This paper summarizes the system concept of the MLA Instrument as developed during the NASA Goddard Space Flight Center Instrument Definition Study (Contract No. NAS5- 26588). The purpose of this study was to encompass the definition, design and analysis of a candidate instrument to support an Experimental Land Observing System. In the process of this instrument definition, trade-off analyses have been conducted to identify and quantify technological risk, cost and performance with the intent of refining the MLA Instrument. The Multispectral Linear Array (MLA) instrument is a combined multi-channel radiometer and wide field imaging telescope for earth resources observation from space. The MLA sensor is designed to provide high resolution images as well as absolute radiometric measurement in 6 discrete spectral bands ranging from 0.45 to 2.35 μm. The Honeywell design centers around a reflective off-axis anastigmat that is well corrected for a 15° linear field-of-view. The major subassemblies are the telescope structure, focal plane/dewar assembly, radiative cooler, calibration sphere, stereo pointing mirror assembly, missed scene roll mechanism and the signal processing electronics module.

A concept for an advanced earth resources system was described in May 1981 at the ERIM Symposium in Ann Arbor, Michigan. Based on those concepts, this paper discusses the improved performance obtained in a modified instrument that includes new optics and mecha-nisms. An essential element of this design is a combined VIS/SWIR detector array with integral color filters that is described in other papers in this seminar. This array is cooled to 125 K by a passive radiator for which a design and thermal analysis are presented. A system is outlined for absolute calibration on orbit, using reflected sunlight. The predicted performance of this instrument is presented in graphs and simulated pictures.

The heart of the MLA design problem is to develop an instrument concept which concurrently provides a wide field-of-view with high resolution, spectral separation with precise band-to hand registration, and excellent radiometric accuracy. Often, these requirements have conflicting design implications which can only be resolved by careful tradeoffs that consider performance, cost, fabrication feasibility and development risk. The key design tradeoffs for an MLA instrument are addressed in this paper, and elements of a baseline instrument concept are presented.

A major requirement of multispectral imaging systems for advanced Earth remote sensing is the provision for greater spectral resolution and more versatile spectral band selection. The imaging spectrometer instrument concept provides this versatility by the combination of pushbroom imaging and spectrally dispersing optics using area array detectors in the focal plane. The Shuttle Imaging Spectrometer concept achieves 10-and 20-meter ground instantaneous fields of view with 20-nanometer spectral resolution from Earth orbit. Onboard processing allows the selection of spectral bands during flight; this, in turn, permits the sensor parameters to be tailored to the experiment objectives. Recent advances in optical design, infrared detector arrays, and focal plane cooling indicate the feasibility of the instrument concept and support the practicability of a validation flight experiment for the Shuttle in the late 1980s.

An earth radiation budget sensor is proposed which is optimized to yield shortwave and total flux measurements over the entire earth on a spatial scale of ≈500 km without mechanical scanning. The elimination of mechanical scanning reduces instrument complexity and increases reliability and expected operational lifetime. A mosaic array of detectors views the total earth in contiguous elements with 250 km resolution at nadir and gradually increasing resolution towards the edges. The pattern produced at the top of the atmo-sphere is chosen to meet scientific and data reduction goals. The sensor also includes complete earth viewing channels which provide an independent check against the integral of the mosaic detector measurements. An effective spatial scale of ≈500km is achieved by sorting the data according to geographic location and inte-grating over the entire range of outgoing angles at each location. Over a month's period, sufficient data are collected to fill the range of outgoing angles so that it becomes unnecessary to use bidirectional reflectance models to estimate the outgoing flux. All channels, including the complete earth viewing channels, are calibrated by the same source. Calibration against the sun is provided on a regular basis without rotation of the instrument. The detectors to be used are an advanced, low noise version of the active cavity radiometers used on the present ERBE nonscanning instrument. These detectors are sufficiently sensitive so that no imaging optics is required, thus minimizing polarization and spectral flatness problems. The cavity type detectors provide maximum spectral flatness from 0.2 to 50 μ by means of their radiation trapping geometry. Sketches of the proposed instrument and its calibration mechanisms are presented.

The spectral and spatial requirements for remote sensing in the next decade are presented. The requirements presented were obtained through extensive literature research and discussions with leading members of the various remote sensing research communities. In the 0.35 - 2.5μm region of the spectrum, numerous bands will be needed at bandwidths as narrow as 10 - 20 nanometers. There is also growing interest in the thermal infrared (8 - 14μm). Spatial resolution (instantaneous field of view) of 5 to 10 meters will be of great benefit to many fields of remote sensing.

An approach has been developed for real-time processing of land observing satellite imagery transmitted downlink at rates of 300 megabits per second or higher. This approach eliminates the necessity for large-scale storage of unprocessed data, and affords the capability for near-real-time display of the collected and processed image data. A test bed for softcopy display of digital imagery, developed in house by E-Systems, has been used to demonstrate the concept and the capability of achieving the required processing speeds.

A computational model of the deterministic and stochastic processes involved in remote sensing is used to study spectral feature identification techniques for real-time onboard processing of data acquired with advanced Earth-resources sensors. Preliminary results indicate that: Narrow spectral responses are advantageous; signal normalization improves mean-square distance (MDS) classification accuracy but tends to degrade maximum-likelihood (MLH) classification accuracy; and MSD classification of normalized signals performs better than the computationally more complex MLH classification when imaging conditions change appreciably from those conditions during which reference data were acquired. The results also indicate that autonomous categorization of TM signals into vegetation, bare land, water, snow and clouds can be accomplished with adequate reliability for many applications over a reasonably wide range of imaging conditions. However, further analysis is required to develop computationally efficient boundary approximation algorithms for such categorization.

The visible and reflected infrared channels of the NOAA-7 Advanced Very High Resolution Radiometer (AVHRR) are used for environmental monitoring, including agricultural crop assessment. An index consisting of a combination of digital data from these channels is used to screen the AVHRR data for cloud. In this paper, we consider the effects on spectral signature of sub-pixel sized cloud, which does not fill the entire IFOV of the instrument. It is shown that, for two different targets, for clear and turbid atmospheric conditions and for various scan angles, cloud which would not be detected during the screening process will cause substantial effects (up to a five-fold decrease) in the data.

An end-to-end digital computer simulation can be a useful tool to determine the relationship between a system's performance and design parameter choices. Properly written, the simulation can allow easy parameter adjustment, and can permit the analyst to quickly com-pare both quantitative and qualitative imaging performance over a broad range of system configurations. To test the utility of such a simulation, we developed a model of a Forward Looking Infrared (FLIR) Imager. The model is based on image processing routines that translate a digital image used as an input to another image that represents the system output. The simulation design is discussed and examples of its performance are presented.

A two mirror flat field anastigmatic telescope was designed for multispectral sensing. The design was adapted to prism-type beamsplitting arrangements without loss of multispectral image quality by the addition of one refractive element. In addition to being relatively simple and mechanically insensitive, the design is immune to focus shift caused by index of refraction variation with temperature.

When individual detectors in scanned focal plane arrays are sampled at a rate of once per instantaneous field of view (IFOV), a problem arises in measuring detector optical modulation functions, (SWR, MTF). Ideally, one would like to pass a slit illumination or knife edge illumination across an IFOV, generating respectively, a line spread function or knife edge response curve from which optical modulation could be calculated versus spatial frequency. Use of sampling phase delays is a possible solution, but is subject to phase errors resulting from scan rate variations. The method described here is a "phased knife edge" approach. It employs a multi-bar reticle image, designed so that successive bar edges are slightly shifted in phase with respect to an IFOV area as the IFOV signal is sampled during a single scan. The resulting data, after conversion to a knife edge response function, is convolved with a sliding phase, computer-generated square wave image to find SWR versus spatial frequency. MTF is obtained by converting the knife edge response to a line spread function and applying standard algorithms for convolution of the LSF with sinusoidal bar patterns of desired spatial frequency.

This paper discusses the conceptual design of an electronic subsystem that processes information from a six-band linear array focal plane with 92,160 detectors contained with 20 IFOV in the along-track direction. Charges generated in the focal plane detector chips are converted from a voltage to current in electrometers integral to the detector chip. These low-level analog signals are amplified in analog signal processors. The data processor converts the detector analog information into digital form for processing and formatting prior to transmission over the spacecraft communications links. The processor is required to compress the data so that it does not exceed downlink capacity, to implement on-board offset and gain correction, and to format the data with error correction code, housekeeping data, and synchronization bits.

The Multispectral Linear Array Sensor is being developed by NASA as a next generation remote sensor. The development is presently in a competitive design phase. This paper describes the on-board signal processing electronics for the Honeywell design. The signal processing includes focal plane electronics for push broom arrays in six spectral bands, on-board radiometric calibration, digitization, bandwidth compression, and formatting for transmission over a high bandwidth data link. The signal processing must adapt to different detector dwell times and bandwidth compression ratios for three different orbital altitudes. A high degree of fault detection and redundant circuitry is incorporated into the design to meet the high reliability requirements of a space sensor.

A general purpose "pushbroom" sensor will have parameters determined by the needs of the majority of potential users. These parameters may not satisfy the needs of certain users: Agriculture requires a very short return visit interval; cartography requires a very small pixel size; Land Use and Geology would be satisfied with moderate resolution and seasonal return times. The aggregate solution of these needs would produce a sensor with extremely high data rates. A sensor concept is proposed which may meet the combined needs without the extreme data rate.

With the advent of new satellites with mappers having wider spectral distribution than previously available a new remote sensing radiometer was needed. In addition to requiring additional spectral channels, improved sensitivity and stability was required. This paper describes the design to meet the goals and evaluates how well the first units met these goals.

This paper discusses the conceptual design of a focal plane for a spaceborne instrument that provides six rows of visible and shortwave infrared sensing elements on two intimately assembled rows of detector chips. The six color bands are all within 20 IFOV in the along track direction. This unified "silicon based" focal plane uses interference spectral filters deposited directly on a linear array of buried channel CCD visible detector chips and palladium silicide back-biased Schottky diode SWIR detector chips. The paper will emphasize the technology associated with detector arrays and deposited stripe filters. The multispectral sensing focal plane design described here was developed partly on Westinghouse funds and partly under subcontract in support of the Eastman Kodak Multispectral Line Array Instrument Definition Study, Contract NAS5-26589, and experimental evaluation of the design technology was supported in part under contract NAS5-25622, Stripe Filters on Multispectral Linear Arrays. (This paper is one of a group of three related Westinghouse papers with 345-14, Multispectral Linear Array Focal Plane Signal Processing, and paper 345-19, Multispectral Linear Array Focal Plane Mechanical and Thermal Design. The overall instrument concept for which this focal plane was designed is described in paper 345-02, Improved Earth Resources Sensing Instrument, by George Keene of Eastman Kodak.)

The mechanical and thermal design of an integrated focal plane subsystem of a nultispectral Linear Array (MLA) instrument is discussed in terms of focal-plane alignment, thermoelastic performance, and thermal requirements. The modular construction and thermal control of the focal plane array are discussed.

Monolithic 32 x 64 and 64 x 128 palladium silicide (Pd2Si) interline transfer IRCCDs sensitive in the 1-3.5 pm spectral band have been developed. This silicon imager exhibits a low response nonuniformity of typically 0.2-1.6% rms, and has been operated in the temperature range between 40-140K. Spectral response measurements of test Pd₂Si p-type Si devices yield quantum efficiencies of 7.9% at 1.25 μm, 5.6% at 1.65 μm and 2.2% at 2.22 μm. Improvement in quantum efficiency is expected by optimizing the different structural parameters of the Pd₂Si detectors. The spectral response of the Pd₂Si detectors fit a modified Fowler emission model. The measured photo-electric barrier height for the Pd₂Si detector is ≈0.34 eV and the measured quantum efficiency coefficient, C₁, is 19%/eV. The dark current level of Pd₂Si Schottky barrier focal plane arrays (FPAs) is sufficiently low to enable operation at intermediate tem-peratures at TV frame rates. Typical dark current level measured at 120K on the FPA is 2 nA/cm². The Pd₂Si Schottky barrier imaging technology has been developed for satellite sensing of earth resources. The operating temperature of the Pd2Si FPA is compatible with passive cooler performance. In addition, high density Pd2Si Schottky barrier FPAs are manufactured with high yield and therefore represent an economical approach to short wavelength IR imaging. A Pd₂Si Schottky barrier image sensor for push-broom multispectral imaging in the 1.25, 1.65, and 2.22 μm bands is being studied. The sensor will have two line arrays (dual band capability) of 512 detectors each, with 30 μm center-to-center detector spacing. The device will be suitable for chip-to-chip abutment, thus providing the capability to produce large, multiple chip focal planes with contiguous, in-line sensors.

The typical Multispectral Linear Array (MLA) Instrument mission would be to gather high-resolution, radiometrically accurate earth resources data in several spectral bands over a prolonged period of time. These bands would include the visible (VIS), near infrared (NIR) and short wavelength infrared (SWIR). Silicon charge-coupled imaging devices (CCDs) can be assembled into contiguous pixel focal planes which will cover the VIS/NIR region and operate reliably for several years in a space environment. A typical MLA focal plane would have approximately 12,000 pixels, with a pixel-to-pixel registration requirement on the order of ±0.1 pixel. Fairchild Weston Systems has developed the technology to assemble such focal planes. This paper describes that technology, and addresses the problem of polarization sensitivity associated with certain types of focal plane assemblies. Since one of the environmental factors affecting the long term (5 years or more) operation of CCD MLA focal planes is space radiation, the paper also discusses radiation effects on CCDs, and describes a practical solution to the problem through the use of shielding.

Indium antimonide (InSb) photodiode technology has progressed to the point where charge integration can now be accomplished directly on the diode junction capacitance. Recent development work at the jet Propulsion Laboratory has combined a 128-element linear array of InSb detectors with a silicon FET switched multiplexer (MUX) and the JFET preamp for readout. This device is characterized by a relatively low read noise of ',1200 electrons, large pixel charge storage capacity of 2 x 10⁷ electrons, 2-3% responsivity nonuniformity, high quantum efficiency (≥0.8, and an NEP of ≈3 x 10^-17W /Hz at 77 K.