EXPERIMENTS WITH MULTI-FREQUENCY AND MULTI-POLARIZATION SAR DATA FOR HYDROLOGICAL 
PARAMETER MODELLING 
Tania Neusch, Manfred Sties 
Institute for Photogrammetry and Remote Sensing 
University of Karlsruhe 
Englerstr. 7, D-76131 Karlsruhe 
email: neusch @ipf.bau-verm.uni-karlsruhe.de 
sties Q ipf.bau-verm.uni-karlsruhe.de 
Commission VII, Working Group 6 
KEY WORDS: SAR Imagery, Airborne Sensor, Soil Moisture, Surface Roughness, Filtering, Modeling. 
ABSTRACT 
The spatial variability of hydrological parameters depends on factors like landscape, topography, soil type and soil surface 
properties. Many solutions of hydrological problems need area wide or spatially distributed input data on the basis of regular 
or irregular grids. One method to provide such data is based on geostatistical extrapolations of point measurements. We 
advocate remote sensing methods to provide such data. Microwaves have the capability to penetrate into the ground and 
are sensitive to hydrological phenomena; thus, active SAR sensors are suitable for this task. 
On April, 17 and 21 of 1997, the Experimental Airborne Synthetic Aperture Radar (ESAR) of DLR Oberpfaffenhofen has 
acquired multi-polarization and multi-frequency radar image data in C-, L-, and P-band over a little watershed in the South 
West of Germany (Weiherbachtal). At the same time, field measurements regarding surface roughness, volumetric and 
gravimetric soil moisture, row directions and land cover type and densities have been made on several representative 
agricultural fields, distributed in the test area of 3 km x 4 km size. 
After slant-to-ground range transformation and georeferencing, the radar image data have been processed to reduce the 
speckle effects, with regards to the ground truth data available. The backscatter values have then been input to empirical 
inversion models for the retrieval of soil water content and surface roughness. The results produced up to now show a 
rough compliance with the field observations. Future work will include model refinement regarding the relief of the test area 
and a more detailed analysis of the wealth of SAR image data. 
In this paper, the image processing steps will be explained, the inversion models and their properties will be discussed and 
  
preliminary results will be shown. 
INTRODUCTION 
The soil water budget is one of the most important elements 
of the natural annual water cycle. An important aspect of 
this budget refers to the soil moisture variation, which is 
hard to describe due to temporal and spatial variability of 
interrelated processes such as evaporation, transpiration, 
infiltration and underground water restoration. However, 
the quantitative description of (at least some of) these pro- 
cesses including their variabilities is required for hydrolog- 
ical, climatological and geological problem solving, which 
has - for a long time - been restricted to point related data 
from field measurements. 
Today, there are efforts to and successes in generating 
area wide input data to hydrological problem solutions prof- 
iting from remote sensing techniques, especially from mi- 
crowave remote sensing. Active microwave sensors not 
only offer weather independent soil surface imaging capa- 
bilities; they also provide subsurface information where the 
surface penetration extent depends on both, sensor and 
surface properties. 
It has been shown in extense (Ulaby et al., 1981, 1982 and 
1986), that microwaves are sensitive to hydrological pa- 
rameters such as vegetation coverage, relief and surface 
roughness. We work on the problem of separation of these 
interrelated parameters from series of radar images. 
Intemational Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998 
1 TEST AREA AND EXPERIMENTS 
In contrast to airborne sensors, spaceborne sensors oper- 
ate from extremely steady and fixed orbits and time cycles 
and imagine the earth - up until today - in one frequency 
band, at one polarization at - usually - at moderate geo- 
metrical resolution, only. Hence, we decided to hire an air- 
borne SAR sensor for imaging of our test area in order to 
get multi-polarization, multi-frequency radar imagery at ac- 
quisition dates of our choice. 
A 3km x 4km watershed of the Weiherbach (Southwest 
Germany) was selected as test area, the reason being, 
that since 1989 in this area, many field measurements and 
hydrological experiments have been performed in the con- 
text of multidisciplinary research projects. Image acquisi- 
tion was realized on April, 17 and 21 of 1997 with the air- 
borne sensor system E-SAR (Experimental Synthetic Aper- 
ture Radar) of DLR Oberpfaffenhofen (German Center for 
Air- and Space-Borne Operations). This sensor system op- 
erates in the X-, C-, L- and P-band, where C-band takes 
only co-polarized and L- and P-band take fully polarized 
radar data. 
Two imaging campaigns with a mere 4 day time delay were 
launched to ensure, that variations of the soil moisture were 
caused by possible precipitation only, excluding influences 
of variable vegetation cover or variable soil surface rough- 
ness. To enforce at least one precipitation event, a small 
549