.c.Raco, Michigan


Titles of Investigations:

I.  Polarimetric Radar Observations of Forest State for Determination
of Ecosystem Process

II.  SIR-C Polarimetric Radar Image Simulation and Interpretation
Based on Random Medium Model

III.  Multifrequency Imaging Radar Polarimetry: Geophysical
Factors from Penetration Phenomena

Principal Investigators:

I.  Dr. Fawwaz Ulaby/Craig Dobson
  University of Michigan

II.  Dr. Jin Kong
  Massachusetts Institute of Technology

III.  Dr. Howard Zebker 
  Jet Propulsion Laboratory

Site Description:

 The Raco supersite is at the eastern end of Michigan¹s Upper
Peninsula.  It is contained within a box defined by 46°40'N
to 45°50'N latitude by 83°50'W to 85°10'W longitude; center
coordinates are 46°15'N latitude and 84°30'W longitude. 
The site is located at the ecotone between the boreal forests
and northern temperate forests, a transitional zone that
is expected to be ecologically sensitive to anticipated global
changes resulting from climatic warming.  Baseline studies
of vegetation communities are essential in monitoring such
changes.
 The site contains most boreal forests species as well as
many of the temperate species, thus studies here serve to
link studies performed well to the north and south.  The
distribution of forest communities is largely determined
by a soil-controlled resource gradient established by glacial
deposits such as moraines and outwash.  Landscape patch sizes
are large; homogeneous forest stands typically exceed 4 ha
in size.  Represented communities include northern hardwoods
(sugar maple, red maple, beech, eastern hemlock, balsam fir),
pines (jack pine, red pine, white pine), conifer bogs (black
spruce, white spruce, northern white cedar, tamarack), hardwood
swamps (red maple, white birch, aspen), and extensive areas
of disturbance-induced pioneer species (trembling and bigtooth
aspens).  Importantly for development of robust radar scattering
models for forests, these species represent two major branching
architectures (excurrent and decurrent tree forms) and leaf
types (broadleaf and needle-leaf).
 The site is not totally forested.  Numerous large areas
of prairie and hayfields provide areas of relatively low
biomass for biomass and soil moisture studies.  These areas
are scattered across the expected SIR-C/X-SAR image swath,
and also provide sites for the deployment of point calibration
targets.  Arrays of trihedral reflectors and active transponders
are deployed for this purpose over a 70 km x 100 km region
and use both grass-covered surfaces and several large airfields.
 The target arrays are being used by US and Canadian teams
for cross-calibration of the JPL AIRSAR, CCRS airborne SAR,
ERS-1, and SIR-C/X-SAR.
 The site has been imaged by the JPL AIRSAR (April and July
1990, June 1991), the CCRS airborne SAR (June 1989, October
1991), Seasat, and ERS-1.  Located at the intersection of
the ERS-1 ascending and descending nodes, the site was often
imaged twice per day (at 12:30 and eleven hours later at
23:30).  This multitemporal coverage provided by ERS-1 during
its commissioning and multidisciplinary phases (with 3-day
and 35-day repeat intervals, respectively) will complement
the shorter, more intensive observation periods provided
by SIR-C/X-SAR; and thus provides an excellent basis for
examination of seasonal change over time scales which the
shuttle-based SAR cannot match.
 Over the last three years, a major investment has been made
to develop the site for SIR-C/X-SAR and ERS-1 forest studies.
 The University of Michigan, in cooperation with Michigan
Technological University, the United States Forest Service,
and the Michigan Department of Natural Resources has established
a series of well characterized forest ³training² stands as
a stratified sample of the various forest communities.  At
present, 30 stands (each 4 ha) have been inventoried using
a 10% sample population for characterizing relatively static
properties such as tree height, diameter, and stem count
(by specie).  By launch date, it is anticipated that approximately
100 forest stands will be inventoried.  Alometric equations,
based upon destructive sampling of the various species, are
used to estimate aboveground biomass on a per tree basis
and then summed within a stand to yield measures of dry biomass
contained within tree trunks, branches, and foliage.  Additional
measurements include: tree architecture, leaf size and total
leaf area, soil properties, and organic litter.  Global Position
System (GPS) coordinates have been obtained for the corners
and sampling grid within each forest stand.  These coordinates
define one overlay in a Geographic Information System (GIS)
for a 60-km x 60-km area which also contains overlays of
digital elevation, surface geology, soils, land-use, land-cover,
transportation networks, surface hydrology, and forest inventories.
 This GIS is linked to a relational data base containing
time varying quantities for each stand such a leaf area,
snow cover, soil and vegetation water status and microwave
dielectric properties, etc.

Objectives:

I.  a) Validate a vector, radiative transfer model (MIMICS)
for estimating radar backscatter from forested terrain. 
Use this model to simulate expected SIR-C/X-SAR sensitivity
to forest biophysical properties such as aboveground standing
biomass and canopy water status and to surface hydrologic
properties.

  b) Develop inversion techniques for estimation of forest
biophysical and hydrologic properties from SIR-C/X-SAR data
using MIMICS simulations and airborne SAR data.

  c) Test and evaluate the inversion techniques using SIR-C/X-SAR
data.

II.  a) Demonstrate the applicability of the random medium
model in simulating SIR-C Supersite imagery.

  b) Analyze and interpret SIR-C imagery for remote sensing
applications.

  c) Investigate seasonal variations and atmospheric effects.

III.  a) Model, experimentally characterize, and verify penetration
phenomena in hyperarid and vegetated regions using the SIR-C
multiparameter radar system and ground-based receivers.

  b) Invert measured radar backscatter as a function of frequency
and polarization in terms of geophysical parameters of the
surface, subsurface and vegetation canopy such as surface
roughness, subsurface geomorphology, or tree height and density.

  c) Display subsurface and within-canopy features in an
image format, thus easing the interpretability of the results.

Field Measurements:

Kong, please provide if applicable.

I.  a) For the thirty forest stands, the following properties
are measured: 
   €  stocking density, height and diameter by specie; 
   €  leaf area index; 
   €  soil type; 
   €  litter depth and 
   €  surface roughness

   Where appropriate, these measurements are updated to reflect
changes resulting from interannual growth, phenologic development
or disturbances.  Meterologic conditions are monitored at
Sault Ste. Marie and Strongs, MI.  Additional moisture-related
quantities are measured in conjunction with SAR operations
and include: (1) snow extent, depth, density, temperature,
and liquid water content; (2) soil moisture, density, and
dielectric properties; (3) moisture content and density of
the litter layer and; (4) moisture and dielectric properties
of vegetation components (bark, sapwood, branches and foliage).

III.  a) Deploy ground receivers during the experiment to
measure field strengths at both vertical and horizontal polarizations.
 These in situ measurements will constrain and confirm our
theoretical models. 

Crew Observations:

1) Crew Journal:  Document cloud types, presence of rain
clouds, and extent of cloud cover, clearcutting activities,
presence of fires and/or smoke,  and snow extent.

2)  Cameras: Hercules and Hasselblad will be used to photograph
the cloud cover, forests, and vegetation at the site.

Coverage Requirements:

   Minimum coverage requirements for the Raco, Michigan site
are three (3) ascending and three (3) descending at a range
of incidence angles.  Both day and night observations are
highly desirable in order to evaluate effects of daily max./min.
moisture and temperature conditions on SAR data.  A launch
window of 0600 to 0900 hours is ideal.

Anticipated Results:

I.  a) Improve understanding of radar scattering by forests
as demonstrated by modifications to and validation of a vector,
radiative transfer model (MIMICS) as a robust approach for
prediction of radar backscatter from a variety of forest
and tree architectures.

  b) Ascertain the dynamic range of radar backscatter from
northern deciduous and coniferous forests in response to
diurnal, daily, and seasonal dynamics of the canopy layer
and the surface.

  c) Define and evaluate techniques using SAR data to retrieve
estimates of cover type, canopy structure, aboveground biomass,
soil moisture, and plant water status for forested conditions.

II.  a) Predicted multi-frequency, multi-incident angle,
fully polarimetric supersite imagery prior to the actual
SIR-C/X-SAR mission.

  b) Interpreted SIR-C/X-SAR imagery with applications to
vegetation classification, crop type, snow depth, seasonal
and diurnal change studies.

  c) Radar image simulation algorithm based on the random
medium model for future radar sensor development.

  d) Improved radar image processing algorithms for terrain
classification.

  e) Multi-layer random medium model for general earth terrain
scattering.

III.  a) An increased understanding of penetration phenomena
in scattering.

  b) Identification of sources for backscatter in hyperarid
subsurface imaging, and quantitative assessment of the relative
contribution of canopy top, volume, and ground surface scattering
components to the return signal from vegetation canopies.

  c) Solutions for descriptive geophysical parameters in
the supersites studied.  These results would be of great
use to any investigators interpreting images acquired over
similar targets by SIR-C/X-SAR or any other radar system.