Morteza Chalak

Economics of controlling invasive species: A stochastic optimization model for a spatial-dynamic process

&NA; We analyze the dynamic process of invasive‐species control in a spatially explicit and stochastic setting. An integer optimization model is applied to identify optimal strategies to deal with invasive species at a steady state. Optimal strategies depend on the spatial location of invasion as well as on stochastic characteristics of spread and control. Previous studies of invasive‐species control have been stochastic or spatial, but not both. We model a landscape as consisting of multiple cells, each of which may be subject to border control or eradication within the cell. Optimal strategies from the model are characterized as eradication, containment, or abandonment of control. Representing the rate of species spread as stochastic rather than deterministic results in less‐intensive control becoming optimal at equilibrium. The optimal strategy may switch from eradication to containment or from containment to abandonment. If an infestation occurs at the boundary of the region within which it may spread, it is more likely to be optimal to eradicate or contain the species, compared to an infestation in the interior of the region. If the effectiveness of border control is stochastic, then containment is not feasible in the long term, but it is still optimal as a temporary measure in some scenarios.

Restoring Native Vegetation in an Agricultural Landscape: Spatial Optimization for Woodland Birds

In heavily cleared agricultural landscapes, decline of biodiversity could be prevented by restoring native habitat. In this paper, we develop a spatially explicit bioeconomic model that optimizes ecological restoration of habitat for woodland-dependent birds in the Australian state of Victoria. Spatial optimization identifies strategies that would generate substantially greater environmental benefits than are likely to be achieved in current programs. Greater biodiversity outcomes can be expected where restoration is optimized across multiple species rather than just individual species, and if the program does not require an even distribution of restoration effort among farmers. (JEL Q57, R14)

Comparing supervised algorithms in Land Use and Land Cover classification of a Landsat time-series

Machine learning algorithms (MLAs) are often applied to identify Land Use and Land Cover (LULC) changes, but typically to only a limited set of imagery. This leaves the consistency of MLAs performance through time poorly understood. The research objective was therefore to compare the performance of six MLAs across a time-series of Landsat imagery (1979, 1992, 2003, 2014), all processed in the same manner. Here Support Vector Machines (SVM), K-Nearest Neighbours (KNN), Random Forests (RF), Learning Vector Quantization (LVQ), Recursive Partitioning, Regression Trees (RPART) and Stochastic Gradient Boosting (GBM) were evaluated. The results demonstrated that SVM achieved higher overall accuracies and kappa coefficients, and a slightly improved fit at individual class level, than the second best classifier RF. Both classifiers clearly outperformed the other algorithms. These results suggest that SVMs (or RFs) should be prioritised when classifying time-series imagery for LULC change detection.