Effect of thematic map misclassification on landscape multi-metric assessment

Abstract

Advancements in remote sensing and computational tools have increased our awareness of large-scale environmental problems, thereby creating a need for monitoring, assessment, and management at these scales. Over the last decade, several watershed and regional multi-metric indices have been developed to assist decision-makers with planning actions of these scales. However, these tools use remote-sensing products that are subject to land-cover misclassification, and these errors are rarely incorporated in the assessment results. Here, we examined the sensitivity of a landscape-scale multi-metric index (MMI) to error from thematic land-cover misclassification and the implications of this uncertainty for resource management decisions. Through a case study, we used a simplified floodplain MMI assessment tool, whose metrics were derived from Landsat thematic maps, to initially provide results that were naive to thematic misclassification error. Using a Monte Carlo simulation model, we then incorporated map misclassification error into our MMI, resulting in four important conclusions: (1) each metric had a different sensitivity to error; (2) within each metric, the bias between the error-naive metric scores and simulated scores that incorporate potential error varied in magnitude and direction depending on the underlying land cover at each assessment site; (3) collectively, when the metrics were combined into a multi-metric index, the effects were attenuated; and (4) the index bias indicated that our naive assessment model may overestimate floodplain condition of sites with limited human impacts and, to a lesser extent, either over- or underestimated floodplain condition of sites with mixed land use.

An Explorative Study on Estimating Local Accuracies in Land-Cover Information Using Logistic Regression and Class-Heterogeneity-Stratified Data

“…As mentioned above, unless field-measured data that are sufficiently accurate are used [53], spatial analyses and modelling based on remote-sensing data and land-cover information estimated from them are subject to uncertainty. There is impressive literature on uncertainty in landscape pattern indices (or metrics) and analyses due to misclassification errors in land-cover maps [54][55][56][57][58][59]. As landscape pattern indices were used as explanatory variables for logistic modelling of accuracy in this paper, existing methods in the literature listed above may be usefully explored for analyzing sensitivities of relevant pattern indices to misclassification errors.…”

“…The work by Carroll and Wand [62] and Yi et al [63] may serve as good starting points for further research. On the other hand, simulation-based approaches (e.g., [59]) also merit consideration. We can simulate (land-cover) maps containing misclassification errors.…”

Mapping Mediterranean Wetlands With Remote Sensing: A Good-Looking Map Is Not Always a Good Map

“…Key elements for delineating and mapping wetlands using RS are quality images; rules for detection of significant changes in wetland habitats; a wetland nomenclature; procedures for interpreting habitat classes; and robust validation procedures. Despite precautions, land cover misclassifications always occur (Kleindl et al, 2015), and a known margin of error is usually accepted by specialists. However, known and unknown type errors can easily outweigh the credibility of habitat maps if the mentioned key elements are not carefully developed and implemented.…”

“…This could be done by aggregating habitat types into broader classes (higher levels in hierarchical nomenclatures), when dealing with regional assessments involving many sites and limited ancillary data. Aggregating land cover categories into less numerous classes has been shown to increase thematic map accuracy (Kleindl et al, 2015), thereby reducing the classification errors and increasing time efficiency for multiple sites assessments (e.g. MWO, 2014).…”

“…Despite having diligently applied nomenclatures using the best available RS and ancillary data with interpretation protocols, errors can still occur (Kleindl et al, 2015; see also Box 1). To keep them within acceptable limits, the produced maps should always be checked and validated.…”