Extending Earth Observation to a Tropical Mountain Forest Ecosystem using Functional Indicators
Brenner, Silva; Bendix, Jörg
University of Marburg, GERMANY

The biodiversity hotspot of the tropical Andes and its ecosystem services are severely threatened due climate change and unsustainable land use. Hitherto scientific investigations in those areas are restricted, which also hinder the application of current and future earth monitoring satellites with worldwide coverage. In the eastern Andes of South Ecuador, a long-term ecological research project - the Research Unit-816 funded by the German Research Foundation - has provided fundamental data and experience for understanding the functioning of this megadiverse ecosystem. The present work aims to extend this knowledge to the investigation of local impacts of climate change in the future using remote sensing and vegetation modelling. An indicator that can reveal the response of the tropical mountain vegetation to the current and future local climate conditions is the ratio between net primary productivity and evapotranspiration, known as water use efficiency. As a first step of the present investigation, the potential primary productivity and evapotranspiration of plant functional types, adapted to local species, have been simulated using a dynamic vegetation model, which runs with realistic forcing data. The results show the range of water use efficiency that can be expected in the study site for two woody species and two grassland species, which characterizes the study site. We show how these local plant-functional-types respond to climate predictions and stress issues on climate regulating mechanisms of the local forests and grasslands. Since the model can be forced using satellite data, we also show how a species-specific approach can be transferred to an area-wide application using remote sensing. This approach requires the on-going investigation in the variability of proxies for water use efficiency using synoptic data. Therefore, two proxies - one for leaf water content and another for leaf area index - have been estimated using broadband spectral indices (respectively the normalized vegetation index and the normalized water index) from operational high- and medium-resolution satellite data. Ancillary state-of-the-art airborne data (light detection and ranging and hyperspectral scanning) and ground observations have been used to investigate the variability of these proxies at landscape- and tree -crown scales, which are for instance due to land cover gradients in biomass and vegetation structure. Results show how the spatial variability of satellite derived indicators can influence on the calculation of water use efficiency using the modelling approach. This novel and descriptive investigation can serve as basis to extend the application of earth observation programs to those areas, as well as for future monitoring of the response of the tropical mountain forest to climate change and land use change in the future.