Landscape as a model: the importance of geometry

Area of Science:

• Ecological modeling
• Geographic Information Systems (GIS)
• Spatial analysis

Background:

• Spatially explicit models are crucial for predicting uncertain processes like climate change and species invasions.
• Bias in these models can confound results, particularly those supporting decision-making.
• The geometry used to represent virtual landscapes is a potential, often overlooked, source of bias.

Purpose of the Study:

• To compare alternative landscape geometries for spatial models.
• To introduce a mechanism for incorporating landscape representation uncertainty.
• To assess the impact of geometry on movement and landscape representation biases.

Main Methods:

• Comparison of regular (square, hexagonal) and irregular geometries in virtual landscapes.
• Testing cell-to-cell movement across homogeneous landscapes.
• Evaluating landscape representation of real-world scenarios.
• Analyzing bias reduction through irregular geometry subdivision.

Main Results:

• Regular geometries systematically bias movement direction and distance.
• Individual instances of regular geometries create qualitative and quantitative errors in landscape representation.
• Irregular geometries represent complex landscapes without error.
• Subdividing with irregular geometry eliminates bias from regular geometries.

Conclusions:

• Regular landscape geometries introduce significant bias in spatial models.
• Irregular geometries offer accurate landscape representation and unbiased movement simulation.
• Irregular geometry is recommended for all spatial models, especially predictive ones for decision-making.