Land use models do work (at least sort of)

By Corey Allan and Suzi Kerr

This post first appeared on Agricultural Emissions Dialogue.

Land-use models are used to explore possible futures, anticipate and diagnose problems, and simulate the effects of different policies. ‘All models are wrong but some are useful’ and more carefully developed and rigorously tested models are more useful. If you are interested in how land-use models are and should be used, our recent paper on land-use modelling provides a non-technical overview of the land-use models currently used in New Zealand. 

If you are interested in the more nitty-gritty detail of modelling, Motu has recently released a working paper that documents and validates the Land Use in Rural New Zealand (LURNZ) model. LURNZ is used to simulate the impacts of climate change policies (such as the inclusion of agriculture in the NZ ETS) on rural land use in New Zealand. The paper tests the projections of the model against reality.  The Land Use in Rural New Zealand model (LURNZ) is based on a heuristic model of dynamic land-use optimisation with conversion costs. It allocates land-use changes to each pixel using a combination of pixel probabilities in a deterministic algorithm and calibration to national-level changes. We simulated land-use change out of sample between 2002 and 2008 and compared the simulated changes to observed land-use change. We show that the allocation algorithm assigns changes in land use to pixels with similar characteristics to those where land-use changes are observed. We also show there is a strong positive relationship between actual territorial-authority-level dairy changes and simulated changes in dairy area. As a result of the model construction, we underestimate the “churn” in land use. You can see the most recent simulations from the model on the LURNZ website.