Sensitivity of Modeled C- and L-Band Radar Backscatter to Ground Surface Parameters in Loblolly Pine Forest

Variations and fluctuations of forest floor scattering was soil rms height, whereas, at C-band, all parameters were about equal in influence. The modeled sensitivities properties can lead to variability in radar backscatter from forest, thereby interfering with SAR backscattersuggested that the surface-related uncertainty in forest backscatter at L-HH and L-VV, and at C-HH and C-VV based estimation of forest biophysical characteristics. Understanding how forest backscatter varies over the range at steep incidence angles, may be unacceptably large for many applications. Elsevier Science Inc., 1998 of probable surface conditions is important in the selection of appropriate radar bands and incidence angles for forest sensing applications and for quantifying uncertainty of the derived forest characteristics. Using a canINTRODUCTION opy backscatter model, we examined the sensitivity of Many studies have explored the potential of imaging ramodeled Cand L-band backscatter to five surface padar for estimating forest biophysical characteristics, esperameters in three loblolly pine stands representing differcially biomass (Dobson et al., 1992, 1995; Harrell et al. ent stages of development in Duke Forest (NC). The pa1997; Hussin et al., 1991; Kasischke et al., 1994a; Ranson rameters were litter depth and moisture content, soil rms and Sun, 1997). In reviewing ecological applications of height and correlation length, and soil moisture content. synthetic aperature radar (SAR) backscatter, Kasischke et Of the bands considered, L-HH backscatter had the highal. (1997), though acknowledging that their opinion repest sensitivity to the surface parameters, followed by resents the optimistic side of an ongoing debate, stated L-VV. In the incidence angle range of 208–408, L-HH that radar appears to offer the greatest promise of remotevaried by 5.3–9.6 dB as the surface parameters changed sensing techniques for estimating the forest biomass. Inover their range, whereas L-VV varied by 3.7–4.5 dB. At deed, the ability of microwaves to penetrate through cloud shallower incidence angles, the sensitivities were lower cover and respond to branches and trunks deep in the but probably not negligible. C-HH and C-VV were sensiforest canopy gives SAR a potential advantage over optitive only at steep incidence (h05208–308) for the lowest cal sensors such as the Landsat thematic mapper (TM). biomass stand studied. C-HH and C-VV sensitivity fell However, deep penetration also carries with it a drawoff with increasing incidence angle and stand maturity. back: it ensures that forest backscatter is a complex mixL-HV showed slight sensitivity for the low-biomass stand ture of backscatter from several scattering mechanisms and at h05208–308 but was otherwise insensitive. C-HV that vary in relative importance, depending on forest was insensitive to the surface for all stands and angles. structure, moisture content, and ground surface condiAt L-band, the most influential of the five parameters tions. Whether the many sources of variability in forest backscatter can be brought into tight enough control to permit accurate synoptic biomass assessment is an im* Department of Geography, East Carolina University, Greenville, portant research question. North Carolina † Institute for Computational Earth System Science, University of Recent studies indicate that backscatter response to California, Santa Barbara forest biomass depends on stand-to-stand differences in Address correspondence to Y. Wang, Department of Geography, stand structure (tree height, diameter, and stocking denEast Carolina University, Greenville, NC 27858. Received 28 May 1997; revised 7 July 1998. sity) and on species-related differences in growth form