Intern Report: Wolfgang Gebauer
This report gives a short view of my
Winter Internship
January 12th to April 19th
at the
Alaska Satellite Facility (ASF)
Geophysical Institute University of Alaska Fairbanks Fairbanks, AK 99775-7320 USA
I want to thank all the people working at ASF, who supported me during my internship. To Parker Martyn, who supported me during the one week survey for RADARSAT ScanSAR modes. And of course a big thanks to my supervisor Jason Williams, who gave me the opportunity to come to Alaska, which is very fascinating during even in the winter time.
Table of Contents:
1. Alaska SAR Facility (ASF)
2.Synthetic Aperture Radar (SAR)
3. SAR Satellites
3.1 SAR Satellite RADARSAT
4. Global Positioning System (GPS)
5. Calibration
5.1 Calibration arrays in Delta and Toolik
5.2 Corner Reflectors
5.3 Maintenance of Corner Reflectors
5.3.1 Battery Discharger for Rechargeable Batteries
6. Radar Cross Section (RCS)
7. Measurement Validation Tests
7.1. Errors in Measurements
7.2. Repeatability and Accuracy
7.3. Impact on Calculation
7.4 Inclination Measured with a Digital Inclinometer
8. ScanSAR Survey
8.1. Reason for the RADARSAT ScanSAR Survey
8.2. Geometric Calibration Targets
8.3. Analyzing ERS1 Lo Res Images
8.4. Site Locations in ERS1 Images
8.5. GPS Settings
8.6. Survey Sites
8.7. Survey Planning
8.8 List of Abbreviations
9. Conclusion of my Winter Internship
10. Useful Literature
1. Alaska Satellite Facility (ASF)
ASF is a sub-branch of the Geophysical Institute at the University of Alaska Fairbanks. ASF calibrates and processes SAR images, that are used by scientists all over the world to survey the earth surface. Those images are recorded by observing satellites ERS1, ERS2, JERS1 and RADARSAT. These satellites transmit SAR data to a 10m dish on the top of the Geophysical Institute building. A second dish with an 11.5m antenna was installed in 1995 200m to the west of that building to track two satellites simultaneously. The received raw SAR data is first recorded on high density tapes and then processed by the specially developed SAR processor. A new IBM PP SP2 computer is in its installation phase and should be ready to process its first SAR image this summer. After processing and calibrating, the images are available for further observations. Scientists use these images to observe forest fires, sea ice, glacier movements, volcanoes etc.
[return to top]2. Synthetic Aperture Radar (SAR)
The Synthetic Aperture Radar is mainly used for remote sensing of the earth by aircraft and satellites. It allows a small resolution in range and along-track (azimuth). The radar waves scanning the earth surface have the advantage to be less influenced by atmospheric changes. This enables an observation of particular areas of the earth, e.g. rain forests and polar ices. The observation of those areas is nearly impossible in a visible or infrared spectrum because of fog and clouds covering humid areas most of the time.
To achieve a high resolution in high altitudes, a large antenna would be required to generate a narrow beam. This of course is impossible to carry with a spacecraft into an orbit. To increase the along-track resolution with antennas of only 10 meters, the SAR principle was developed. It observes each point for a longer period of time, which requires an additional processor system, simulating a long antenna. The range resolution depends only on the time length of each impulse. A shorter impulse increases the resolution in range.
An interesting point is that the best possible resolution of SAR always equals half of the antenna aperture and is independent of all other parameters like wavelength or distance to the target.
SAR images are processed by using the information of amplitude and phase of the received signals. There can be a large number of parameters influenced during this process. If they aren't set correctly, the image could be useless for further observations. Here are some examples of important parameters, affecting SAR image processing:
-Location, altitude and velocity of the SAR sensor
-Movement of the earth
-Profile of the ground (mountains)
-Refraction in the atmosphere
-Variable attenuation in the atmosphere
-Perspective distortions
Some of these parameters can be calculated, but for others there are not any calculation models available. Parameters that do not have models available can only be found by detailed field measurements.
[return to top]3.SAR Satellites
All received and processed images are send by the observing satellites, ERS1, ERS2, JERS1, and RADARSAT. The Canadian satellite RADARSAT, which was launched in September 1995 began to transmit its first data to ASF in March. The following table compares these satellites.
Comparison of SAR satellites, transmitting data to ASF:
| ERS1 / ERS2 | JERS1 | RADARSAT | |
|---|---|---|---|
| Altitude | 785 km | 568 km | 793-821 km |
| Inclination | 98.516[[ring]] | 97.662[[ring]] | 98.594[[ring]] |
| Repeat Cycle | 3, (35, 176) days | 44 days | 24 days |
| Ground Track Velocity | 6.628 km/s | 6.883 km/s | 6.576 km/s |
| Frequency | C-Band, 5.3 GHz | L-Band, 1.275 GHz | C-Band, 5.3 GHz |
| Wavelength | 5.66 cm | 23.5 cm | 5.66 cm |
| Polarization | VV | HH | HH |
| Average Power | 300 W | 71 W | 300 W |
| Swath Width | 100 km | 75 km | see table below |
| Max. Resolution | 12.5 x 12.5 m | 7 x 7 m | 10 x 10 m |
| Pule Compression Ratio | 580:1 | 525:1 | 491, 727, 1260:1 |
| On board Storage | None | Tape 20 min. |
Table 3.1
[return to top]
3.1 SAR Satellite RADARSAT
RADARSAT differs from all satellites launched earlier because of its various beam modes to observe the earth surface. All modes and their location over the swath are shown in the figure 3.1 and table 3.2
Table 3.1
| Beam Mode | Beam Position | Incidence Angle Range (degree) | Resolution (m) | Nominal Area (km) |
| Fine | F1 F2 F3 F4 F5 | 37 - 40 39 - 42 41 - 44 43 - 46 45 - 48 | 10 | 50 x 50 |
| Standard | S1 S2 S3 S4 S5 S6 S7 | 20 - 27 24 - 31 30 - 37 34 - 40 36 - 42 41 - 46 45 - 49 | 30 | 100 x 100 |
| Wide | W1 W2 W3 | 20 - 31 31 - 39 39 - 45 | 30 | 165 x 165 150 |
| ScanSAR Narrow | SN1 SN2 | 20 -40 31 -46 | 50 | x 150 130 x 130 |
| ScanSar Wide | SW1 | 20 - 50 | 100 | 300 x 300 |
| Extended High Incidence | H1 H2 H3 H4 H5 H6 | 49 - 52 50 - 53 52 - 55 54 - 57 56 - 68 57 - 59 | 25 | 75 x 75 |
| Extended Low Incidence | L1 | 10 -23 | 35 | 170 x 170 |
Table 3.2
With all these different types of SAR modes various possibilities in scanning the earth's surface are offered and can be used for studies in:
-Agriculture
-Cartography
-Coastal Zone
-Sea Ice
-Forestry
To study these areas SAR data can be ordered as film/print or in digital form.
[return to top]4. Global Positioning System (GPS)
The GPS is a radio navigation system, operated by the US Department of Defense. It is used to get detailed geographic position information all over the world. The GPS constellation consists of 24 NAVSTAR satellites in 12-hour orbits. A receiver located on earth can receive information from several satellites at the same time, depending on how many satellites are visible and how many can be tracked simultaneously by the receiver. The position is then calculated using distances between each satellite and the receiver. The distances are calculated by using the time delay of each received signal.
Solar winds, gravity, Selective Availability (SA) and type of receiver being used, together can affect the accuracy of GPS measurements. Errors can be removed using differential surveying techniques, which require at least two GPS receivers. One receiver is set up at a particular point where the geographic position is well known. This unit is often designated as a base station. The second GPS unit, called the rover, collects all field measurements at the same time as the base station. To calculate differential corrections, both GPS receivers must collect position information simultaneously. After collecting all the data and sending base station and rover files to a PC, the differential correction program can be started. The differential correction program uses the base station coordinates to erase all random failures and deviations in the rover data file during the time data was collected. With this differential correction mode an accuracy of centimeters is possible. However it must be stressed that base station coordinates must be well known and the rover has a limited range in which it should travel. Also such surveys are often very time consuming and expensive.
Due to the lack of known positions, time constraint and limited budget, the ScanSAR survey (see 8. ScanSAR Survey) only used one standalone GPS receiver to collect data. If Selective Availability was turned on (a worst case scenario), the Pathfinder GPS receiver used in this survey could miscalculate positions by up to 100 meters. For most of the sites in this survey, calculations for a single point ranged between +/-50 meters, in some cases even +/-25 meters. Compared to the 100m wide and the 50m narrow ScanSAR modes, these measurements are within an acceptable range for position measurements used in geometric calibration of ScanSAR data.
[return to top]5. Calibration
By using measurements from corner reflectors, the received raw data images can be corrected within the SAR processor. This is necessary to ensure that all images are accurate to within pre-established tolerances. Corner reflector allow two types of calibration for SAR images.
-geometric calibration is used to see that the location in a SAR image matches with the real geographic location. There could also be a distortion in the image, that could only be corrected by several different well known global position targets, that are visible in a SAR image (e.g. corner reflectors).
-radiometric calibration is to ensure a correct image processing with respect to the returned power of a well known target. The knowledge of the back scatter characteristics of a certain target, in this case a corner reflector, allows one to correct SAR image processing.
To get all essential parameters and settings of corner reflectors, some fieldwork is necessary.
[return to top]5.1 Calibration arrays in Delta and Toolik
The ASF Calibration department has two arrays of corner reflectors to calibrate SAR images. All reflectors used are triangular trihedrals, with each axes measuring 2.44 meters.
The first array is installed near Toolik on the north side of the Brooks Range. This calibration array is maintained only once a year. The climate and the long distance to Fairbanks does not allow maintenance year around. Because of the harsh climate radiometric calibration can only be practiced for a short period of time after each maintenance trip. Geometric calibration can be done as long as the reflectors are visible in a SAR image.
The reflectors of the second array are installed around Delta Junction and are maintained frequently, at least once per month. During my work at ASF we changed this array and expanded it in north direction. All generation one (GENI) reflectors and some GENII reflectors were dismantled and new GENIII reflectors were deployed during my internship at ASF. In the modified array are now eight GENIII reflectors and only one GENII reflector for geometric and radiometric calibration. The SAR image on the next page shows the expanded and modified Delta Junction array.
First ascending SAR image of the expanded Delta array with all corner reflectors.
GIF(173 KB)
or
JPG(95 KB)
figure 5.1
[return to top]5.2 Corner Reflectors
During my internship, a third generation (GENIII) of corner reflectors were installed in the field. These are mainly built from GENI and GENII reflectors. The reflector itself is a slightly modified GENI reflector and the base comes from a GENII reflector. It was necessary to built a new generation because of two main reasons:
- GENII reflectors are useless for radiometric calibration because of bent panels.
- For the Canadian satellite RADARSAT, rotateable reflectors are necessary (see SAR satellites) to point them fast and easily for each satellite pass and mode.
The following three paragraphs will show the development of corner reflectors used at ASF:
1. GENI reflectors are stationary reflectors. For each ascending and descending one reflector was necessary. This is sufficient for ERS1, ERS2 and JERS1, that have only one particular mode. ERS1 and ERS2 fly in the same orbit. The new satellite, RADARSAT, with 7 different beam modes will change the demand on the calibration array (see table 3.2). Between these modes the inclination alone can change from 10 to 60 degrees. RADARSAT additionally uses a different orbit than ERS1, ERS2 and JERS1.
The following sketch shows a GENI reflector, that is still used at the Toolik calibration array.
figure 5.2
2. Generation two (GENII) reflectors (see figure 5.3) were newly designed compared to GENI reflectors. The reflectors were set on bases so that they could be easily moved and pointed toward each satellite. The reflectors were made out of solid aluminum plates. Because of a manufacturing error, a 3 - 5 cm curve in the pannles werre introduced. This had a negative affect on radiometric calibration and was the main reason to design new reflectors. However, one reflector (DJR7) of this type is still installed for radiometric calibration in Delta Junction. The three GENII reflectors removed from the array during my winter internship will be reinstalled for geometric calibration purposes this summer.
Generation two corner reflector design.
GIF(293 KB)
or
JPG(65 KB)
figure 5.3
3. GENIII reflectors are mainly modified GENI reflectors with GENII bases. It was also necessary to weld three aluminum bars on each GENI reflector to decrease its weakness in the windy surrounding of Delta Junction. Because of a different fastening to the bases, new mountings had be designed and installed on all GENI reflectors being used for the GENII upgrade.
The most difficult part of this project were their installation in the wintertime. The weather presented us with a lot of snow in the spring, which is unusual at this time of the year. So snowmobiles were needed to reach 4 of our sites in order to install the new reflectors. It was also impossible to anchor the bases into the frozen ground. Because the ground was frozen we put 180 kg of sand, packed into 6 sandbags on the base. This should make the reflector robust against wind and storm until summer.
Picture of a new GENIII reflector
GIF(172 KB)
or
JPG(43 KB)
figure 5.4
[return to top]5.3 Maintenance of Corner Reflectors
Additional to our deployment tours it was also necessary to maintain the reflectors once a month, for normal calibartion activities. This was one part of my work at ASF. The maintenance consists of the following activities:
-check and change reflector orientation
-adjust this orientation
-determine the reflector location with GPS (only once per new installation)
-tighten up the loosen cables
-clear snow out of the reflectors
-if necessary replace damaged reflector panels (e.g. damaged by firearms, animals)
With the new reflector style, it is possible that there will be even more tours to the Delta Junction array to maintain the reflectors.
[return to top]5.3.1. Battery Discharger for Rechargeable Batteries
For the fieldwork in Delta Junction various rechargeable batteries are used. These batteries were never completely discharged and only used a few times during a year. A battery that is only used a few times during a year and never completely discharged will keep a memory effect, which decreases the battery power. To avoid memory effects and to easily discharge batteries, I built this circuit:
The next picture shows a layout of the battery discharger
figure 5.5 Circuit explanation:
To avoid a low discharging, I used a relay that shuts all power components off if the battery power decreases under a chosen voltage level. In which case the battery always keeps a safe voltage. The circuit is set up to turn it self off if the battery power falls short of 11.5 V. This circuit was mainly developed for 12 V batteries but with a slight change of the potentiometers R4 and R5, it can also be used for 9 V batteries. It runs from 14 to 11.5V (10 to 8.5V for 9 V batteries) depending how high the battery power is at the time the process was started.
To start the process a short press on the push button(PB1) is necessary and the battery will be discharged over the three resistors R1 to R3. All three resistors together will produce 2 W of heat. If the discharge is in process, the green LED lights up. When the battery is completely discharged the LED will turn off. If plus and minus are exchanged, the red LED will light up as long as the push button is pressed down. A complete discherge of a 12 V electric drill battery takes ~5 hours.
[return to top]6. Radar Cross Section (RCS)
The Radar Cross Section (RCS) in general describes the intensity of radar waves reflected by a fictitious area back to the radar. For SAR image calibration the RCS is calculated out of two measured values on corner reflectors. These two values are the inclination and satellite ground track angle.
[return to top]7. Measurement Validation Tests
For all measurements on corner reflectors, a compass is used with a readable scale down to 1[[ring]]. It is also possible to get a more precise reading about the azimuth angle of each corner reflector by interpolation between two degree lines. This interpretation depends on each persons visible observations. The decimal value depends mainly on the person taking the measurements.
The inclination is the second value measured at each reflector. It was carried out with an inclinometer that was built-in to the compass, but is now preformed using a more accurate digital inclinometer.
The accuracy of the compass and its built-in inclinometer is described in the following sections:
[return to top]7.1. Errors in Measurements
The compass attracts magnetic and metal objects very well, as shown in the following table:
| observed objects | without affect | largest affect |
| (range in cm) | in ([[ring]]) | |
| flashlight | > 20 | 45[[ring]] |
| pencil | > 10 | 15[[ring]] |
| clipboard | > 20 | 30[[ring]] |
| camera | > 30 | 20[[ring]] |
| belt buckle (metal | > 25 | 15[[ring]] |
| objects on clothing) | ||
| automobile | >1500 | 10[[ring]] |
table 7.1
As seen from this table, metal and magnetic objects such as cameras, flashlights, belt buckles, etc. should be kept at least 30 cm away from the compass during the measurement. Also the automobile normally used to drive to the reflectors should be parked more than 15 meters away from the current measurement site. All these things affects the measurements up to 45[[ring]].
A second affect on measurements is the deviation between magnetic north and true north. I wanted to measure a line between two well known benchmark points in Delta Junction, but too much snow this spring made this confirmation impracticable. The deviation between magnetic and true north in Delta Junction is set up on the compass at 29.5[[ring]]. It changes approximately 0.5[[ring]] per 50 km. The confirmation of the magnetic deviation to true north is worth for further examinations.
The pressure placed on the corner reflector by the technician reading the inclinometer will affect the final reading. Each kilogram, that is pressed on the front edge of the reflector decreases the inclination. Low temperature, that is below -35 [[ring]]C will affect the inclination as well. At those extreme low temperatures, the liquid in the inclination tube will turn into a viscous medium, that changes its position slowly. These points should all be taken into account to avoid errors in measurements.
[return to top]7.2. Repeatability and Accuracy
Measurements were performed by 4 persons to find out how repeatably and accuratly measurements can be made. We observed 10 corner reflectors. One reflector had too large of a deviation because of a reading error. This reflector, DJ3 is therefore left out of the following calculations.
The first angle we looked at was the azimuth angle. The deviation of the measured angle betweenthe mean of all measurements on the same reflector, measured by 4 persons is shown in table 7.2. The measurements are all given in chronological order, because the observation went over a one day period. Because of a temperature around -35 [[ring]]C, the concentration of each person could be reduced. That result should be seen in larger deviation of the measurements at each reflector by the end of the day. As the standard deviation of each reflector shows the measurements aren't affected by the time of day the measurements are taken. The standard deviation for each reflector varies between 0 and 2.2. The standard deviation of all reflectors together is 1.1[[ring]].
Azimuth Residuals
table 7.2
The deviation of the average inclination measured during this particular field trip, is shown in table 7.3.
Inclination Residuals
table 7.3
[return to top]7.3. Impact on Calculation
The affect of measurement deviations on the RCS was the main reason why I observed all the results of our measurements. As I explained in the first paragraph, it is necessary to know the exact pointing of each corner reflector. Knowing how accurate the measurements are we can calculate the accuracy in the RCS.
All corner reflectors in Delta Junction are triangular trihedral with a 3 dB beamwidth of approximately 40[[ring]] and have a side length of 2.44 m. These values are necessary to calculate the boresight angle, that describes the exact pointing of the reflector to the satellite with the brightest RCS.
Phi: The angle between the satellite ground track and the azimuth pointing angle. Theta: The angle between the reflectors vertical edge and the vector from the relfector apex to the satellite
L: One side length of one reflector panel.
The equations to calculate theta on boresight are:
Phi on boresight, for the corner reflectors used in Delta Junction is 45 degrees. The equation used to calculate the RCS is
With an standard deviation of 1.074[[ring]] in satellite ground track angle and 1.137[[ring]] in inclination, the RCS changes to the values shown in table 7.4.
table 7.4
In the worst case the RCS changes about 0.056 dB from its optimal theoretical value. In comparison with the total admissible RCS deviation value of about 0.5 dB, is that 11.2 %. The total admissible RCS deviation includes all differences between the theoretical RCS and the RCS calculated out of the radar response, e.g. curved reflector panels, angle between panels varying from 90[[ring]] and deviation in measurements as calculated above.
[return to top]7.4. Inclination Measured with a Digital Inclinometer
Because of the difficult reading, and the large deviations with the inclinometer built into the compass, we bought a digital inclinometer, with an accuracy of +/-0.1[[ring]]. As we could see in our fieldwork, a temperature under -10[[ring]] could also affects its measurements. The accuracy then decreases to +/-0.2[[ring]]. Because of the digital display , there aren't any reading errors any more. So we can say the accuracy is +/-0.2[[ring]] or less, depending on the surrounding temperature.
RCS calculations with digital inclinometer.
table 7.5
With the digital inclinometer the worst case deviation decreases to 0.022 dB, which is now only 4.4 % of the total admissible RCS deviation value. In the case with STD even only 1%.
[return to top]8. ScanSAR Survey
The following part is an extract of the RADARSAT ScanSAR survey folder. All worksheets with GPS measurement data, site descriptions and U2 photos are included in this folder.
[return to top]8.1. Reason for the RADARSAT ScanSAR Survey
The Canadian satellite RADARSAT has several beam modes which differ from the ERS1, ERS2, JERS1 satellites. Two of these modes are the ScanSAR's wide and narrow beam modes (see table 2.1). These modes can observe a 500 x 500 km and a 300 x 300 km area with a resolution of 100 m and 50 m, respectively. Both areas are larger than the ASF calibration array in Delta Junction, which is used to calibrate the ERS1, ERS2, JERS1 and RADARSAT satellites. The Delta Junction calibration array measures approximately 60 x 60 km.
However, the ScanSAR beam modes require an area 25 times larger for calibration than that of the Delta array. For this reason, a triangle of the three Highways - the Richardson, Glenn and Alaska Highways - was chosen for this survey. The area enclosed by these highways measures about 300 x 300 km. With all the geometric calibration points measured in this area during the ScanSAR survey, it is now possible to calibrate images recorded with ScanSAR wide and narrow beam modes.
[return to top]8.2. Geometric Calibration Targets
Geometric calibration is used to verify that the location in a SAR image matches its real geographic location. There is also often a distortion in the image that could be corrected by well known target coordinates. If any differences occur, the image can be reprocessed using slightly different parameters to correct deviations.
Ground targets were chosen using satellite images from the European satellite ERS1, because of their similar characteristics to RADARSAT ScanSAR beam modes. ERS1 has been in orbit since 1992. Given this fact calibration values for this particular satellite are well known. ERS1 Lo Res images have a resolution of about 80m, which is close to the RADARSAT ScanSAR wide mode with a resolution of 100m, and also for the narrow beam mode of 50m. ERS1 Lo Res images are processed out of 64 Mbytes Full Res images. Each pixel represents a power average of a 8 x 8 pixel region from the corresponding Full Res image. The Lo Res pixel width is 100 m whereas the range resolution is only 200 m. The absolute location errors of ERS1 SAR data are +/-200 m. The magnitude of these errors are such that ERS1 images could not be used to determine detailed global position information for calibration targets.
Targets were chosen based on their relative brightness or darkness compared to their surroundings in SAR images. Two different types of targets were chosen, 1) area targets like lakes, and 2) point targets such as buildings and road intersections. There are several advantages and disadvantages of using lakes. Lakes can easily be identified in images of summer months, as long as their surfaces are flat and protected from wind. However, they may be invisible in winter images, when water bodies are frozen and snow-covered. Farm fields also often make excellent area targets because of dark appearance in SAR images. Airport landing strips show up also very dark in SAR images. Point targets may be observed by using U2-aerial photography and by physically visiting such sites with a GPS receiver. By using U2-photos it was possible to find point targets like buildings and road intersections and to see if trails lead to measurement points. Further, physically visiting such sites with a GPS receiver allowed their geographic position to be measured.
The area around Delta Junction was left out because there are all the corner reflectors installed. They give us both: geometric and radiometric information whereas all points along the Highways can only be used for geometric calibration.
[return to top]8.3. Analyzing ERS1 Lo Res Images
GIF-files were viewed with the XV software. This program allowed exact image pixel coordinates (x and y) to be measured. It also allowed images to be zoomed in on and their windows cut and pasted into worksheets that were taken into the field. The TWOWAY_TLR program, written by Jason Williams, was used to convert these x, y image coordinates into longitude and latitude. Elevation values were also needed to convert the coordinates. These values were obtained from an atlas with a deviation of +/- 50 m, depending on the area surface.
GIF files were created from raw data files by using the following steps:
1. The header files in the Lo Res images were extracted with the following step:
extract imagefile.dat tempfile
2. To convert the tempfile into a imagefile.gif several programs were used:
raw 1024 1024 tempfile | pgmtoppm white | ppmtogif >imagefile.gif
[return to top]8.4. Site Locations in ERS1 Images
The following ERS1 Lo Res images were used to find point and area targets. This table lists all targets which were chosen before the survey started:
| Image Files Name | Flight Directions | GPS File Names (see Abbreviations) | Date of Flight |
| 192793200 | ascending | FBKSDJ1 | 10/14 1995 |
| 188623200 | descending | FFBKSDJ2 | 12 / 15 1995 |
| 185362200 | ascending | FBKSDJ3, 4, 5, 6, 9 | 10 / 11 1995 |
| 197570200 | descending | FBKSDJ7, 8, 10 | 9 / 22 1995 |
| 150693200 | descending | DJG1, 3, 4, 5, 6, 7, 8 + DJD1 | 11 / 9 1995 |
| 150694200 | descending | DJG9, 10, 11 | 11 / 9 1995 |
| 183618200 | ascending | GGLEN1, 2, 3 + GLENN2, 3, 4, 5 | 10 / 13 1995 |
| 189291200 | ascending | GGLEN4, 5, 6 | 11 / 8 1995 |
| 150695200 | descending | GLENN1 | 11 / 9 1995 |
| 183417200 | descending | GTOK1 | 8/ 28 1995 |
| 183566200 | ascending | GTOK2, 3 | 9 / 22 1995 |
| 183119200 | descending | GTOK4, 5, 6, 7, 8, 11, 12 | 5 / 31 1995 |
| 191870200 | ascending | GTOK9, 10 | 10 / 24 1995 |
| 183118200 | descending | GTOK13, 14 + TOKE2 + TOKDJ1, 2, 8 | 5 / 31 1995 |
| 192371200 | descending | TOKE1 + TOKDJ3, 4, 5 | 10 / 2 1995 |
| 191434200 | ascending | TOKE3 | 10 / 5 1995 |
| 19284200 | descending | TOKDJ6, 13 | 10 / 22 1995 |
| 192952200 | ascending | TOKDJ7, 9, 10, 11, 12 | 11 / 12 1995 |
| 400 km2 image | ascending | DJG2, 12 |
In total, 52 ERS1 Lo Res images were ordered and analyzed to cover all sites. However, 25 images either did not cover the surveyed area or they were supplementary images.
It was also possible to obtain one additional image with an approximate size of 400 x 400 km. This image was a mosaic of several ERS1 images which offered two more sites and confirmed that the chosen survey area was the correct size to calibrate ScanSAR images.
The picture below shows the 400 x 400 km mosaic of ERS1 images.
GIF(1.1MB)
or
JPG(328KB)
figure 8.1
The following two indexes illustrates all ERS1 Lo RES image data ordered for the ScanSAR survey. The first index illustrates ascending images and the second illustrates descending images.
Ascending Image Index
Descending Image Index
figure 8.3
[return to top]8.5. GPS Settings
More precise positions could be controlled by changing two settings in the GPS unit: 1)How many satellites are tracked simultaneously by the receiver and 2) how those satellites are geometrically distributed in the sky. Position Dilution of Precision (PDOP) is a measurement that expresses the relationship between the error in user position and the satellite geometry. It is calculated by the GPS to define the accuracy of the collected data during the measurement. A maximum PDOP can be set in the GPS unit, which allows the GPS only to record data if the PDOP is under this limit. PDOP values of 4 or below gives excellent positions. The Pathfinder software PROPLAN allows PDOP predictions for planning GPS surveys by using almanac files downloaded from satellites. Plots of each days PDOP readings were printed out and used to plan GPS surveying activities.
An additional option to get a more accurate position is to set the signal to noise ratio (SNR) to a higher value so that the GPS receiver only selects those satellites with a strong signal. This minimizes the possibility to collect reflections by mountains or metal objects near the measurement.
The GPS was always run in three dimensional mode (3D), to collect additional elevation data. The 3D mode also gives a more accurate position than the two dimensional mode, because of more satellites are tracked. At least four satellites are necessary to get a representative position.
On every site except on GTOK7 the following GPS settings were used to collect global position data:
Max. PDOP: 04 Elev. Mask: 20 SNR Mask: 14 dB Recording Position: every 1 second.
GTOK7 had to be measured between trees, to enable the receiver tracking more satellites the SNR had to be decreased to 4 and the maximum PDOP setting was increased to 6. In this case the GPS receiver was able to track 5 satellites.
[return to top]8.6. Survey Sites
Fifty-two sites were measured during the ScanSAR survey. These sites can be divided into the following subgroups:
| Area | Lake | Gravel Pit | Airport | Pipeline | Bridge | Buildings | Total |
| FBKSDJ | 4 | 1 | 1 | 4 | 10 | ||
| DJG | 5 | 1 | 1 | 2 | 9 | ||
| DJD | 1 | 1 | |||||
| GGLEN | 2 | 1 | 1 | 4 | |||
| GLENN | 1 | 2 | 1 | 2 | 6 | ||
| GTOK | 2 | 2 | 4 | 8 | |||
| TOKE | 1 | 1 | 1 | 2 | 5 | ||
| TOKDJ | 2 | 3 | 2 | 2 | 9 | ||
| Total | 15 | 9 | 5 | 2 | 3 | 18 | 52 |
table 8.2 Unreachable sites:
| Travel Route | Abbreviations | Site Number (not surveyed) |
| Fairbanks-Delta Junction | FBKSDJ | 9, not all sites surveyed yet |
| Delta Junction-Glennallen | DJG + DJD | 1, 4, 9, 11 + 2 |
| Glennallen | GLENN | all selected sites were surveyed |
| Glennallen-Gulkana | GGLEN | 3, 4, 6, |
| Gulkana-Tok | GTOK | 3, 4 , 6, 8, 11, 12 |
| Tok East | TOKE | 6 |
| Tok-Delta Junction | TOKDJ | 2, 6, 11, 12 |
table 8.3
[return to top]8.7. Survey Planning
The measurements from Fairbanks to Delta Junction could be divided into 3 to 5 parts. In which case it would be possible to survey those parts of the project during a normal Delta maintenance trip. For the other survey points I calculated 15 minutes for each site near the Highway would be required, and the rest of these sites far off the Highway are shown in the table below:
All the following numbers are estimates for how long it could take to perform the GPS measurements for each site:
| Road Trip | Miles | Driving Time | Survey Time (Hours | Hikes (Hours) | Total Time Needed |
| Fairbanks-Delta Junction | 100 | 2 | --- | --- | 2 |
| Delta Junction-Glennallen | 200 | 4 | 9*15=>135 min. | 2 + 1 + 1 | 10 |
| Glennallen | 100 | 3 | 10*15=>150 min. | 2 | 7.5 |
| Glennallen-Tok | 175 | 3 | 10*15=>75 min. | 1 + 2 + 1 | 9.5 |
| Tok | 100 | 3 | 11*15=>165 min. | 1 + 1 | 8 |
| Tok Delta- Junction | 125 | 3 | 6*=>120 min. | 2 | 7 |
| Delta Junction- Fairbanks | 100 | 2 | 2*15=>30 min. | 1 | 4 |
| Total | 900 | 21 | 12 hours | 15 hours | 48 hours |
table 8.4
- day: Fairbanks - Glennallen
- day: Glennallen - Gulkana plus first survey site to Tok
- day: Gulkana - Tok, leave the last 2 sites out
- day: Tok + East of Tok + the 2 sites form the 3 day + 5 sites to Delta Junction
- day: Tok to Fairbanks if possible some sites between Delta Junction and Fairbanks
8.8. List of Abbreviations
FBKSDJ: Richardson Highway (Fairbanks to Delta Junction)
DJG: Richardson Highway (Delta Junction to Gulkana)
DJD: Denali Highway
GGLEN: Richardson Highway (Gulkana to Glennallen)
GLENN: Glenn Highway (Glennallen, west direction)
GTOK: Tok Cutoff, or Glenn Highway (Gulkana to Tok)
TOKE: Alaska and Taylor Highway (east of Tok)
TOKDJ: Alaska Highway (Tok to Delta Junction)
[return to top]9. Conclusion
of my winter internship
Even if I was the first winter intern and everybody thought it might be boring during the winter here in Fairbanks, I can prove the opposite. There are many events that take place only during the wintertime, e.g. Yukon Quest race, Ice Art Competition... and Tom's flight over Fairbanks, that I enjoyed very much! Also a big thanks to Parker, who supported me during the ScanSAR survey and to all other people working at ASF, they helped me to have such a great time in Alaska. The working conditions are excellent at ASF.
I want to thank especially my supervisor Jason Williams, who always understood my problems very fast (despite the language problems) and gave me helpful advise how to proceed. His patience is unbelievable!
I had planned to do my "Praktikum" in a foreign country in order to improve language skills and get in contact with a different life style of people, let's skip the language, the latter was very interesting.
Thank you all for the great time!!!
[return to top]10. Useful literature
Official Calibration Plan 1.0, Jason Williams
RADARSAT International, World Wide Products & Services, Price List
Synthetic Aperture Radar, John C. Curlander/Robert N. McDonough
Radar Cross Section Measurements, Eugene F. Knott
GPS Pathfinder, Trimble Navigation
[return to top]