®®®® SIIA Público

Título del libro: Environmental Modelling And Software For Supporting A Sustainable Future, Proceedings - 8th International Congress On Environmental Modelling And Software, Iemss 2016
Título del capítulo: Combining Geographically Weighted and pattern-based models to simulate deforestation processes

Autores UNAM:
JEAN FRANCOIS RAYMOND MARIE MAS CAUSSEL;
Autores externos:

Idioma:

Año de publicación:
2016
Palabras clave:

Deforestation; Regression analysis; Sustainable development; Coincidence index; Complex Processes; Geographically weighted regression models; Pattern-based models; Region-based models; Spatial data mining; Spatially explicit; Weights of evidences; Data mining


Resumen:

In this study, two approaches were compared to carry out spatially explicit deforestation model using data driven pattern-based models. In the first approach, model was trained globally: A single matrix of change and set of weights of evidence was obtained taking into account the entire study area. Therefore, the relationship between change potential and expected amount of change and the drivers of change was established for the entire study area and used to simulated deforestation process. In the second approach, sub-regions which present different patterns of deforestation were first identified using a Geographically Weighted Regression model. Then model was trained and deforestation was simulated independently for each one of the region. Performance of both approaches was assessed through the comparison between simulated and true deforestation using a fuzzy coincidence index. The coincidence obtained by the region-based model was sightly superior to the global model. The coupling between spatial data mining techniques as Geographically Weighted Regression models can contribute to help understanding the land changes as complex processes involving both social and natural systems and increasingly develop models which take into account the processes of change and not only the patterns. © Environmental Modelling and Software for Supporting a Sustainable Future, Proceedings - 8th International Congress on Environmental Modelling and Software, iEMSs 2016. All rights reserved.


Entidades citadas de la UNAM: