Airborne Hyperspectral Imaging Using the CASI1500

Transcription

1 Airborne Hyperspectral Imaging Using the CASI1500 AGRISAR/EAGLE 2006, ITRES Research

2 CASI 1500 overview A class leading VNIR sensor with extremely sharp optics. 380 to 1050nm range 288 spectral bands ~1500 spatial pixels up to 30cm spatial resolution Water Quality - Land Cover - Forestry

3 Sensor Head Unit Instrument Control Unit (PCU) Sunlight- readable Display POS 410 v5. IMU Keyboard w. tracking ball

4 Mission Parameters System to be installed coincident with AHS on CASA-212 AGRISAR nm FWHM 6 flight lines per sortie EAGLE -60 Predetermined bands within nm (VNIR). Aircraft velocity slowed for maximum spatial resolution Nominal pixel resolutions in raw data 288 band = 1.5 * 6 m 60 band = 1.3 * 0.5m / 1.3 * 1.3m

5 CASI 1500 Installation (Quatro Vientos )

6 Acquisition Summary - Agrisar Flight line configuration for AgriSAR acquisitions. High overlap = less bidirectional effects between flight lines Best-fit flightline design.

7 Acquisition Summary- EAGLE Flight line configuration for EAGLE acquisitions, Cabauw Block. Flight line configuration for EAGLE acquisitions, Speulderbos and Loobos Blocks High overlap = less bi-directional effects between flight lines Best-fit flightline design.

8 June 01, 2006 CASI integration with AHS in CASA June 06, 2006 AGRISAR 1 st Mission 6 288b, acquired in afternoon (~13:00 UTC) due to Wx isolated clouds in area. CASI Hardware problem does not allow for subsequent CASI images for 1 st Agrisar campaign (no data June 10 th ) June 13, 2006 Eagle data acquisitions 11 lines over three areas of 60bands, acquired in afternoon (~11:00 UTC) Isolated clouds in area, increasing during the acquisition, hazy atmospheric conditions throughout. July 04 & 05, 2006 AGRISAR 2 nd Mission 6 lines 288b, acquired at solar noon (~10:00 UTC). Clear conditions on both days Bundle Adjustment Flight performed on July 06, 2006 over NeuBrandenburg.

9 1 4 Radiometric Corrections : Raw to Pix 3 Removal of Instrument Noise & Dark Current effects Traceable Standard!Persistent Blue end calibration source effect 2 Raw vs. Radcorr Spectra Comparison (DN vs. SRU (1 SRU = 1.0 µw cm-2 sr-1 nm-1 )

10 Notes on Imagery Calibration CASI1500 imagery acquired at maximum spectral resolution & minimun integration time can be expected to have some image artifacts (noise) in parts of the spectrum where the SNR is lower. Data sets where imagery is summed spectrally (i.e 48 spectral bands = 6 rows summed) do not contain the same noise effects due to the reduction of the SNR through summation. Calibration of CASI1500 done to standardized light source with an accuracy of ± 2% over the spectral range of 470 to 800 nm. Below 470 nm and above 800 nm, the quoted accuracy is ± 5%.

11 Effects of Clouds in Imagery L4 June 6 th Example of flight line illumination difference in mosaiced CASI1500 data (0705 L5/6) BiDirectional effects minimized, not cancelled out

12 Geocorrection process involves the integration of four separate data streams: Radiometric correction software extracts GPS time stamp from CASI data Aircraft GPS data are differentially corrected (DGPS) using data from a nearby GPS base-station. DGPS positions with aircraft attitude data from the Inertial Measurement Unit (IMU) are generated. A Digital Elevation Model (DEM) is applied during geocorrection to remove topographic effects and facilitate the final orthorectification of the CASI imagery. Position and attitude data are optimized using sensor misalignments determined in bore-site calibration (bundle adjustment process) and applied to the CASI image data. = A north-up image with square fixedsized pixels is populated using a nearest neighbor algorithm. OR Unaltered image pixels mapped to a Geographic Lookup Table

13 Bundle Adjustment : inputs Ground Control Point (GCP) XYZ Coordinates with reference map (Courtesy of University of Neubrandenburg ) Identify GCP in nongeoreferenced imagery, record image X,Y values for each point Through use of multiple points, each in +2 adjacent flight lines, Sensor/GPS/IMU angular offset parameters are iterated. Any data flown during same installation can be solved using offsets.

14 Bundle adjustment : Output XYZ offsets, ωs, φs, κs values entered into navigation data refinement program > Geocorrection is done on these output files. Boundary between two flight lines note building offsets due to coarse DEM! DEM coarseness affects linear offsets between flight lines.! Very high resolution DEMs slow down processing if integrating data into a CASI image.

15 + Pos Data + SAR Data = + + Bundle Adjustment Radiometric-corrected CASI image

16 Radiometric / Geocoded hyperspectral Imagery Copyright ITRES 2006

17 Fused SAR/CASI mosaic- AgriSAR Optical N 2x Vertical Exaggeration Note CASI line boundary 11x Vertical Exaggeration

18 Delivered Data format: 16-bit unsigned integer image files (BIP) Accompanying.glu file ( with ENVI header file).glu file : output Geographic lookup table file 2 channels : Easting, Nothing values in UTM WGS84 for center of each pixel. No resampling! Average terrain Height used in Geocorrection due to extents of imagery being outside DEM range introduces errors in.glu file write out ~28 Gb delivered data

19 SENTINEL 2 Simulations CASI imagery for July 5 th, 2007 acquisition used to simulated Sentinel- 2 VNIR band data. Imagery averaged over rows that best fit bandset using bandmath function in ENVI. Some non-noise image artifacts enhanced by summing, instrument noise averaged out Spatial resolution left at original setting for analyst manipulation. Image files contain both bands 8 & 8a.

20 Conclusions & Acknowledgements Copyright ITRES 2006