Appendix I - Description of the AISA Instrument and Data Set

This hyperspectral aircraft imagery of the Milford Neck Wildlife Area, Delaware was acquired on 25 September, 1999 using an AISA imaging spectrometer flown by the 3DI corporation of Easton MD. The AISA is a PC controlled push-broom imaging spectrometer giving 12 bit data within the 430 to 900 nm spectral range in 288 spectral bands. There are 360 spatial pixels per swath in a 23 degree field of view. This yields 1 meter wide pixels when the aircraft altitude is 1000 m. Pixel width increases linearly with altitude. Pixel length is determined by airspeed and integration time and is usually adjusted to be similar to the pixel width.

The detector is thermo-electrically temperature stabilized. Incident solar radiation is collected by a sensor on the top of the aircraft and fed to the spectrometer in an additional 20 pixels so that a full spectrum of incident solar radiation is recorded with each scan to allow post processing computation of apparent reflectance. Differential global positioning data and inertial aircraft attitude information are recorded with each scan to allow geo-referencing each scan during post processing.

In post processing by the 3DI company the spectral observations were converted to reflectance at the aircraft level by using the in-flight observations of incident solar radiation. Image reflectance, over the range 0 to 1.0, is scaled in integers from 0 to 10,000. Since the observations are quantized in 12 bits, each bit is equivalent to 2.44 reflectance units. This level of precision is often described as 3 ½ significant digits.

In-flight observations of differential GPS and aircraft motion were used to georeference the position of the observations and correct for aircraft motion. The flight strip data were delivered as generic binary files referenced to UTM, meters, Zone 18.

Table I-1 shows the spectral channels assigned for this flight. Note that the gain of the shorter wavelength bands was increased to compensate for lower sensor sensitivity in that spectral region. Doing so resulted in observable noise in the data of bands 1 to 4.

Table I-1
Wave-band AISA Chan - Min. wvl Center wvl Max. wvl Chan. width Avg. offset Avg. gain
1 22 4 458.54 461.78 465.02 6.48 0 26.5205
2 29 4 469.89 473.13 476.37 6.48 0 22.9378
3 35 4 479.61 482.86 486.1 6.48 0 18.4235
4 41 4 489.34 492.58 495.82 6.48 0 14.5105
5 47 4 499.07 502.31 505.55 6.48 0 11.8414
6 53 4 508.79 512.03 515.28 6.48 0 10.0721
7 59 4 518.52 521.76 525 6.48 0 8.8712
8 66 4 529.87 533.11 536.35 6.48 0 7.8921
9 72 4 539.59 542.83 546.08 6.48 0 7.2363
10 78 4 549.32 552.56 555.8 6.48 0 6.6292
11 84 4 559.17 562.55 565.93 6.76 0 6.1127
12 90 4 569.31 572.69 576.07 6.76 0 5.7445
13 96 4 579.45 582.83 586.21 6.76 0 5.553
14 102 4 589.59 592.97 596.35 6.76 0 5.5433
15 108 4 599.73 603.11 606.49 6.76 0 5.6339
16 114 4 609.87 613.25 616.63 6.76 0 5.704
17 125 4 628.46 631.84 635.22 6.76 0 5.6234
18 137 4 648.74 652.12 655.5 6.76 0 5.2642
19 149 4 669.02 672.4 675.78 6.76 0 4.8171
20 155 4 679.16 682.54 685.92 6.76 0 4.6681
21 161 4 689.3 692.68 696.06 6.76 0 4.6157
22 167 4 699.44 702.82 706.2 6.76 0 4.6444
23 172 4 708.45 711.97 715.49 7.04 0 4.6891
24 178 4 719.01 722.53 726.05 7.04 0 4.7774
25 184 4 729.57 733.09 736.61 7.04 0 4.9535
26 189 4 738.37 741.89 745.41 7.04 0 5.1291
27 195 4 748.93 752.45 755.97 7.04 0 5.3861
28 212 4 778.85 782.37 785.89 7.04 0 6.2647
29 227 3 805.25 807.89 810.53 5.28 0 9.4876
30 246 3 838.69 841.33 843.97 5.28 0 11.2467