Estimating Soil Moisture in a Boreal Old Jack Pine Forest

Polarimetric Land P-band AIRSAR data, corresponding model simulations, and classification algorithms have shown that in a boreal old jack pine (OJP) stand, the principal scattering mechanism responsible for radar backscatter is the double-bounce mechanism between the tree trunks and the ground [1]. The data to be used here were taken during six flights from April to September 1994” as part of the BOREAS project. The dielectric constants, or equivalently moisture contents, of the trunks and soil, can change rapidly during this period. To estimate these dynamic unknowns, parametric models of observed radar backscatter for the doublebounce mechanism are developed by using a series of simulations of a numerical forest scattering model. The resulting simulated data are used to derive polynomial fits of backscattering cross section as a function of the ground and trunk dielectric constants. Empirical and field data are used to relate the real and imaginary parts of the dielectric constants, and hence formulate the parametric model in terms of two unknowns only. Three data channels, P-HH, P-VV, and L-HH are used to solve for the two unknowns. A nonlinear optimization procedure is used to estimate the dielectric constants, and hence, in particular, soil moisture. Point ground measurements are used to verify the results of the estimation algorithm. INTRODUCTION AND RATIONALE The boreal ecosystem atmosphere research (BOREAS) project is a multidisciplinary effort to study the interactions between the boreal forest biome and the atmosphere to determine their role in global change. BOREAS is f~ cused on two principle study areaa in central Canada, one near the Prince Albert National Park in Saskatchewan, or the Southern study area (SSA), and the other near Thompson, Manitoba, or the Northern study area (NSA). During the time period of April 1994 to September 1994, several intensive and focused campaigns were carried out, in which several remote sensing instruments made measurements. In particular, the NASA/JPL airborne synthetic aperture radar (AIRSAR) collected polarimetric C-, L-, and P-band data during several flights in this time period. In this work, we concentrate on the data acquired over an old jack pine (OJP) stand in the SSA. The SSA contains several conifer forest types such as young and old jack pines and black spruce. There are also deciduous species such as aspen, as well aa stands consist ing of mixtures of these species. A major goal of performing remote measurements, such as radar measurements using the AIRSAR, was to attempt to retrieve forest parameters that play significant roles in the functioning of the ecosystem. An important such parameter is soil moisture under the forest canopy. Soil moisture content has implications in the rate of transpiration, on water stress, and in the growth rate of forests. The latter is especially important in studying the factors limiting the growth of older trees. In this work, an estimation algorithm was developed to obtain the soil moisture of an old jack pine stand for six different dates in the Spring and Summer of 1994. The site was chosen due to its specific structure, which as described below, enabled a simplified modeling of the scattering process. As shown in [1] and further substantiated using the classification algorithm of [2], at Pand Lbands (P-HH, P-VV, and L-HH), the mechanism almost entirely responsible for the backscattered signal over the OJP site is double-bounce scattering between trunks and ground. For L-VV, the branch-ground interactions also become significant. Therefore, a parametric scattering model was derived in terms of the dielectric constant of the trunks and the dielectric constant of soil. This model was then used in a nonlinear estimation algorithm to estimate the two dielectric values from P-HH, P-VV, and L-HH AIRSAR data. A few point comparisons of the results with ground-truth measurements were performed and good agreement observed. With the availability of further ground-truth data, more thorough validation will be performed.