Forest biometrics from ERS and JERS in Michigan

Operational forest information gathering in the US is moving fast toward the increased use of remote sensing data for a broad variety of applications in the private and public forestry sector. To achieve the best information extraction from remotely sensed data the focus needs to be on combining the strength of active and passive sensor systems. While optical data have demonstrated strength, e.g., in forest species distinction and forest health monitoring questions, the structural composition of forests can be better detected with radar data. The excellent correlation of the radar signals with forest biophysical parameters has been demonstrated through multi-frequency and multi-polarimetric AIRSAR and SIR-C/X-SAR data and lead to significant knowledge about the specifications for the design of future SAR missions for natural resources mapping and monitoring. To test the capabilities of current SAR missions which provide global. coverage, i.e. ERS, JERS and Radarsat, we conducted an intensive study with multi-seasonal ERS and JERS data in the Raco test site in the Upper Peninsula of Michigan. The Raco test site was a SIR-C ecological supersite and intensive ground campaigns were conducted over the course of five years where more than 80 stands were repeatedly sampled. with fixed plots leading to an unprecedented set of ground truth data for research with the SIR-C data. This paper discusses the results of studies on forest parameter retrieval (basal area, height, stem density, biomass) with ERS-1 C_vv and JERS-1 L_hh data for upland pines (jack pine, red pine, white pine). Statistical models (log-log) were generated to measure the correlation between the various forest biophysical parameters and the radar backscatter