Assessing and predicting biodiversity in a floodplain ecosystem: Assimilation of net primary production derived from imaging spectrometer data into a dynamic vegetation model

New concepts for river management in northwestern Europe are being developed which aim at both flood protection and nature conservation. As a result, methods are required that assess the effect of management activities on the biodiversity of floodplain ecosystems. In this paper, we show that dynamic vegetation models (DVMs) in combination with regional scale derived remote sensing products can be adopted to assess both current and future ecosystem development and biodiversity status of a complex floodplain ecosystem in the Netherlands. The dynamic vegetation model SMART2-SUMO2 in combination with the nature valuation model NTM3 predicting potential floristic diversity was applied to simulate the biodiversity status of the Millingerwaard floodplain along the river Rhine in the Netherlands. Estimates of net primary production (NPP) derived from airborne HyMap imaging spectrometer data were used for validation of the simulated NPP by the DVM at the time of data acquisition in 2004. Imaging spectrometer derived NPP was in good agreement with the SMART2-SUMO2 modeled results. The NTM3 derived nature valuation in 2004 expressed as plant diversity for the floodplain was high and well in agreement with field observations. In a next step, the DVM was re-initialized using imaging spectrometer derived NPP in 2004 and a forecast of plant diversity and biomass development in 2050 was made. A comparison was performed for three pre-defined floodplain management scenarios using a data-assimilation based approach as well as one without. Significant differences in biomass development can be observed between the scenarios. Predicted plant diversity for individual ecosystems in 2050 shows increased variability for forest ecosystems compared to grass ecosystems. This shows that floodplain management should take advantage of spatiotemporal dynamics of the floodplain as a basis for fostering the development of increased biodiversity. The results of this study demonstrate that imaging spectrometer derived products can be used for validation and initialization of DVMs.