Thematic resolution influence in spatial analysis. An application to Land Use Cover Change (LUCC) modelling calibration

Abstract

Understanding the uncertainty of every spatial analysis is a required step to be aware of its usefulness and limitations. The chosen scale of analysis determines part of this uncertainty. Thematic resolution refers to the grain component of scale, regarding the thematic detail at which categorical maps are made. The higher the thematic detail or resolution, the larger the amount of information we have and the more complex the analysis will be. Through this paper, we aim to assess how the calibration of a Land Use Cover Change (LUCC) model differs according to variation of the thematic resolution of input Land Use Land Cover (LULC) maps. To achieve this objective, we have set up four modelling exercises in Metronamica, each one calibrated with maps at different thematic resolutions. Obtained simulations were compared in terms of allocation of changes, modelled quantities, explanatory power of the model and pattern of simulated changes. Results show that the model behaves differently depending on the selected thematic resolution. Lower thematic resolution led to simpler simulations that, however, get better validation scores. High thematic resolution maps introduce more complexity and information in the analysis. If not correctly managed, this complexity and information can translate to model noise, worsening the model performance. The paper ends with a proposal of criteria to follow to aid modellers and researchers in the selection of the proper thematic resolution for their analyses.