Document: RAL-3-89-20

Initialization and validation of global terrestrial ecosystem models using lidar remote sensing.

DUBAYAH, R.* ¹, G.HURTT ² and J.DRAKE ¹

University of Maryland, College Park, MD 20742 USA ¹
University of New Hampshire, Durham, NH 03824 USA ²

Abstract:
Terrestrial ecosystem models are central to efforts to improve our understanding and prediction of vegetation dynamics and associated biophysical and biogeochemical fluxes. However, their potential has been limited by our inability to obtain data to initialize present conditions and to validate model outputs. Without such data, it is difficult to go beyond predictions about "potential vegetation", to predictions of ecosystem dynamics and fluxes on the current landscape and across policy relevant time scales. For example, the current above ground biomass and carbon stocks of the land surface are not accurately known, and the continuation of small, plot-based field studies will do little to help initialize and subsequently validate global model predictions. While optical and radar remote sensing have promised to provide the needed observations, they have not yet met the challenge. The Vegetation Canopy Lidar Mission (VCL), a joint effort by the University of Maryland and NASA, will produce global data sets of canopy height and canopy vertical structure. In this paper we explore the potential of VCL for global ecosystem modeling, and describe our initial efforts to link available lidar data to a new terrestrial biosphere model that predicts height-structured vegetation dynamics and associated fluxes. We first tested model predictions of aboveground biomass for a tropical forest in Costa Rica, initiallized using climate data, with estimates derived from both ground allometry as well as airborne lidar measurements similar to VCL as a function of stand height. Field, lidar and model predictions all agreed within 10%. The biomass ranged from approximately 75 T/ha for average height of 17.8 m, to 165 T/ha for average height of 31.5 m. We next linked the model with the lidar observations by inputing sub-grid cell canopy heights as derived by lidar into the model, and letting the model predict biomass for the forest based on these heights. Biomass predictions for the study area were again within 10% of ground based predictions, confirming the ability of the lidar to provide important land surface parameters for global modeling applications. Our future work is focused on assessing the affects of using lidar heights as a surrogate for land use history for model predictions of net carbon flux.

Keywords: global ecosystem modeling, lidar remote sensing

This abstract is being presented at: 10:30 AM in session:
REMOTE SENSING