Using fully polarimetric SAR data for the retrieval of soil surface roughness: potentials and limitations for an operational use
P. Marzahn(1) and R. Ludwig(1)
(1) Ludwig-Maximilian University of Munich, Luisenstrasse 37, 80333 Munich, Germany
Soil surface roughness plays an important role in numerous physical processes as well as in the
remotely sensed derivation of geophysical parameters using microwaves. In the literature, various
empirical as well as polarimetric roughness descriptors exist and are deployed mostly in qualitative
assessments successfully. However the temporal behaviour of those roughness descriptors is not
tested very intensive and little knowledge about this issue is known.
To bridge this scientific gap, the framework of the ESA founded AgriSAR 2006 campaign gives an
excellent basis for testing the robustness of such polarimetric roughness descriptors under different
crops, phenological stages, soils and tillage systems.
During AgriSAR 2006, various bio- and geophysical parameters such as soil moisture, soil surface
roughness as well as vegetation parameters (wet/dry biomass, vegetation heights) have been
sampled over the Demmin test site, located in northern Germany, for a whole phenological cycle
from April to August 2006. Simultaneous to the weekly ground truth measurements, multi
parametric radar imagery has been acquired using DLR E-SAR system in X-, C- and L- Band.
In this paper, the robustness of various well established polarimetric surface roughness estimators
(Anisotropy A, Entropy H, |ρRRLL|, Re|ρRRLL|, S.E.R.D) is evaluated by using intense highly accurate
photogrammetric AgriSAR roughness ground truth data as well as by analytical solving Oh`s model
using the derived roughness values as a priori information for an estimation of soil moisture.
The paper discusses the potentials and limitations of those algorithms, shows dependencies of the
retrieval algorithms to different frequencies and highlights the role of the in field roughness
description for an enhanced roughness retrieval from the applications point of view.