve model output accuracy to gain users’ acceptance as users consider spatial accuracy is an important aspect of landscape-based models. In data-scarce situations, model users considered model ‘robustness’ in responding to new situations to be more important than ‘precision’. Scenario analysis proved to be an effective tool to examine interactions in a complex landscape, including their consequences for trade-offs (e.g. farmer’s welfare versus landscape carbon stocks) and synergies (e.g. fodder availability and farmers’ welfare). Analysis of uncertainty of landscape C emission during land use changes can provide guidance in developing appropriate natural resource management interventions. Although model users may perceive model validation as a product, it is in fact a process.