Application and assessment of a dynamic riparian vegetation model to predict the spatial distribution of vegetation in two Japanese river systems

Abstract

Riparian ecosystems which are highly sensitive for hydrological processes are being altered worldwide, largely because of anthropogenic activities that include the construction of dams, the afforestation of river basins and the mining of river gravel. During last 60years, many of the Japanese rivers have been severely forested, and the gravel habitats that were dominant before the 1950s no longer exist. The restoration of these habitats to their former condition to conserve the diversity associated with undisturbed gravel habitats is a growing concern. Therefore, the influences that alter these riparian habitats must be recognized, and the development of a management tool to predict possible changes and evaluate causes and effects is emerged. In this context the numerical models used to describe the riparian vegetation dynamics is important. Therefore, the objective of this study was to evaluate the application of a recently developed dynamic riparian model (DRIPVEM) under Japanese river conditions to predict the spatial distribution of riparian vegetation. The model simulation was conducted for two Japanese river reaches, and the compatibility of the model was verified with observations on the distribution of vegetation. Based on these visual observations, the simulated results showed reasonable agreement with the observations. Furthermore, the compatibility of the model was evaluated with the help of a GIS tool. The model compatibility was greater than 70% for the simulation of both tree and herb distributions; however, for the bare areas, the compatibility was approximately 50%. The accuracy of the model was greater than 0.85 and based on the Kappa coefficient of agreement, the performance of the model was moderate to substantial for the conditions of the two river systems. Although mimicking complex ecosystem processes is a challenging task for any type of dynamic model; the DRIPVEM has great potential to predict the distribution of riparian vegetation.