Global Landcover Prediction and Natural Landcover Vulnerability Mapping for 2050: Regional Level Modeling
Organization: Clark Labs/IDRISI
Position: Remote Sensing Research Assistant
Project: Global Landcover Prediction and Natural Landcover Vulnerability Mapping for 2050
Date: May 2019 - present
Background: Clark Labs/IDRISI received a grant from ESRI to conduct a global land cover change and natural vulnerability model and predict the results to 2050 using TerrSet's land change modeler. As a researcher, I am responsible for regional and country-level modeling and variable exploration using ESRI's Climate Change Index data.
Data: Research assistants utilized 28 independent variables (17 Quantitative & 10 Qualitative) to model transitions from Natural Landcovers to Cropland and to Artificial Surfaces. These variables were chosen based on published studies concerning global and regional-scale models and based on publicly-available global-scale datasets. Then, in order to determine which variables should be prioritized, we utilized the back-propagation multi-layer perceptron (MLP) neural network available in TerrSet’s Land Change Modeler to identify the relationships between landcover transitions and the independent variables.
Position: Remote Sensing Research Assistant
Project: Global Landcover Prediction and Natural Landcover Vulnerability Mapping for 2050
Date: May 2019 - present
Background: Clark Labs/IDRISI received a grant from ESRI to conduct a global land cover change and natural vulnerability model and predict the results to 2050 using TerrSet's land change modeler. As a researcher, I am responsible for regional and country-level modeling and variable exploration using ESRI's Climate Change Index data.
Data: Research assistants utilized 28 independent variables (17 Quantitative & 10 Qualitative) to model transitions from Natural Landcovers to Cropland and to Artificial Surfaces. These variables were chosen based on published studies concerning global and regional-scale models and based on publicly-available global-scale datasets. Then, in order to determine which variables should be prioritized, we utilized the back-propagation multi-layer perceptron (MLP) neural network available in TerrSet’s Land Change Modeler to identify the relationships between landcover transitions and the independent variables.
Methods:
- Evaluated over 10 GIS datasets from government, academic, and private databases based on the data’s geographic coverage and potential relation to natural landscape loss for use in empirical, machine learning-based models
- Explored Climate Change Index data to produce regional and country-level modeling and variable exploration
- Utilized TerrSet’s Land Change Modeler and the MLP neural network algorithm to model natural landscape loss to agricultural expansion and urbanization and predict landcover at 2050 at a continental and individual country scale.
| turkey_land_change_prediction_-_2050.pptx |
Research featured in research articles:
Clark Labs forecasts how land use will change across the globe by 2050
ClarkNow, 8/20/20
Clark Labs forecasts how land use will change across the globe by 2050
ClarkNow, 8/20/20
|
|