Modelling spatial association in pattern based land use simulation models.

Area of Science:

• Environmental modeling
• Geographic Information Science
• Spatial analysis

Background:

• Pattern-based land use models are crucial for forecasting land use change, often relying on landscape driving variables.
• Existing models struggle to incorporate neighbor effects due to the categorical nature of land use and endogenous interactions.

Purpose of the Study:

• To develop and evaluate a novel single-variable measure for capturing neighbor interactions in land use patterns.
• To assess the impact of dynamically updating this measure on land use change model forecasts.

Main Methods:

• Developed a single variable measure to quantify neighbor interactions within land use patterns.
• Employed a stepwise updating process to dynamically adjust the neighbor interaction measure.
• Applied the CLUE-S (Conversion of Land Use and its Effects at Small regional extent) system for forecasting in the Deep Creek watershed, British Columbia.

Main Results:

• The proposed measure significantly improves model calibration compared to approaches ignoring neighbor effects.
• Dynamic updating of the neighbor interaction measure enhances the accuracy of land use change forecasts.
• Failure to account for changing spatial influences leads to biased land use change predictions.

Conclusions:

• The developed measure effectively captures crucial neighbor interactions in categorical land use data.
• Dynamic updating of spatial influences is essential for reliable land use change modeling.
• This approach offers a more robust method for forecasting land use change, improving ecological and planning insights.