Apr 10, 2001 IIP Proposal Summary 1

Unified Onboard Processing and Spectrometry

Unified Onboard Processing and Spectrometry

Murzy Jhabvala (550)

Sarath Gunapala (JPL)

Peter Pilewskie (Ames)

Michael King (900)

Warren Wiscombe (913)

Peter Shu (553)

Pen-Shu Yeh (564)

PI: Si-Chee Tsay (913)

Apr 10, 2001 IIP Proposal Summary 2

PrologueRemote sensing in the Earth sciences has grown remarkably over the last decade. This growth has been spurred by remarkable advances in technology... Of these advances, the merging of spectroscopy and imaging has been the most important. Spectroscopy has been used as a quantitative tool in the laboratory for many years and there exists a wealth of understanding and analysis strategies for such data. Although early imaging spectrometer instruments suffered through the usual development problems, these systems are now approaching the spectral resolution and quality of laboratory measurements, blurring the distinction between the two but also bringing some of the most advanced laboratory spectral analysis techniques to bear on complex Earth science problems. The most advanced sensors and instruments are currently mounted on aircraft, but there are exciting plans to integrate the best of these into orbiting platforms which will facilitate greater accessibility and wider geographic coverage.

Remote Sensing for the Earth Sciences: Manual of Remote Sensing, 3rd ed., Vol. 3, A. Rencz, ed., Ch. 5

Apr 10, 2001 IIP Proposal Summary 3

Proposed Activities

• Select a combination of “dumb” (proximate differencing) and physics-based lossless spectral compression algorithms for shortwave spectra that will be widely accepted within the community

• Develop custom chips for onboard processing of spectrometer data using those algorithms

• Integrate those chips into:– Leonardo Airborne Simulator

– Quantum Well Infrared Photometer

• Conduct flight validation tests

Apr 10, 2001 IIP Proposal Summary 4

Why is onboard compression needed?

• Data rate!– think of MODIS; then, multiply by 10+

• Generic spectrometer data rate– 1Kx1K 12-bit image, 200 wavelengths = 2.4 Gb

– one such image every 100 km (14 s) => 170 Mb/s

– total data in 94-min orbit: 1 Tb

– 10-min download requires 1.6 Gb/s (ultra-high rate)

• The “archive all the raw bits” paradigm has reached the end of its utility (EOSDIS x 10?)

Apr 10, 2001 IIP Proposal Summary 5

AVIRIS Image Cube Shows the Problem: Firehoses of Data

Apr 10, 2001 IIP Proposal Summary 6

What is Compression, Really?

• Just a way of “flattening” a data object

• A grey or flat object compresses perfectly

• A spectrum with only a few mild ripples is much more compressible than one with big variations...so, get rid of the variations!

• Think of it as removing known information– model spectra, lab spectra, empirical spectra...

– why send known information to ground 1B times?

Apr 10, 2001 IIP Proposal Summary 7

How to Compress?

• using a priori knowledge; divide out– extraterrestrial solar spectrum

– known absorption spectra

– known scattering spectra (e.g. Rayleigh)

• use current knowledge– e.g. divide by spectra contiguous in space or time

» MPEG, HDTV have pioneered this road

– principal components

• best: combinations of both strategies

Apr 10, 2001 IIP Proposal Summary 8

How to Compress in Hardware?

• ASIC: Application Specific Integrated Circuit

• FPGA: Floating Point Gate Array

• Pen-shu Yeh/553 designed a JPEG ASIC

• Code 935 is studying image navig’n using FPGA’s

• DoD is now coding numerical methods like tridiagonal solvers into ASIC’s and FPGA’s

• This technology has come of age!

• Onboard processing is a big NASA goal, but no one has done much about it yet.

Apr 10, 2001 IIP Proposal Summary 9

Candidate Spectrometers

• Leonardo Airborne Simulator (Tsay/Shu)– single array detector, 0.4 to 5 microns—a first!

– operational: aircraft version flew in SAFARI 2000

• Quantum Well Infrared Photodetector (Jhabvala)– the thermal IR, in chunks (3-5, 8-10, 10-12, 14-16 m)

– operational: aircraft version

10

4x8 ch. pre-ampsdewar

data storage4x 26 GB

2x16 ch. A/D converters14 bits, 2MHz10 frame/sec

computerdetector/optics module

80°

FO

V

13.5° FOV

18”

SAFARI-2000SAFARI-2000South Africa OR

Leonardo Airborne SimulatorLeonardo Airborne Simulator

Apr 10, 2001 IIP Proposal Summary 11

• ALADDIN astronomical-quality 1024 x 1024 InSb detector; high quantum efficiency in 0.4–5.5 µm.

The Heart of Leonardo Airborne Simulator: ALADDIN Array Detector

Apr 10, 2001 IIP Proposal Summary 12

0.4 - 2.5 µm

or, 2K x 2K HgCdTe array detector @ 150 Kelvin

1K x 1K InSb array detector @

50 Kelvin

90º FOV

wedge filter

Leonardo Spectrometer for Space

Apr 10, 2001 IIP Proposal Summary 13

– Quantum Well Infrared Photodetector

– 256 x 256 GaAs detector array @ 65K

– 9° FOV

– 16 Hz frame rate

– programmable integration time

– calibration: cold and hot blackbody sources in lab

Flies on Aerocommander

(10 km ceiling)

QWIP Airborne Spectrometer

Apr 10, 2001 IIP Proposal Summary 14

Quantum Well Infrared Photodetectors

Apr 10, 2001 IIP Proposal Summary 15

QWIP Optical Layout

Apr 10, 2001 IIP Proposal Summary 16

The deer is over 100 m away and not visually discernible.

QWIP as Deer Detector