Radiometric Calibration, Spatial Characterization, and Spectral Evaluation of the Advanced Land Imager and Hyperion Sensors

Principal Investigator

Dr. Stuart Biggar, Remote Sensing Group, Optical Sciences Center, U. Arizona

Co-Investigators


 

Spatial Performance Analysis

1.0 Calibration/validation of ALI spatial response and MTF and comparison to ETM+

We propose to measure the ALI spatial response to validate the pre-launch predicted performance and to monitor any on-orbit changes over time.

1.1 Background

Previous experience with Landsat-4 and -5 TM indicate that the spatial response of that design extends over about 40-45m, rather than the 30m pixel often assumed.
Because of the whiskbroom design of TM, the spatial response is notably asymmetric in-track and cross-track (Schowengerdt, 1997), with a wider response cross-track due to an electronic low-pass filter in the data stream.
The spatial response characteristics of ALI are expected to be substantially different from TM and ETM+ since ALI is a pushbroom array imager. For example, if time integration is used in-track to increase SNR, then a broader spatial response will be expected in-track than cross-track. At this time, we do not have the sensor engineering details necessary to make such a conclusion, and will need relevant engineering design and test data for ALI to support our on-orbit tests.

1.2 Approach

The ALI, Hyperion and ETM+ spatial response functions will be measured with three techniques:

1) Analysis of linear targets such as bridges over water

This technique has advantages of high signal contrast and subpixel sampling dependent on the angle of the target to the in-track and cross-track directions (Schowengerdt et al, 1996). However, it requires imagery of specific targets and only permits 1-D measurements orthogonal to the target. Possible targets are the San Mateo Bridge over San Francisco Bay (18.3m wide at 31.1° to cross-track), which we've used previously to evaluate TM (Schowengerdt et al, 1985), and the causeway over Lake Pontchartrain, Louisiana (two 10m spans, separated by 12m, at <1° to in-track ). The latter target has several unique characteristics, including a straight 26 mile length and alignment to the TM path within 1° (see Fig. ?). These features, combined with the 12-bit quantization of ALI and Hyperion data, allow high precision cross-track subpixel sampling (the incremental subpixel phase from line-to-line is about 1/70 pixel for TM) for alias-free MTF analysis and precise unmixing tests (described below). Furthermore, the nearly in-track orientation of the Causeway means that we will be using only a few adjacent detectors in the pushbroom array, minimizing any required relative calibration effort.

Both of these targets are near the center of the Landsat path. To acquire them with ALI and Hyperion will require "rolling" of the spacecraft so both sensors view nadir, which we understand can be done by special request.

The radiometric profile across a bridge over water will vary with solar angle due to shadows (see Fig. ?). We will calculate the size of these shadows for given image acquisitions based on the bridge geometry and height above water. Spectroradiometric ground measurements will also be used to determine the relative reflectances of the bridge and water. The MTF analysis then requires a correction for the radiance profile of the bridge as described by Schowengerdt (1985).

2) Comparison of relative spatial frequency content of ALI to ETM+ images

This technique can be used on any near-coincident pair of ALI and ETM+ images. Temporal changes are not of concern because of their one minute separation. The spatial response of ALI relative to that of ETM+ will be measured in terms of a relative MTF (Schowengerdt et al, 1985) and can be obtained for bands ALI-2 (TM-1), -3 (TM-2), -4 (TM-3), -8 (TM-5), and -9 (TM-7). The other ALI bands do not match those in ETM and a relative image comparison is therefore not valid. Information on relative SNR may also be obtained from this analysis. A drawback is that the scene must have high spatial detail and radiance contrast in order to obtain reliable MTF measurements.

3) Comparison of relative spatial frequency content of ALI-pan and ALI-2, -3 and -4 bands

Any ALI image set that includes data from these four channels can be used for this technique. The MTF of the 30m bands (in the visible) relative to the 15m panchromatic band will be obtained. This information will be useful in fusion of ALI-pan and ALI-MS bands. The same advantages and drawbacks as in (2) above apply to this technique as well.