A comparison of three ways to assemble wall-to-wall maps from distribution models of vegetation types

Distribution modeling methods are used to provide occurrence probability surfaces for modeled targets. While most often species, distribution can also be applied vegetation types. However, provided by need transformed into classified wall-to-wall maps of types useful practical purposes, such as nature management and environmental planning. The paper compares the performance three assembling predictions multiple types, individually, a map. authors grid-cell based from models 31 test assembly methods. first, probability-based method, selected each grid cell type with highest predicted in that cell. second, performance-based assigned ordered high low model performance, fraction cells given type’s prevalence study area. third, prevalence-based differed method assigning order prevalence. Thus worked two principally different ways: cell-by-cell manner, both them type-by-type manner. All were evaluated use reference data collected field, more or less independently parameterize vegetation-type models. Quantity, allocation, total disagreement, well proportional dissimilarity metrics, evaluation Overlay analysis showed 38.1% agreement between all had lowest disagreement with, from, datasets, but differences small. strongly respect on its quantity allocation components: assignment favored disagreement. best reproduced general pattern variation across area, at cost many rare which left out assembled By contrast, represented accordance nationwide area statistics. results show coverage individual quality acceptable indicative tested currently have shortcomings. indicate specific points methodology map may improved.