During the same time period, remote sensing data was collected from airborne platforms. One of the instruments, the Laser Vegetation Imaging Sensor (LVIS) operated by NASA's Goddard Space Flight Center, used a full-waveform lidar for measuring the height and density of the canopy over a 10km swath for each of the northeastern U.S. study sites. This data was complemented by overflights of the UAVSAR instrument, an L-band repeat-pass synthetic aperture radar, which measured the fully polarimetric scattering matrix of the regions over a 16 km swath. Through a careful desing of the UAVSAR flight pattern, single-pass observations are able to be combined together to form interferometric images which are sensitive to the ground topography and the vegetation vertical structure. Data from both of these observing platforms is now being processed and analyzed for better understanding of the remote sensing techniques sensitivity to the forest biophysical characteristics of interest (e.g. height, density and biomass) with the anticipation that such analysis and algorithm development can ultimately be applied to the design and application of an ultimate space-borne application.
In the work described above, the Harvard forest has served as one of the key sites for this algorithm development. During the 2009 period, the University of Massachusetts conducted an inventory of 15 1-ha plots in the Harvard forest and surrounding region, and has been actively working on the analysis and use of the UAVSAR and LVIS remote sensing data for characterizing the forest structure in the way outlined above. In addition to a direct analysis of the remote sensing instruments direct measure of height and reflectance, which can be empirically related to the forest characteristics of interest, our group has been investigating the use of a forest dynamics model to complement the remote sensing observations. Such a model is capable of directly estimating the forest characteristics through simulations of forest growth constrained by water and light resource availability. While the output of such models is a direct simulation of the forest structure, simple physical models of the remote sensing measurement techniques can be applied to the simulated structures, and thereby related to the remote sensing measurements. By combining the simulations with the remote sensing observations, it is then possible determine the best explanation for the remote sensing observation, informed by the forest dynamics model, and thereby include a degree of ecological realism into the estimation problem.
In this talk, we will provide an overview of the 2009 field campaign conducted at the Harvard forest, discuss the remote sensing data types and their application for the estimation of forest structural characteristics. A discussion will also be given regarding ongoing research for combining remote sensing data and the forest dynamics model, and indicate areas of future research.