QuestionsPrevious studies have shown that mire vegetation responds to fine‐scale differences in topography; however, very few studies have examined the relationship between multiple topographic conditions and vegetation distribution patterns by using topographic information with a resolution finer than microtopographic patches of sub‐meter scale in mires. Is it possible to predict vegetation distribution patterns in a mire using topographic information derived from drone imagery? How do topographic conditions affect the emergence of vegetation types?Location3 km2 of sloping mire in eastern Hokkaido, northern Japan.MethodsDigital surface models (DSMs) and orthomosaics were created using an affordable drone (Phantom3 advanced) and photogrammetry. We calculated 23 variables reflecting topographic conditions from DSMs. The locations of four vegetation types (Moliniopsis japonica–Eriophorum vaginatum, ME; Moliniopsis japonica–Carex lasiocarpa, MC; Sphagnum fuscum–Ledum palustre, SL; and Carex lyngbyei–Thelypteris palustris, CT) were recorded by field survey and orthomosaics. Random forest models that predicted vegetation types from topographic variables were created and we evaluated the prediction accuracy, importance of variables and effect of variables on the occurrence probability of the vegetation types.ResultsAll predictive indicators (recall, precision, overall accuracy, kappa) were very high (>0.9). Important variables for modeling were altitude, relative height to the surroundings, slopes at a scale of several 10 m, soil wetness index and terrain ruggedness. Each vegetation type appeared under different topographic conditions; altitude was a key factor in ME and MC, and slope in CT. SL showed patchy distribution predictions on hummocks.ConclusionsVegetation distribution in the mire was strongly related to topographic information based on the DSMs. Analysis of the variable effects indicates that the emergence of the vegetation types was explained by a combination of multiple topographic conditions. This approach has the outstanding advantages of obtaining many survey points quickly and retrieving topographic conditions difficult to measure using on‐ground field surveys.