Geoscience Applications of SAR:

SIR-C/X-SAR is an imaging radar system that flew aboard the NASA Space Shuttle in 1994. It consists of a radar antenna structure and associated radar system hardware that is designed to fit in the Space Shuttle's cargo bay.

SIR-C/X-SAR's unique contributions to Earth observation and monitoring are its capability to measure, from space, the radar signature of the surface at 3 different wavelengths, and to make measurements for different polarizations at 2 of those wavelengths. SIR-C image data is helping scientists understand the physics behind some of the phenomena seen in radar images at just one wavelength/polarization. Investigators on the SIR-C/X-SAR Science Team are using the radar to make measurements of the following:

1. vegetation type, extent and deforestation
2. Soil moisture content
3. Ocean dynamics, wave and surface wind speeds and directions
4. Volcanism and textonic activity
5. Soil erosion and desertification
   
   

The SIR-C/X-SAR antenna structure is show if the picture below, as it looks in the Shuttle Cargo Bay in orbit around the Earth. It actually consists of 3 individual antennas, one operating at L-band (23.5 cm wavelength), one at C-band (5.8 cm wavelength) and the third at X-band (3 cm wavelength).

SIR-C/X-SAR System Parameters

Orbital Altitude :225 Km

Resolution: Typically 30m X 30m on the surface

Look angle range: 17 to 63 degrees from nadir (straight-down)

Bandwidth: 10, 20 and 40 MHz

Pulse Repetition Rate: 1395 to 1736 pulses per second

Total Science Data: 50 hours/channel/mission

Total instrument mass: 11,000 Kg

DC Power Consumption :3000 to 9000 W

Our Test Site is near Raco, Michigan. This is in the eastern part of the Upper Peninsula of Michigan.

This region contains numerous lakes, scattered grasslands, marsh, agricultural areas, and is dominated by forests: northern hardwoods, aspen, upland conifers, and lowland conifers. A ground-measurement campaign measured specied composition and biomass for many test areas over a period of 4 years and thsese test stands are located as shown in the figure. The detailed biomass data obtained was used in devlopment of a biomass estimation algorithm as discussed below.

There was considerable activity on the ground during the 2 SIR-C/X-SAR overflights. A team of 20 or so researchers and graduate students gathered data concerning snow depth, precipitation, soil moisture, moisture inside trees, and other variables that change hourly or daily.

Below is a picture of one of the graduate students getting the dielectric constant of a tree during a full diurnal cycle, and so she is ther near midnight changing batteries on the data recorders.

---

Classification Results

---

Aftern elaborate pre-processing procedure that is alluded to in the following figure, the multi-channel data is then classified and validated against our numerous training areas for accuracy.

At level 2, where we distinguish 5 different classes, 3 of them trees, we can do very well, better than 95% accuracies. When adding in various texture measures, we can further distinguish 6 more classes, as shown in the following "level-3" classification figure. The accuracies are generally very high, and we believe we have a general method for classifying this kind of landscape very accurately.

Changes in yellow.

excluding big Aspens: avg tree accuracy = 96%

Biomass

Given the extensive measurements made on the ground to quantify biomass, we also developed a biomass estimation procedure. It was about as accurate as the ground measurements: 15%. From this we can also estimate the carbon, which is directly related to biomass.

Note, however, that we can only detect living biomass, as the dead biomass is generally not wet enough to be seen by the radar, which is mostly interacting with the moisture in the plants to produce the backscatterd signal.

Clearcuts

Below can be seen an image where the trees are colored as green and the trees that have been cut down since the spring are shown in red. This algorithm is fairly heuristic at the moment, but was quite accurate in identifying the new clearcut areas. We have also quantifyied the carbon loss through the use of the biomass estimation algorithm on the pre-clearcut SAR image.