John Bolton's Initiatives - Appendix B

APPENDIX B: Other Proposals, and Papers

 

 

* Report On The Development Of The Airborne Imaging Spectrometer
* New Systems And Technologies For Optical Remote Sensing
* A Rough Draft Of A Preliminary Conceptual Design For A High Spatial Resolution Hyperspectral Imager (HSRHI)
* Strawman Design For A Hyperspectral Transfer Radiometer And Follow-Ons
* Plans For An Extended And Operational Airborne Remote Sensing Program
* And Now For Something Completely Different

 

 

Report On The Development Of The Airborne Imaging Spectrometer
John Bolton
Goddard Research Fellow
Technical Research Center of Finland
Instrument Laboratory
Ittuulentie 2
SF-02100 Espoo
FINLAND
tel: +358 0 456 4369 fax: +358 0 456 4496
e-mail: "john.bolton@vtt.fi" or bolton@ins.vtt.fi

Permanent Address:
NASA/Goddard Space Flight Center
Systems Engineering Office
Code 704
Greenbelt, MD
20771 U.S.A.
tel: (301) 286-8547 fax: (301) 286-1600
e-mail: “jbolton@gsfcmail.nasa.gov” or “n2jfb@vax720.gsfc.nasa.gov



Introduction

The goals of my Goddard Research and Study Fellowship were 1) to develop cooperation between Finnish institutions doing space research and NASA/Goddard, and 2) to help to develop the capability to build space qualified instrumentation in Finland. The means to accomplish these goals was to design and build an airborne imaging spectrometer (AIS) at the Technical Research Center of Finland (VTT).

The AIS is an instrument which would be very useful for work being done at NASA/Goddard. For years, a similar instrument, the ASAS, has been used for airborne campaigns. This instrument, being rather old and using obsolete technology, should be replaced by a modern instrument.

One of the reasons for proposing to build an AIS in Finland is that the technology for such an instrument is now available off-the-shelf to a large extent. The only instrument sub-system which cannot be bought directly is the optical system and spectrometer. There are several solutions to this problem which will be discussed in detail later. The areas in which development is needed, microprocessor technology and control software development, can be easily handled by the Finns, who are certainly the equal of anyone in the world when it comes to this type of technology.

Background

During the work on the conceptual design of the MODIS-T instrument at Goddard, it was suggested that an airborne prototype be built. This would have provided excellent experience in the development of an imaging spectrometer, and would have given Goddard researchers a state-of-the-art airborne instrument.

Unfortunately, this method for development of the MODIS-T instrument was not followed. It is quit likely that if this method would have been used, the development of MODIS-T would have proceeded much more successfully.

There are several commercially available airborne imaging spectrometers. Most are very expensive systems which are not really imaging spectrometers but scanners. The only instrument which is equivalent to the AIS is the CASI instrument build by ITRES Research in Calgary, Canada. This instrument employs some of the same technology as the AIS, but does not have all the capabilities that up-to-date technology would allow. There are several other problems with the CASI that would be solved by the Finnish-NASA/Goddard cooperative development of the AIS.

The development of the AIS in Finland proceeded rather slowly at first, due to circumstances which are described in a separate report (1). Briefly, the laboratory where I worked during my fellowship, the Remote Sensing Section of the Instrument Laboratory (INS) of the Technical Research Center of Finland (VTT), had not made any preparation for the work that I had proposed, even though we had discussed the project for more than a year before my arrival in Finland.

Therefore, my first task after getting organized in Finland, was to write a proposal for the development of the AIS. A proposal was necessary because in order to do the project at VTT, funding had to be secured from the Finnish technical development agency, TEKES. During initial discussions with TEKES, it was suggested by TEKES that my year in Finland should be spent doing a feasibility study for the AIS. It was clear that nobody at TEKES or at VTT had any idea of the background work that had already been done for the AIS. The writing of the proposal was completed within one month2. The proposal contained a very complete explanation of the need for the AIS, the applications of the AIS, and the advantages for Finnish business to building the instrument in Finland.

The next step was to decide how the instrument would be built and who would build it. As was mentioned earlier, no preparation had been made by the Instrument Laboratory (INS) at VTT. Discussions with the management and staff of the Laboratory showed that it was not possible to do the development of the AIS using the staff and facilities of the Instrument Laboratory.

Fortunately, it turns out that there is another laboratory of VTT which does instrument development work. It is the Optoelectronics Laboratory (OPE) of VTT and it is located in Oulu. The OPE has, in fact, better capabilities for building instrumentation than the Instrument Laboratory. The OPE has capabilities in optics, electronics and microprocessor technology which the INS does not have.

In addition to the instrumentation capabilities of the OPE, the management was much more open to the idea of innovative new projects. The OPE had, in fact, recently built an instrument which incorporated some of the features that would be useful for the development of an imaging spectrometer. Not only that, but the OPE was also considering proposing to TEKES the development of an imaging spectrometer for industrial applications. It was in fact, through TEKES and not VTT, that I learned about the interest of OPE in this type of work. The management of INS, in contrast to that of the OPE, was quite closed-minded and very unreceptive to any suggestions.

In order to secure TEKES funding, it was necessary to secure Finnish industrial involvement. During the writing of the proposal and prior to its submission to TEKES, discussions were held with many Finnish companies that might have had the capability to build the AIS. None were interested in the project. The reasons for this are discussed elsewhere1. A small company, which had previously worked with the Remote Sensing Section in the development of a video based false-color camera, finally agreed to be the industrial sponsor. The company, Karelsilva OY, was probably the best choice, as it turns out. They had already looked into buying a CASI instrument, and their lead applications person, Bart Braam (a Dutch forester working permanently in Finland), had done a considerable amount of research on the availability of airborne instrumentation.

Another problem, and one which was never solved, is that of doing the instrument performance analysis and, once the instrument is built, the instrument characterization. It was originally supposed that the Remote Sensing Section of the INS, in cooperation with Karelsilva, would do this work. The need for this work was not recognized by anyone at the INS. The OPE management recognized the need, but was not able to find anyone who could do thework.

Consequently, this basic component of the instrument development work has not been done. The solution that I have proposed is that this work be done by Goddard. The plan for the continuation of the development of the AIS3 describes the work that will be done by VTT and by Goddard and the schedule for this work.

Justification

It is quite clear that this type of AIS will be built by someone very soon. Now that nearly all of the technology is available off-the-shelf, it is a simple matter for a company with the appropriate expertise, to build an AIS. This instrument could very easily be built by ITRES, for example. The reason that I am proposing that Goddard cooperate with the Finns in the development of the AIS, is that it would be a very straightforward arrangement. Goddard would get an instrument, which presumably would be continuously developed by the Finns. The Finns would get the experience, working with Goddard, in instrument characterization and applications.

The advantage for Goddard is that it could more or less freely specify the instrument. This would not be possible with a commercially developed instrument, unless NASA was paying the bills. Needless to say, NASA will pay nothing for the Finnish part of the AIS development. The advantage for Finland is that they will get a crash course in AIS applications and characterization, plus the international recognition that their instrument will receive if it is used in Goddard airborne campaigns.

An article on the AIS, prepared for the IGARSS'92 Conference in Houston, gives a general overview of the instrument4. The interest in this instrument at IGARSS was amazing. It is clear that people in the airborne remote sensing business really want a good, cheap, versatile airborne instrument. The advantage to having an instrument of your own, which can be flown over targets that you have tested and understand cannot be underestimated.

Instrument Specifications

The following table lists the general specifications of the prototype AIS. It should be borne in mind that this is a prototype instrument. The specifications and capabilities of the second generation instrument, an instrument which will be suitable for operational use, are discussed in the summary of this report.

Parameter

Value

Units

Comments

Spectral Resolution

1.3

Nanometers

per pixel

Spectral Registration

0.5

Nanometers

 

Minimum Wavelength

450

Nanometers

may be 400

Maximum Wavelength

900

Nanometers

may be 1000

Spectral Range

450

Nanometers

Variable

Spatial Resolution

384

Pixels

 

Field-Of-View (Swath)

28

Degrees

 

Field-Of-View (Track)

45

Degrees

with tilt

Inst. F-O-V

1

Milliradians

 

Dynamic Range

5000

N.D.

 

A/D Quantization

12

Bits

 

A/D Conversion Rate

171

Ksamples/Sec

 

Data Rate

2.084

Mbits/Sec

 

Data Capacity

2

Gigabytes

Exabyte

On-Chip Spectral Summing

7

Pixels

 

Radiometric Accuracy

5

Percent

 

Radiometric Stability

1

Percent

 

Weight

<50

Kilograms

 

Power

<200

Watts

 

Input Voltage

24

Volts

220 volt
Inverter

 

Operating Modes

Imaging

384 Pixels

1 – 16 Bands

Multispectrometer

1 – 47 Pixels

288 Bands

Full-Frame

384 Pixels

288 Bands

 

Dispersion System: Volume Holographic Grating
CCD: Thomson TH7863

Instrument Technical Description

The instrument will be described in several parts. These are:

1. The Scan Subsystem
2. The Calibration Subsystem
3.
The Fore-Optics
4.
The Field-of-View Definition Subsystem
5.
The Dispersing Subsystem and Associated Optics
6.
The Detector
7.
The Detector Drive Electronics
8.
The Preamplifier/Charge Converter
9.
The Analog-to-Digital Converter
10.
The Experiment Controller
11.
The On-board Processor
12.
The Data Storage and Compression Subsystem
13.
Ground Data System

Parts 1 through 8 comprise the instrument front-end or sensor head. This is a shoe box size unit (330 x 230 x 180 mm) with a lens on one end and cables coming out of the other end. There are provisions for mounting brackets on the sides, top, and bottom of the sensor head. Parts 9 through 12 are contained in the instrument controller which is basically a desktop, IBM compatible, personal computer (PC).

he front-end is made as small as possible to facilitate mounting. The prototype AIS is fitted to be mounted in a hole in the floor of a single-engined Cessna aircraft. It can be tilted fore and aft to a maximum of 45 degrees.

The PC is a normal desktop, 33 MHz Compaq 80486, EISA bus, personal computer. The decision to use a standard desktop PC rather than a ruggedized unit was made because experience has shown such units to be rugged enough for highway operation (which is more severe with respect to vibration), and to save money.

1. The Scan Sub-System

The AIS uses a pushbroom type target viewing system so there is no scan subsystem. The pushbroom optical system simply images the ground passing below the instrument onto the entrance aperture (field-of-view definition subsystem) of the spectrograph (dispersing subsystem). Modifications of this subsystem for roll, tilt, and pitch compensation, as well as for off-nadir viewing, are described below.

2. The Calibration Subsystem

There is no optical calibration subsystem in the prototype AIS. It was considered that the portability of the AIS would allow radiometric calibration immediately before and after each flight using a relatively simple, portable calibration unit. The calibration could also be done with the instrument in place in the aircraft, to take into account any degradation of the window in the floor of the aircraft. On-board calibration lamps may be added to the AIS at a later stage of development. The spectral stability should not need regular calibration.

Electronic test signals may be introduced into the signal processing chain for testing the system from the CCD output to the data storage subsystem. This is an "aliveness" test which can be conducted at any time to check the instrument electronics and data recording system.

3. The Fore-Optics

A short focal length (24mm) wide FOV, 35mm camera lens is used as the foreoptics for the AIS. For the prototype instrument the only factors which were considered important were the focal length and the spectral transmission of the lens. The lens is mounted directly to the spectrometer housing. The spectrometer s mounted inside the spectrometer housing and rigidly attached to the lens mounting.

4. The Field-of-View Definition Subsystem

The entrance slit for the spectrometer is a laser etched rectangle in a reflectively coated glass disk. This provides a simple alternative to the traditional slit. The transmission and reflection properties of this device have yet to be carefully characterized.

5. The Dispersing Subsystem and Associated Optics

Several options were considered for the dispersing subsystem (spectrometer). One option was a simple prism. The disadvantages to this concept are the non-linear dispersion, and the size and weight of the optics. A linearized prism system was also considered, but this was thought to be too complicated for the prototype development phase. A dispersing system using a transmission diffraction grating was selected.

The spectrometer consists of a pair of triplet lenses as collimating and focusing lenses, a shortwave cutoff filter, a volume holographic transmission diffraction grating as the dispersing element, and a pair of prisms for alignment of the optical path. All of the components in the optical system have broad-band anti-reflective coatings.

The Cooke type triplet lenses are standard items from Melles-Griot. Future models of the AIS may require specialized coatings to enhance the spectral throughput of these lenses.

The shortwave cutoff filter is used to eliminate any possibility of interference from the second order of the diffraction grating's dispersion. This could be a problem if the instrument is operated at higher altitudes and if there is a lot of scattered light. The cutoff filter is a Hoya Y44 with a nominal cutoff wavelength of 440nm.

A ray trace analysis of the spectrometer was done using the Kidger lens design and analysis software. The spot sizes for the various off-axis locations and spectral dispersion (wavelength) locations were obtained. The geometric distortion of the system for various configurations was also checked.

In order to facilitate the alignment and adjustment of the prototype instrument, five micrometer adjusting mechanisms are employed. Three of them are used to adjust the position of the CCD. The fourth micrometer adjusts the slit angle with respect to the spectrograph, and the fifth adjusts the angle of the slit and spectrograph with respect to the CCD. Once the problems of alignment are studied with this system, it will be possible to devise a much simpler and permanent alignment system.

6. The Detector

The detector for the prototype AIS is a frame-transfer, television type, charge coupled device (CCD). It is the Thomson TH7863. It has 288 pixels in the spectral dispersion direction and 384 pixels in the spatial direction. The frame transfer mode of operation is ideal for an imaging spectrometer as it serves as an effective shutter. The TH7863 was selected because, in addition to the frame transfer feature and the proper frame rate, it is possible to obtain a complete set of drive electronics for it from Thomson.

7. The Detector Drive Electronics

As mentioned above, the basic detector drive electronics, consisting of a TH7996 sequencer and a TH7995 TTL/MOS interface, were obtained from Thomson. The sequencer is programmed using a FIFO structure which is operated by the microcontroller. The command sequence is selected by serial uplink from the PC controller.

8. The Preamplifier/Charge Converter

The low-level signal processing is done by a Thomson TH7992 CCD Signal Preprocessor. This device provides a low impedance output which goes via a coaxial cable to the A/D converter in the PC.

9. The Analog-to-Digital Converter

Analog signals from the front-end pass through a coaxial cable to the instrument controller. These are digitized by the EISA-A2000 personal computer add-on board made by National Instruments. This analog to digital (A/D) board has the capability to convert 1 megasample per second at 12-bits. To enable the data which is acquired at this rate to be processed by the PC and recorded, an EISA computer bus is required.

In addition to the A2000 board, an MC-MIO-16 data acquisition board, also from National Instruments, board is used to acquire auxiliary data inputs. Included in these inputs are airspeed, roll, pitch, yaw, GPS coordinates, etc.

10. The Experiment Controller

The PC is the experiment controller. Through a system of menus, the operational mode of the AIS can be selected. Once selected, the pre-programmed operation modes are downlinked to the front-end via the PC's serial port.

During operation, the controller monitors the operation of the AIS. Real-time images in any one of the selected spectral bands maybe displayed on the monitor.

11. The On-board Processor

The only on-board processing of the data which is done in the prototype system is formatting of the data, both from the CCD and from the auxiliary sources, before it is stored on tape. Some of the tape unit control software comes from the tape drive manufacturer, but a lot more has to be written to format the data and to control the drive so that it accepts the data smoothly. This is facilitated by a built-in buffer in the tape drive.

12. The Data Storage and Compression Subsystem

The data storage device for the AIS is an Exabyte 8500 streaming tape system. For the prototype device, no data compression is used. Depending on the method for formatting the data, it is likely that a considerable amount of data compression will be possible, due to the highly redundant nature of the spectrometer scenes.

A second analog-to-digital converter board in the PC is used to acquire auxiliary data. This data is sampled once per CCD frame, formatted, and included in the records which are transmitted to the streaming tape unit. Some of the data which is critical for interpretation of the image data, such as roll, pitch, and yaw, are included in these records.

13. Ground Data System

Any image processing system which can accept data on the standard 8mm Exabyte digital tape cartridges can be used to process the data from the AIS. The Remote Sensing Section of the Instrument Laboratory is working on developing software for this part of the project.

Instrument Operations

The prototype instrument operates in four modes. These are called: 1) the full-frame mode, 2) the full spatial, reduced spectral mode, 3) the full spectral, reduced spatial mode, and 4) the variable spectral, variable spatial mode.

In the first mode, full spectral and spatial information is available. This means 384 pixels in the spatial direction and 288 pixels of spectral information. As the frame rate in this mode is rather low (approximately 8 frames per second maximum) this mode is primarily used for calibration and ground based experimentation. It could also be used for very high altitude (low spatial resolution) or low speed (helicopter) operation.

The second mode, and most common mode of operation, acquires full spatial information (384 pixels) but is limited to a maximum of 16 spectral channels. This limit is set by the aircraft speed (approximately 55 meters/second) and by the spatial resolution (1 meter). These spectral channels can be selected from anywhere within the spectral range of the instrument, which is 450 to 900 nanometers in the prototype. Each channel can have a bandwidth of up to 6 pixels, or about 10 nanometers. The summing of the pixels to give the bandwidth of up to 6 pixels is also limited by the on-chip summing capability of the TH7863 CCD. This capability is approximately 1.1 full-well pixel capacity, so highly reflective scenes may saturate if summed to the maximum.

The third mode acquires full spectral information at a limited number of "look directions" or spatial stripes. The prototype system is limited to a minimum stripe spacing of 7 pixels. This is due to the timing capability of the pixel clock. There is also the possibility to sum spatial pixels, but the TH7863 chip only allow one full-well pixel to be summed on-chip. The on-chip summing capability in both the spectral and spatial directions is a feature which needs to be enhanced in future instruments.

The fourth mode allows the user to specify both the spectral and the spatial resolutions. The only limiting factor is the detector integration time. This mode is only suitable when the acquisition parameters are exactly known as it reduces the amount of irrelevant data that will be acquired.

Instrument Control

The interface between the instrument and the host computer is kept as simple as possible. It consists of three signals: 1) the CCD analog signal, 2) the trigger signal from the CCD for pixel conversion timing, and 3) the serial port for uploading of the instrument control parameters.

The host computer is responsible for counting the pixels and sorting out the lines of spectral and spatial information. This is easily coordinated as the operation mode of the CCD is determined by the PC, so the sequence of the output signals is well known.

Performance Analysis

The performance analysis for the system is done using a spreadsheet which was specifically created for the AIS. The spreadsheet is based on similar performance analyses for spaceborne imaging spectrometers, also using spreadsheets.

Performance Analysis Spreadsheet

The performance analysis spreadsheet attempts to thoroughly model the performance of the airborne imaging spectrometer. The spreadsheet is modeled after the MODIS-T performance analysis spreadsheet, which included an atmospheric transmission simulation. As this is not important for a low-flying AIS, the reflected light values in the AIS spreadsheet were taken from tables generated by a PC version of LOWTRAN 7, called PCTRAN 7.

Instrument Enhancements

Swath Width: The prototype AIS is the simplest, functional instrument that could be made and still incorporate most of the features of a fully functional airborne imaging spectrometer. In the next development stage several improvements will be made. The most important of these is increasing the swath width. This improvement will be accomplished by using a CCD with more than 1000 pixels in the spatial direction, a nearly threefold increase in spatial coverage over the prototype.

Image Stabilization: Another important enhancement is the addition of an image stabilization subsystem. The simple, brute-force, solution is to mount the compact instrument front-end on a small stabilized platform. The problem with this solution is that the advantage of compact size will be lost. It will no longer be possible to mount the AIS in a small hole in the floor of a small aircraft. One solution to this problem is to acquire the roll, pitch, and yaw data for the aircraft and record it with the AIS data stream. This information can then be used in the processing of the data to restore the registration. The disadvantage to this technique is that, primarily due to aircraft roll, some of the data across the swath may be lost and the full field-of-view cannot be used.
Another solution to this problem is to separate the front-end optics and FOV definition subsystem from the spectrometer and link them with a coherent fiber optics bundle. This very small subsystem may then be mounted in a image stabilization mechanism. The spectrometer and front-end electronics may then be mounted some distance (the length of the fiber optic coupling) from the aperture in the aircraft floor. Another application of the fiber optic coupling is discussed below.

Infrared Focal Plane: Another important enhancement of the instrument is the addition of a second focal plane for the extension of spectral coverage into the near infrared (NIR). The idea to add focal planes by having multiple spectrometer systems fed by fiber optic couplings to the focal plane of the front-end optics is considered to be a better solution than having multiple detectors in the same spectrometer. Multiple spectrometers will allow the subsystems to be optimized for each wavelength range.

On-chip Summing Capability: An enhanced on-chip summing capability is a feature which needs to be enhanced in future instruments. This will allow the summation of more spectral bands, and combination of spatial resolution elements. The more advanced CCDs which have more spatial pixels usually have better on-chip summing capabilities.

Conclusions

The tangible result of my Research and Study Fellowship in Finland is the completion of the prototype airborne imaging spectrometer. The more ambitious goals, to develop cooperation between Finnish institutions doing space research and NASA/Goddard, and to help to develop the capability to build space qualified instrumentation in Finland, have yet to be accomplished.

References

1) Report on the Status of the Space Business in Finland, John Bolton, July 1992
2) Proposal for the Development of an Airborne Imaging Spectrometer, John Bolton, October 1991
3) Plan for the Continuation of the Development of the AIS, John Bolton, June 1992
4) An Airborne Imaging Spectrometer for Forest and Pollution Measurements, John Bolton, Bart Braam, and Jukka Okkonen, May 1992

Finnish Staff for the Prototype AIS
Aikio, Mauri Optics, Researcher (Oulu)
Braam, Bart Applications, Researcher (Karelsilva OY)
Haapalainen, Ahti Optomechanics, Technician (Oulu)
Mäkisara, Kai Data System, VTT Management (Tapiola)
Meinander, Marko Data System, Student (Tapiola)
Okkonen, Jukka Electronics, VTT Management (Oulu)
Pylkkö, Pentti Applications, General Management (Karelsilva OY)
Rantasuo, Markku General Systems, Technician (Tapiola)
Sipola, Kaarlo CCD control electronics, Technician (Oulu)



Original: 18 April 1995


New Systems And Technologies For Optical Remote Sensing


This paper will describe a concept for remote sensing that has existed for some time, but has not yet been thoroughly exploited. This concept is hyperspectral imaging. It is now possible because of advances in technology. These technological advances are not, however, enough to build a complete hyperspectral remote sensing system. The new technologies and the associated systems required for their development and exploitation will be described. The primary problems associated with hyperspectral imaging will be addressed.

 

What is Hyperspectral Imaging?

The traditional spectral measurements in the laboratory have typically been made with instruments that scan the spectral range of interest. These full spectrum measurements have been the source of fundamental information regarding most organic and inorganic materials. Early remote sensing instrumentation did not have the capability to make continuous spectral measurements. Limitations in optics, detectors, and data storage and transmission forced researchers to compromise and to select a few spectral bands. The present state-of-the-art in remote sensing is based on the selection and manipulation of these bands.

This band selection approach in remote sensing has persisted until the present. It is no longer forced by technology, but is an artifact of the original technology limitations for remote sensing. Technology developed over the past few years has allowed us to acquire nearly continuous spectral data. This is hyperspectral imaging. The primary problem that we are now faced with is the selection of the spectral and spatial resolution and the trade with the amount of data that we wish to handle.

Why use Hyperspectral Imaging?

As mentioned in the last section, hyperspectral imaging provides nearly continuous spectral coverage. In addition, the design of hyperspectral instruments is simpler than that of the traditional multi-spectral band instruments. A hyperspectral imaging instrument is an imaging spectrometer. The most efficient implementation of an imaging spectrometer requires operation in the "pushbroom" mode. This has several advantages over the traditional multi-band scanner, but there are also some problems that must be addressed. The imaging spectrometer technology can be applied to all "optical" instruments covering the spectral bands from the blue (400nm) to the thermal infrared (>15µ). This technology has been successfully demonstrated in several projects for different applications.

What Problems are Associated with Hyperspectral Imaging?

One problem associated with hyperspectral imaging, as mentioned above, is the high data rate. There are several ways to address this problem in addition to the brute force application of recently developed technology. The simplest option is to select only the bands that have traditionally been used. This is the way in which the LANDSAT data sets can be preserved with the new instrumentation. Many options for reducing the data rate while preserving the full spectral data set have been proposed but it is not appropriate to go into them here.

Another problem with hyperspectral imaging is calibration of the sensors. The traditional multi-spectral scanners are relatively easy to calibrate using on-board radiometric and spectral sources. The small number of sensor elements and the rotating scan mirror provide easy access for all detector elements to the calibration sources. The pushbroom type imaging spectrometer cannot practically be calibrated this way. Uniform illumination across the entire field-of-view by the on-board sources would be required. This is both impractical and unreliable.

Solution to the Calibration Problem

The straightforward solution to the calibration problem is to eliminate the on-board calibration capability for all of the pushbroom sensors. This seemingly radical approach opens the way for the development of a remote sensing system that can provide many collateral advantages. This approach encourages calibration of the sensors by reference to ground sites (a technique called vicarious calibration). This requires that the suite of instruments on the spacecraft have the capability for the transfer of calibration from ground site to ground site, and from instrument to instrument. This, in turn, requires an accessory instrument, a transfer radiometer, designed for the transfer of calibration. Other targets such as the Moon and dark space, and some on-board sources, may also be used. In addition to this calibration transfer function, the instrument can serve other purposes as it would be a pointable, high performance, high spatial resolution, imager.

Spin-Offs of the Solution to the Calibration Problem

 


Original: 26 April 1995



A Rough Draft Of A Preliminary Conceptual Design For A High Spatial Resolution Hyperspectral Imager (HSRHI)


John Bolton Systems Engineering Office
Code 704 NASA/Goddard Space Flight Center
Greenbelt, MD 20771 U.S.A.


personal telephone: (301) 286-8547
secretary telephone: (301) 286-2269 (Michelle Loyd) or (301) 286-8279 (Cheryl Carr-Huggins)
office telefax: (301) 286-1653
e-mail: "john.bolton@gsfc.nasa.gov" or n2jfb@vax720.gsfc.nasa.gov


Introduction

This is a preliminary draft of a conceptual design for a high spatial resolution hyperspectral imager (HSRHI). This is a very rough first draft and cannot proceed until science requirements and instrument constraints are further defined. The key science drivers and the key instrument features will be described. The key features are selected from many options that have been considered in other instrument designs. A complete trade study will have to be done eventually, but this paper will serve to highlight some of the features and the technologies that may be used.

Innovative approaches to meeting the requirements will be described. Technologies that have been proved, but not necessarily applied to remote sensing instrumentation, will be used. The technologies will be kept as simple as possible. This simplicity will both reduce cost and risk, and will help to assure high quality measurements. The most significant performance characteristic of the instrument is its capability for continuous spectral coverage. Through uploadable commands, the instrument can be programmed to observe any combination of spectral and spatial regimes. Alternatively, the instrument can be operated in the full data acquisition mode (full spatial and full spectral resolution) and the output data can be buffered, sorted, selected, and downloaded by the spacecraft processor. Some of the concepts described in this paper were implemented and proved during the development of the AISA airborne imaging spectrometer, during my Goddard Research and Study Fellowship in Finland during 1991, -92.

As the instrument operates in the pushbroom mode, it will have no continuously operating scan mechanisms. The only mechanisms needed will be those to deploy the instrument on-orbit, such as aperture and radiator covers. A mechanism will be used to point the instrument (actually only the fore-optics part of the instrument) along and cross-track. This mechanism will serve an important role in the data collection process, as described below.

There will be no on-board, internal, calibration sources. A ratioing radiometer, or equivalent, may be included if desired. One function of the instrument will be to transfer measurements from ground-based standards to satellite-borne instruments. The instrument will be calibrated by ground truth observations and by prepared ground sites. Other ground- and space-based calibration techniques will be considered. The calibration via ground sites is discussed in Appendix 1 of this paper.

The basic instrument characteristics can be suggested without full knowledge of the science requirements. The general science requirements are fairly well understood from work that has been done with the HIRIS instrument and follow-on proposals, for example. The basic idea for the HSRHI is to incorporate innovative ideas (not necessarily state-of-the-art) to simplify the instrument, improve its performance, and to make it more versatile.

Performance Requirements

The preliminary, rough, performance requirements are as follows:

Parameter

Requirement

Swath width

20 Km

Spatial resolution

10 m

Spectral range

0.4 to 1.0 microns (0.9 to 2.5 µ option)

Spectral Resolution

256 (approx. 2.3 nm for 0.4 to 1.0 µ)

Pointing

± 40 ° cross-track, +40° : -40° along-track

Radiometric accuracy *

± 3%

* The radiometric accuracy is the accuracy that must be maintained between the observation of a calibration site or source, and the transfer of the measurement (cross-calibration), or collection of scientific data at another site. See Appendix 1 for further discussion.

Instrument description

To describe the design of the instrument in an orderly manner, it must be broken down into subsystems. When the requirements are known better, these subsystems can be optimized individually and then the entire instrument performance can be optimized using and end-to-end performance analysis. This same procedure was used for the conceptual design of the MODIS-T instrument. Many of the tools and concepts developed for the MODIS-T work are directly applicable to this study. The following are the generic instrument subsystems. Several of them do not apply to the HSRHI instrument, as will be explained below.

1. The Scan Sub-system (for the HSRHI this is The Pointing Sub-system)
2. The Calibration Sub-system
3.
The Fore-Optics
4.
The Field-of-View Definition Sub-system
5.
The Dispersing Sub-system and Associated Optics
6.
The Detector
7.
The Detector Drive Electronics
8.
The Detector Readout Electronics
9.
The Analog-to-Digital Converter
10.
The Experiment Controller *
11.
The On-board Processor *
12.
The Data Compression Sub-system *
13.
The Structure and Mechanisms

* do not apply to the HSRHI instrument

1. The Pointing Sub-system
The HSRHI instrument will use pushbroom technology and will therefore have no continuously operating scan subsystem. The instrument will, however, have a pointing capability requiring a pointing mechanism. As the high spatial resolution will require large fore-optics, it will not be possible to simply point the fore-optics as in other, smaller, hyperspectral imagers. A pointable mirror ahead of the fore-optics will be required. Image rotation due to cross-track pointing can be eliminated by innovative use of the field-of-view definition sub-system.
This implies that the HSRHI will also need a system to determine in which direction it is pointed. There are several options for this function, including a simple spacecraft-based position encoder, a star tracker, and an image correlation tracker. The image correlation tracker is an important potential technology for the HSRHI. The large number of spectral bands and high spatial resolution translate to an extremely high data rate, unless an innovative approach is used. The image correlation tracker is a part of this technique.
If the HSRHI acquires data at the ground track velocity of approximately 7 kilometers per second, the demands on the detectors and data conversion systems will be excessive. To decrease this rate, the HSRHI instrument can be "back-scanned" as it is passing over the scene to provide longer dwell times and lower data rates. The best way to control the back-scanning is with an image correlation tracker.
The image correlation tracker is a simple imager that acquires sequential frames and compares their displacements. By determining the displacement of sequential images, the correlation tracker can be used to control the pointing of the acquisition system. The implementation of the correlation tracker is fairly straightforward and many systems have been built.

2. The Calibration Sub-system
The HSRHI instrument need not have a traditional calibration sub-system. A ratioing radiometer or equivalent system may be provided for absolute, full field, calibration with respect to the Sun. Internal electronic checking will be available. Through platform rotation and instrument pointing, it will be possible to view the Moon and dark space. The primary instrument calibration will be accomplished by viewing ground sites that have been fully characterized by either ground or airborne measurements (see Appendix 1 below). This provides the database needed for cross-calibration of the other instruments, as well as the calibration needed for high spatial resolution scientific investigations.

3. The Fore-Optics
The fore-optics, together with the field-of-view definition sub-system, form a critical part of the instrument. The HSRHI field-of-view is not particularly wide, so the high spatial resolution, high spectral throughput, fore optics will not be difficult to design and manufacture. Due to the high spatial resolution requirement, the fore-optics must have a rather large aperture, to provide adequate resolution. The fore-optics in combination with the field-of-view definition sub-system is the first innovative part of the HSRHI instrument.
If a SWIR capability is added to the HSRHI, the spatial resolution requirement should be reduced. In order to achieve the same high spatial resolution as in the VIS-NIR, an exceptionally large aperture fore-optics would be needed.

4. The Field-of-View Definition Sub-system
A key innovative part of the HSRHI is the field-of-view (FOV) definition subsystem. Typically, this is a slit aperture leading to the remainder of the optical train. In this instrument this sub-system would be a flexible, imaging, fiber-optic link. The fiber-optic link has many advantages. These include the possibility to mechanically separate the fore optics from the rest of the instrument. By splitting the fore-optics and FOV system from the instrument, the size of the instrument that has to be articulated to point across and along-track can significantly reduced. In the case of the HSRHI, where the fore-optics are not articulated due to the large size, the flexible, imaging, fiber-optic link provides other advantages.
Using a mirror ahead of the fore-optics, to direct the field-of-view of the instrument, results in image rotation when pointed cross-track unless a correction is made. This correction is relatively simple when the flexible, imaging, fiber-optic link is used. The input aperture of the fiber-optic link can be rotated synchronously with the mirror.
Another advantage provided by the fiber-optics link is to split the entrance aperture to several focal planes. This technique can be used to divide the field-of-view and/or to divide the spectral coverage. If the field-of-view is divided, the data acquisition task can be accomplished by smaller, faster, and presumably "better" CCDs. If the spectral coverage is divided, the spectrometer and the CCDs can be optimized for the specific range of wavelengths. Though the light intensity will be split among the focal planes, the registration of the images will be exact. There are other variations on the concepts mentioned above, and there are many more instrument options involving the fiber-optic technology that could be employed, but they will not be described here.

5. The Dispersing Sub-system and Associated Optics
The demands on the dispersing sub-system are rather stringent as it must provide good image quality and spectral purity, as well as help to correct the spectral response non-uniformity of the rest of the instrument sub-systems. In order to optimize the throughput of the instrument, it might be necessary to use more than one dispersing sub-system. This is mentioned in the previous and in the following sections. The exact implementation of this concept, in order to maintain the best image quality and registration, is yet to be determined. This is not expected to be a difficult task.

6. The Detector
For the VIS-NIR portion of the spectrum (0.4 to 1.0 microns) a self-scanned, silicon, CCD array would be used. To minimize image smear, a frame-transfer or equivalent device would be appropriate. Several types of this kind of detector are commercially available. Custom devices are not extraordinarily expensive. Proof-of-concept instruments should use off-the-shelf commercial parts, particularly the CCDs.
In order to optimize the spectral response of the system, it might be necessary to split the spectral range into two parts. This is particularly useful as the spectral range of 0.4 to 1.0 microns covers more than one octave, resulting in order overlap in a grating-based dispersing system. To cover the VIS-NIR spectrum, for example, two detectors could be used, one blue enhanced detector from 400 to 600 - 700 nm, and one red-enhanced detector from 600 - 700 to 1000 nm. This implies, of course, that a second dispersing sub-system will also be needed. The flexible, imaging, fiber-optic link would be bifurcated to split the entrance aperture to the two focal planes.
If the option to cover the spectral range from 1.0 to 2.5 microns SWIR bands is considered, a completely different detector technology (probably HgCdTe (MCT)) would be required. In this spectral range, it would certainly be necessary to split the spectrum into several regions for optimum detection efficiency, particularly as the various regions would require optimized detectors with different cooling requirements. The techniques described above for data reduction in the VIS-NIR regime would also apply in the SWIR.

7. The Detector Drive Electronics
The versatility of the instrument depends on the programmability of the detector readout, which is controlled by the detector drive electronics. This can be a rather simple sub-system, consisting primarily of a FIFO that can be loaded with the appropriate sequence of clock pulses and then clocked out at the appropriate rate. Experience with the airborne imaging spectrometer described in Appendix 1 proves that this can be done relatively easily and reliably. The instrument's operating mode can be changed practically instantaneously.
As it is so simple to have selectable operating modes for the CCD readout, it is a good idea to incorporate this feature, as many users of the HSRHI will prefer to operate it with selected bands. This mode of operation significantly reduces the data rate, which is another reason to have the on-board readout programmability option.

8. The Detector Readout Electronics
The signal-to-noise ratio of the instrument is primarily determined by the quality of the CCD device and the detector readout electronics, and the speed at which the device is read out. The readout technology is well understood and will be relatively easy to implement. The performance of the CCD is another matter and depends to a large extent on the device size (which drives the readout rate) and fabrication technology. Dividing the field-of-view into several focal planes will alleviate this problem as smaller, lower readout rate, higher performance (particularly regarding signal-to-noise-ratio) CCDs can be used.

9. The Analog-to-Digital Converter
The potentially very high data rate of the HSRHI instrument will place extreme demands on the analog-to-digital (A/D) converter(s). A linear 12-bit A/D converter will probably be appropriate, though some sort of non-linear A/D conversion, such as square-root conversion, might be preferable. These options are well documented in other papers. The table below, "Table of Data Acquisition Parameters", summarizes the data conversion situation.

10. The Experiment Controller
This sub-system is outside the scope of the HSRHI instrument. The function will be provided by the spacecraft-provided controller.

11. The On-board Processor
This sub-system is outside the scope of the HSRHI instrument. The function will be provided by the spacecraft-provided controller.

12. The Data Compression Sub-system
This sub-system is outside the scope of the HSRHI instrument. The function will be provided by the spacecraft-provided controller.

13. The Structure and Mechanisms
As the instrument front-end (pointing mirror and fore-optics) is essentially independent of the rest of the instrument, (assuming a flexible, imaging, fiber-optic link) the alignment requirements are significantly reduced. The dispersing sub-system(s) and associated optics, with the detector assembly attached, are compact assembly(s), independent of the front-end regarding alignment. This means that the demands on the structure are reduced and that a lighter, less expensive structure can be used. In addition, alignment and testing of the sub-systems is much easier as their performance can easily be assessed independently.
As mentioned above, the primary instrument mechanisms are for mirror pointing. The only other internal instrument mechanisms are those needed to deploy radiator shields and, possibly, an aperture cover. If the bands from 1 to 2.5 µ are covered, the cooling and radiator issues become more significant. If not, the cooling is not an issue.
The mechanism to tilt the instrument fore-optics along- and cross-track can be a simple flex-pivot assembly.

Instrument Mass, Power and Size

NOTE: The values in the following Table are all rough estimates and further analysis is needed to make accurate estimates.

Table of Instrument Mass, Power and Size

(assuming the focal plane is divided into 8 parts)

Subsystem

Power (Watts)

Mass (Kilograms)

Dimensions (meters)

The Pointing Sub-system

20 (intermittent)

10

0.6 x 0.6 x 0.4

The Fore-Optics

0

3

N/A

The Field-of-View Definition Sub-system

0

3

 

The Dispersing Sub-system and Associated Optics

0

1 x 8

(0.1 x 0.1 x 0.2) x 8

The Detector

0.5

1 x 8

(0.1 x 0.1 x 0.05) x 8

The Detector Drive Electronics

.05

0.025 x 8

1 euro card x 8

The Detector Readout Electronics

2.5

0.025 x 8

1 euro card x 8

The Structure and Mechanisms

0

20

0.7 x 0.7 x 0.5


Table of Innovative Concepts applied in the HSRHI

Concept
Implementation
Hyperspectral imaging
Dispersive optical system, 2-D readout array
Spectral band selectability
Dispersive optical system, 2-D readout array
High radiometric accuracy
Ground truth calibration
Fore-optics decoupling
Flexible, imaging, fiber-optic link
Multiple focal planes
Flexible, imaging, fiber-optic link
Backscanning
Image correlation tracker
Image stabilization
Image correlation tracker

Appendix 1. On the Utility of Ground Reference Sites and Their Calibration with an Airborne Hyperspectral Imager

NOTE: The comments in this Appendix are a rough draft and should only be "taken internally". They will have to be made more "politically correct" if we want to promote this idea along with the concept for the hyperspectral imager.
Introduction

This Appendix discusses the use of accurately characterized ground sites as calibration targets for the HSRHI instrument. The technique for characterizing the sites and the side benefits of the establishment of the sites is discussed.

This discussion is based on the work that I initiated during my Goddard Research and study Fellow Fellowship at the Technical Research Center of Finland (VTT). The goal of the work was (among other things) to develop a system for accurate characterization of ground sites for the underflight calibration of spaceborne sensors. This work received strong "moral" support from many people and organizations, including most of the major U.S. Government agencies (NOAA, USGS, USDA, etc.). Unfortunately, due to severe budgetary limitations, this work was not carried forward and now is being continued primarily by my colleagues at the Finnish Forest Research Institute and at the Finnish Geological Survey.

The key to the site calibration is a high-performance, compact, and economical instrument. This instrument, which has been called "AISA" (Airborne Imaging Spectrometer for various Applications) was built at VTT as a result of my organization of the Project in Finland. Its design is virtually identical to the design that I originally proposed to the VTT staff. The AISA instrument is a very similar to the VIS-NIR HSRHI instrument described in this paper, but somewhat simpler and built almost entirely of off-the-shelf commercial parts. It is simple to use and affordable by any small remote sensing laboratory. As it is fully programmable, it can produce data sets of any sets of spectral bands in the VIS-NIR range, or it can produce full hyperspectral data.

The establishment of ground reference sites would have a side benefit. As the AISA instrument is a true hyperspectral imager (minimum spectral bandwidth approximately 1.5 nm) it can be used to produce experimental hyperspectral data sets. The affordability of the instrument make it possible for small remote sensing laboratories to get one of these instruments for themselves and use it to measure their own sites. This will provide excellent reference information for the development of algorithms for the reduction of hyperspectral data. This will encourage the use of remote sensing data, it will develop expertise in remote sensing in general, and it will provide the ground reference sites.

These concepts are discussed further in the paper entitled, "New Systems and Technologies for Optical Remote Sensing".

Table 1. Data Acquisition Parameters and Rates

(With no data reduction strategies applied)

INPUT PARAMETER

VALUES

UNITS

NOTES

Satellite Height

705

Kilometers

 

Ground Resolution

5

Meters

projection of a pixel on the ground

Pixel Length

25

Microns

pixel size

Pixel Width

25

Microns

pixel size

Pixels per Spectral Band

1

N/A

(used if pixels are summed)

Pixels per Swath

4096

N/A

pixels in swath direction

Spectral Bands

256

N/A

pixels in spectral direction

Swath Length

20

Kilometers

length of a "snapshot"

Number of Bits Encoded

12

Bits

A/D converter quantization

Number of CCDs

1

N/A

Number of CCDs

Backscan Ratio

1

N/A

Scene dwell factor mult by scanning

A/D Converter Speed

20,000,000

Samp/sec

A/C converter sample rate

 

OUTPUT

VALUES

UNITS

NOTES

Detector Area

6.25E-04

mm square

pixel size

Ssubsat. Point Velocity

6.76E+00

Km/second

Ground track speed

Orbit Period

98.86

Minutes

 

Frame Period

7.40E-04

Seconds

time for one spatial resolution element

Pixel Readout Time

1.81E-07

Seconds

time to read one pixel

Swath Width

2.05E+01

Kilometers

 

Swath Angle

1.67E+00

Degrees

 

Pixels per CCD

4096

N/A

pixels in one CCD

Sample Rate / CCD

1.35E+03

Samp/sec

sample frames per second / CCD

Bits / Sample

1.26E+07

Bits

number of bits / sample / CCD

Pixel Rate / CCD

1.42E+09

Pix/sec

pixel readout rate from each CCD

A/D Converter Rate

1.70E+10

Bits/sec

data rate / A/D / CCD

Num. A/D Required / CCD

71

N/A

number of A/Ds required / CCD

Time per Swath Length

2.9601

Seconds

time to cover swath length

Total Data (Bits)

5.03E+10

Bits

total data per swath length

Total Data (Gigabytes)

6.29

Gigabytes

total data per swath length

Table 2. Data Acquisition Parameters and Rates

(With all data reduction strategies applied)

INPUT PARAMETER

VALUES

UNITS

NOTES

Satellite Height

704

Kilometers

 

Ground Resolution

5

Meters

projection of a pixel on the ground

Pixel Length

25

Microns

pixel size

Pixel Width

25

Microns

pixel size

Pixels per Spectral Band

1

N/A

(used if pixels are summed)

Pixels per Swath

4096

N/A

pixels in swath direction

Spectral Bands

256

N/A

pixels in spectral direction

Swath Length

20

Kilometers

length of a "snapshot"

Number of Bits Encoded

12

Bits

A/D converter quantization

Number of CCDs

8

N/A

Number of CCDs

Backscan Ratio

10

N/A

Scene dwell factor mult by scanning

A/D Converter Speed

20,000,000

Samp/sec

A/C converter sample rate


Original: 27 June 1995

Strawman Design For A Hyperspectral Transfer Radiometer And Follow-Ons

One of the key elements in the suite of follow-on LANDSAT/EOS (NOAA/NASA) instruments will be the hyperspectral transfer radiometer (HTR). If we are to achieve dramatic cost reductions through the simplification of the instruments, one feature that can be deleted is the on-board calibration capability for each instrument. The HTR has been proposed to replace this function.

This method of calibration involves an entirely new way of doing the remote sensing business and will be one of the most important issues to be addressed in promoting the new instrument technology. We will, therefore, have to demonstrate a whole series of capabilities before the new technology will be accepted. These capabilities include:

* duplication of the present LANDSAT/EOS bands
* equaling or exceeding the present LANDSAT/EOS signal-to-noise ratios
* improved radiometric accuracy and stability
* ability to replicate existing data sets (combination of the above points)
* procedure for reduction of the hyperspectral data
* strategy for transferring calibrations to accompanying instruments
* technique for handling large hyperspectral data volume
* a suite of instruments with higher performance and lower cost than the existing options

To properly demonstrate these capabilities, we will first of all need a strawman design for the HTR itself. The specifications for the HTR are listed in Table 1. It is very important that this instrument be relatively simple, compact, and reliable. Even though its primary function will be calibration, it will have many scientific uses by virtue of its high spectral and spatial resolution, and its ability to point. This means that it will probably be used continuously. The design of the HTR should not be concerned with the handling of the large volume of data, other than getting it out of the instrument and onto a data buss.

Once a feasible strawman design for the HTR has been presented, we will have to continue to demonstrate the capabilities of the new remote sensing system. Of primary concern is the continuation of the LANDSAT data sets. We must, as a minimum, demonstrate that we can replicate this data. To do this with hyperspectral, pushbroom instruments will be challenging. It will not be possible to demonstrate this capability by analysis alone.

One of the most serious objections to the hyperspectral implementation of the LANDSAT bands is the exact center band wavelengths and bandwidths. Even though there is considerable tolerance (in practice if not in principle) in the LANDSAT spectral bands, it will be necessary to demonstrate using real data, that the new instrumentation provides products equivalent to those acquired with the old instrumentation. This can only be done with data acquired by real hyperspectral instruments flown on aircraft, or in experimental flights on the Shuttle.

As the replacement of the existing LANDSAT and EOS instruments with something completely different is such a radical move, it is expected that extensive demonstrations of the new technology will be required. These demonstrations can serve to do much more than simply show that that the new technology will adequately replace the old. An R&D program to develop the new technologies will pay off if it is properly planned to produce more than just the replacement demonstration. Possibilities for spin-off demonstrations during the R&D program have been suggested and new suggestions are always welcome.

NASA is seeking to form a partnership with industry in the development of the follow-on EOS/LANDSAT instruments. The incremental development of the hyperspectral technology, from simple laboratory and airborne instruments to the full line of space-borne instruments, will minimize the development risk and provide spin-off opportunities with commercial applications along the way.

NOTE: The first step in this process to demonstrate the new technology is to come up with the strawman design for the HTR and a strategy for cross-calibration of the accompanying instruments. Once we are comfortable with the plan for this key technology, it will be possible to move on to the next steps.

TABLE 1.

The preliminary, rough, performance requirements are as follows:

Parameter

Requirement

Notes

Swath width

20 km

 

Swath length

20 km (data acquisition period)

 

Spatial resolution

10 m

 

Spectral range

0.4 to 1.0 microns (0.9 to 2.5µ option)

 

Spectral Resolution

256 (approx. 2.3 nm for 0.4 to 1.0µ)

 

Pointing

±40° cross-track,
+40°:-40° along- track

 

Radiometric accuracy *

± 3%

 

* The radiometric accuracy is the accuracy that must be maintained between the observation of a calibration site or source, and the transfer of the measurement (cross-calibration), or collection of scientific data at another site.



Original: 19 July 1995

Plans For An Extended And Operational Airborne Remote Sensing Program


Note: This is the rough draft of a paper to present the rationale for an extensive, high performance, airborne remote sensing program. The purpose of this paper is to show that the benefits of such a program are very cost effective and would contribute significantly to Earth remote sensing science. The program would also be an integral part of the development of the EOS/LANDSAT follow-on missions.

Introduction

For NASA, airborne remote sensing has always been a minor activity. This is understandable as our activities are primarily space oriented. In order to increase the level of scientific activity with ever decreasing budgets, we are going to have to look at more effective ways to do remote sensing. Airborne operations are an integral part of the solution to this problem.

There are many advantageous features of airborne remote sensing. This paper will seek to list them and to tie them in to the general NASA programs for Earth remote sensing. The benefits and disadvantages (compared to space based activities) will be discussed.

General Plan

The general plan is to develop and coordinate a worldwide network of airborne remote sensing activities. It is important to point out immediately that neither NASA nor the U.S. Government would bear the primary financial burden for these activities. NASA, and the GSFC in particular, would assume the role of coordinator, partner, and advisor. Most of the work would be done by the private sector. Even though some of the funding may come from the public sector, it is expected that the technology and applications development, and the operations of the airborne systems will be done by the private sector in partnership with an airborne remote sensing program at the GSFC. This GSFC airborne program would be closely associated with, and complementary to, the space based remote sensing programs at GSFC.

Specific Features

Development of Remote Sensing Science

The first and most important feature of the airborne remote sensing program would be the development of remote sensing science, applications, and talent in general. The user community ranges from the "top-of-the-line" activities such as those done here at GSFC and at other institutions with specialized airborne remote sensing equipment, to the general "man-on-the-street" activities of small remote sensing laboratories, local governments, and private businesses. The "top-of-the-line" work requires state-of-the-art instrumentation with many special features. The "man-on-the-street" work requires general-purpose, but high quality instrumentation. We must pay primary attention to the "man-on-the-street" as this is where the bulk of the activities will be and where most of the funding will come from. It will also be the source of new talent. The possibility for researchers to have their own remote sensing system, and to fly it over their own sites, and to work with their own data, will provide a great incentive to do new and creative work, and for more people to get into the business.

Development and Commercialization of Airborne Remote Sensing

As airborne remote sensing technology becomes more accessible, more applications will be developed and businesses established. Recent developments in sensor, computer, and recorder technology will bring down the cost of the instrumentation. Once low cost, high performance instrumentation is available, applications will follow which will in turn, stimulate new developments in instrumentation.

In order to encourage the development and commercialization of airborne remote sensing, the GSFC should establish a Center for Airborne Remote Sensing at the Wallops Flight Facility (WFF), in addition to the airborne remote sensing program at the GSFC. The advantages and features of the WFF have been described elsewhere (ref.). This Center should be available to all users of airborne remote sensing, both public and private. The interaction between GSFC and WFF staff and contractors, and the world airborne remote sensing community, will serve as a focal point and proving ground for new technologies and applications. There would, of course, be extensive cooperation with the existing programs for the commercialization of remote sensing at the Stennis Space Center.

Incremental Development of Technology and Applications

The private and commercial sectors are always those who are most responsive to new technological developments. A lot of work has already been done in the development of the technology for remote sensing instrumentation. The technology used in airborne instruments can be used as a basis for the design of space borne instruments. The data collection and processing techniques may also be applied to space borne systems. We are planning to develop future space borne sensors starting with technology and applications that are developed and proven in airborne systems. This is an essential part of the plan to develop the EOS/LANDSAT follow-on missions and has been described in detail elsewhere (ref.).

Supplemental Science

Even though there is a clear distinction in capabilities between the space borne remote sensing capabilities and the airborne, they are both needed to do complete remote sensing science. The space borne systems can give the global picture. The airborne systems will provide the details and help to investigate the mechanisms that influence the global phenomena.

It is often impossible (and sometimes possible but not allowed) to obtain the level of detail with space based systems as can be obtained with airborne systems. Many remote sensing applications require high resolution studies. High spatial and spectral resolution is required. In some cases very good elevation information is also needed. This can be done relatively easily, and in most cases, only with airborne systems.

Vicarious Calibration and Synergistic Science

A worldwide network of sites that are accurately characterized will serve as reference points for space borne sensor calibration. As these sites will typically be spectrally and spatially complex, new vicarious calibration techniques will have to be developed. This development work would be closely connected with studies to understand the details of the remotely observed phenomena.

The Center for Airborne Remote Sensing at the Wallops Flight Facility would serve to develop, establish, and certify techniques for the accurate characterization of these sites. The sites, many of which will have been established for purposes other than the vicarious calibration of space borne sensors (thus saving considerable cost to NASA), will have to be certified before data from them can be accepted into the EOS/MTPE database.

Conclusions

An extensive program for airborne remote sensing should be planned and established at the GSFC and the WFF. This will serve to both augment the Earth Science programs, and to decrease the cost and risk of the space borne segment.



And Now For Something Completely Different

A Proposal for an Alternative Method for the Development of Earth Observation Science and Technology


Cover Message

The following is the initial draft of a paper that I have written to open the discussion of an alternative way of doing NASA’s Earth Observing Science. It is based on my experience working for the EOS Project, and on some years of instrument development experience. It is also based on the many conversations I have had regarding alternative methods for the development of technology and applications in connection with my proposal to establish a Center for Airborne Remote Sensing and Technology and Applications Development (CARSTAD). This plan tries to fit in with the stated goals of the NASA Administration and with the strategic plan of the GSFC.

I would be most grateful if you would read this paper and respond with corrections, and suggestions for modifications and improvements. My experience in all aspects of Earth remote sensing is limited, so I am relying on feedback from those who are more experienced to help me put together a credible proposal. I would like to incorporate direct quotations in this paper, with your permission. In any case I would like to include a list of contributors to the paper; again with your permission.

Introduction

This paper presents an alternative to the present process for the development of NASA’s Earth remote sensing science and technology. The underlying principles of this alternative plan are smaller satellites and a continuous development process. Instead of a few large platforms with many instruments, many small satellites with one, or maybe two instruments would be launched. Instead of a single selection and development activity, instrument development and scientific participation would be a continuously evolving process.

Smaller, Cheaper, and Better

A constellation of smaller satellites would solve many problems. The swath width of the instruments would be narrower, eliminating some atmospheric and angle of incidence effects. The down-linked data rate would be deceased, and spread over more intervals. The opportunity would exist for cross-track pointing of the individual satellites for more frequent revisits of critical scenes. Multiple copies of an instrument would be required to provide global coverage. These instruments would be simpler than the present instruments because of the reduced swath width and decreased data volume requirements as well as a number of other factors associated with this new approach.

Smaller satellites would be cheaper to build because of their simplicity and due to the possibility to use the same subsystems in all of them. These small satellites would basically be the instruments with power, propulsion, guidance, and telemetry subsystems attached. The tedious and expensive process of selecting a major spacecraft bus, and dealing with all the interface and compatibility issues would be eliminated. By building multiple copies of the smaller instruments, it would be possible to incorporate modifications in successive generations that would both improve the performance and reduce the cost.

The downside of the smaller spacecraft (implying fewer instruments) and the narrower swath widths is the problem of synergy and simultaneity of data. Complementary instruments absolutely essential for a measurement will have to be flown together. A step in this direction is already being taken by JPL in their proposed development of a combined AIRS and microwave sounder. The smaller swath width means that passes by different satellites will require more “stitching together” of the scenes to achieve global coverage. This would be significant at the lower latitudes, but would not be a problem at higher latitudes due to the considerable overlap of swaths. A lot of redundant data could in fact be eliminated by selectively not collecting high latitude data. This would further reduce the data downlink requirement.

Continuous Development

Instead of having a “once in a lifetime” opportunity for investigators and institutions to participate in the EOS Program, the process should be continuous. A continuous development program would utilize the best features of NASA, academia, and the private sector. Smaller satellites would make this possible. If a “better idea” came along, it could be utilized because it would not involve changing the entire configuration of a large spacecraft. The lessons learned during the development of the technology and applications for one satellite could be utilized more expeditiously in the development of the succeeding satellite. In this way it would be assured that the state-of-the-art in science and technology would always be employed.

Another advantage to continuous development would be continuity of data. Instead of having major changes in instrument configurations, requiring major changes in data interpretation, the changes would be gradual and could be easily incorporated into scientific programs. Also, when a new satellite was launched it would be ready for use immediately as its performance and the data it produces would be well understood.

By splitting out the instruments from a large platform, development schedule pressures would be reduced. There would be no need for a spacecraft full of instruments to be delayed while waiting for one that was late. Schedule delays for the smaller and simpler instruments would be less likely in any case, particularly as incremental development procedures would make the process less risky.

A final advantage of continuous development is that the resources for the development of instrumentation would be kept in place on a long-term basis. This would eliminate the problems of having to find an organization, or put together some consortium with the capability to produce the hardware. Again, the smaller and simpler instruments would not require the level of expertise from the instrument manufacturer so development could be done by less experienced organizations.

Participation and Selection

A significant part of the Earth Science community is eliminated from participation in NASA’s programs by the present selection process. A process of continuous development would leave the doors open to the entire Earth remote sensing community. This would provide more opportunities for participation and would increase competition for places in NASA’s programs. Rather than simply award contracts to the winning bidders, the continuous development program would form partnerships to develop the new technology and applications. In some cases, where there are no good commercial applications of the technology, NASA will have to bear the majority of the cost. International partnerships would continue more or less as presently, though the instruments would be packaged for flight individually rather than integrated onto large platforms. This would simplify international agreements considerably as in most cases a foreign contribution would simply be the use of an already planned satellite as is.

Integration with the Existing Program

It is assumed that at least the first generation of the present EOS-AM and -PM configurations will be implemented. This will initiate the collection of the long-term global data set required by the MTPE. The continuous development program should be implemented as soon as possible. Some of the alternative planning that has already been done by instrument contractors and institutions is appropriate to this new program, such as the JPL effort already mentioned.

This page was last modified on 17 February 2015

 

John Bolton's Initiatives 1985 - Present

 

HOME