Fake it till you make it: What can we learn about simulating forest structure?

Michael Palace, CSRC-University of New Hampshire, michael.palace@unh.edu (Presenter)
Bobby H. Braswell, Applied Geosolutions, LLC, rbraswell@appliedgeosolutions.com
Stephen Hagen, Applied Geosolutions, LLC, steve.hagen@agsemail.com

The dynamic processes of growth and disturbance are reflected in the structural components of forests, such as tree trunk diameter size distributions and canopy profiles, defined as the vertical distribution of foliage. These structural properties of forests are closely linked with ecosystem functioning. Forest structure represents a critical feature that can be both measured with field-based methods or estimated using remotely sensed data. Often, interpretation of remotely sensed image data is limited because of the inability to completely know all forest structural elements that constitute the landscape and image. An approach to over come this limitation is to synthesize forests and generate images for different forest structural parameters. These images could be used to test algorithms, such as crown delineation, textural methods, and spectral unmixing.

This data could also be used to provide input for radiative transfer models and forest simulation models. Reliable scene modeling is vital in data assimilation with remote sensing. Scene modeling has been elusive because for moderate resolution data it is difficult to map biophysical model parameters to the observed satellite data. At high resolution the process might be more straightforward because the scene-level parameters match (or are the same) as the model parameters.

There are numerous forest structural parameters that could be synthesized in such a forest. We have developed a three-dimensional model that can use geometric series, field stand data, or estimated crown distributions to generate a synthetic forest. We are able to generate multiple synthetic vegetation profiles (1000+) based on different geometric series representing different hypothetical forest structure. Our newest advance is how to space trees and tree crowns using a geometric series. We can either use field-based measurements of tree spatial distribution or use simple rules of how trees space themselves across the landscape. In addition, we have begun to incorporate disturbance regimes to develop another spacing mechanism. Our initial attempts have produced synthetic forests to test algorithm develop, provide input to radiative transfer models.

Presentation Type:  Poster

Session:  Other   (Mon 4:00 PM)

Associated Project(s): 

• Palace, Michael: Estimation of tropical forest structure using multiple remote sensing platforms and field based data ...details
• Palace, Michael: Scaling Forest Biometric Properties Derived from High Resolution Imagery to the Amazon Basin using Moderate Resolution Spectral Reflectance Data ...details

Poster Location ID: 62