Current science projections for future earth-imaging instruments indicate the need for as many as 25 spectral bands, with bandwidths as narrow as 20 nanometers. The desire for a multiplicity of bands has led researchers to study various spectrally dispersive instrument designs as a means of providing the desired future capability. These instrument designs, however, are costly, complex, and of high technical risk. This paper describes a "multiband selection device" containing several spectral filters that can be placed at the exit faces of a broadband multiport beam splitter and thereby provide a multiplicity of spectral bands with a high degree of spatial coregistration while utilizing state-of-the-art linear array detectors. Fabrication of the multiband selection device has been successfully accomplished, and the design and test results are described.

This paper presents a figure-of-merit parameter for the scientific utility of multi-spectral linear array (MLA) instruments for civil remote sensing. The utility parameter is based on the observation that most analyses of earth resource satellite data are based on algorithms which combine measurements from two different spectral bands. Thus, the number of spectral band-pair choices available to the user of a given MLA instrument is offered as one very useful measure of the flexibility and overall utility of that particular instrument design. Also, the cost of a MLA instrument is a strong function of the number of linear arrays due to the requirement for additional electronics, larger optics, added thermal-controls and increased power requirements. However, research requirements currently project a need for a highly flexible instrument with a large number of spectral bands (arrays) for future earth remote sensing. Newly developing algorithms will undoubtedly continue to increase the number of desired spectral bands. Conversely, the bandwidth limitations of data-retrieval systems support an instrument design that has fewer linear arrays and changeable filters. This paper analyzes the Multi-band Selection Device capability mathematically, and provides a basis for a cost-benefit comparison of alternative multi-spectral instrument designs.

Next-generation pushbroom sensors for earth observation require high-performance optics that provide high spatial resolution over wide fields of view. Specifically, blur diameters on the order of 10 to 15 pm are needed over 5° to 15° fields. In addition to this fundamental level of optical performance, other characteristics, such as spatial coregistration of spectral bands, flat focal plane, telecentricity, and workable pupil location are significant instrument design considerations. The detector-assembly design, optical line-of-sight pointing method and sensor packaging all hinge on these secondary attributes. Moreover, the need for broad spectral coverage, ranging from 0.4 to 12.5 μm, places an additional constraint on optical design. This paper presents alternate design forms that are candidates for wide-field pushbroom sensors, and discusses the instrument-design tradeoffs that are linked to the selection of these alternate optical approaches.

All-solid-state pushbroom sensors for Multispectral Linear Array (MLA) instruments are under development. It is planned that these pushbroom sensors will replace mechanical scanners used on current LANDSAT earth resources satellites, providing improved performance and operational flexibility. A buttable, four-spectral-band, linear-format charge coupled device (CCD) and a buttable, two-spectral-band, linear-format, shortwave infrared charge coupled device (IRCCD) are being developed under NASA funding. These silicon integrated circuits may be butted end to end to provide multispectral focal planes with thousands of contiguous, in-line photosites. The visible CCD integrated circuit is organized as four linear arrays of 1024 pixels each. Each linear array views the scene in a different spectral window, resulting in a four-band sensor. The spectral windows are defined by integral bandpass filters. First-generation filters are interference stacks tuned to Thematic Mapper bands 1-4. The pixel center-to-center spacing of 15µm combined with a band-to-band, along-track spacing of only 60 pm provides a compact, attractive focal plane organization. The high quantum efficiency of the backside-illuminated CCD technology provides excellent signal-to-noise performance from 0.4 /.1111 to 0.9 μm. The backside-illuminated technology also results in unusually high MTF in the red. The shortwave infrared (SWIR) sensor is organized as two linear arrays of 512 detectors each. Each linear array is optimized for performance at a different wavelength in the SWIR band (1-3.5 μm). The actual spectral window of each band is defined by bandpass filters placed in close proximity to the chip. This dual-band infrared sensor consists of Schottky barrier detectors which are read out by CCD multiplexers. The detectors and the CCD registers are formed as one mono-lithic structure using standard silicon process technology. These IRCCD focal planes provide radiometric performance at 125K. This operating temperature, and the low power dissipation of 18 pW per detector, make this sensor compatible with satellite passive cooling. The detector center-to-center spacing is 30 Am with a band-to-band spacing of 300 μm.

Interpretation of imagery and photographs of various scales (altitudes) can be valuable in formulating new exploration concepts in both proven and frontier areas of oil and mineral exploration. This synergistic approach utilizes a methodology of proven merit and takes advantage of the growing variety of imagery and photo types and scales that are available in photographic archives. Analysis of satellite imagery combined with high, medium and low-altitude photography permits the interpreter to augment the regional, synoptic views at orbital scale with the magnified detail of lower altitude photography in searching for exploration clues. Three areas are selected for interpretation and preparation of multiple geologic overlays. They are Patrick Draw - West Desert Springs area, Wyoming (hydrocarbons), New Almaden district, California (Hg), and Getchell deposit, Nevada (Au). These examples demonstrate the approach and utility of multiple scales and types of imagery in documenting the geology of both metalliferous and petroleum-bearing areas.

The new LANDSAT-4,-5/Thematic Mapper (TM) land observational satellite remote sensing systems are providing dramatically new and important short wave infrared (SWIR) data, which combined with Landsat's Multi-Spectral Scanner (MSS) visible (VIS), very near infrared (VNIR), and thermal infrared (TI) data greatly improves regional geological mapping on a global scale. The TM will significantly improve clay, iron oxide, aluminum, and nickel laterite mapping capabilities over large areas of the world. It will also improve the ability to discriminate vegetation stress and species distribution associated with lateritic environments. Nickel laterites on Gag Island, Indonesia are defined by MSS imagery. Satellite imagery of the Cape Bougainville and the Darling Range, Australia bauxite deposits show the potential use of MSS data for exploration and mining applications. Examples of satellite syn-thetic aperture radar (SAR) for Jamaica document the use of this method for bauxite exploration. Thematic Mapper data will be combined with the French SPOT satellite's high spatial resolution and stereoscopic digital data, and U.S., Japanese, European, and Canadian Synthetic Aperture Radar (SAR) data to assist with logistics, mine development, and environ-mental concerns associated with aluminum and nickel lateritic deposits worldwide.

Analysis of Landsat and Seasat imagery reveals several structural and geobotanical anomalies in the northwestern lower peninsula of Michigan. Integrating these surface anomalies with regional cross-sections and geophysical studies shows that they lie along predictable subsurface trends. These trends control hydrocarbon migration and accumulation; anomalies occur in both producing and untested portions of the trends. Subsequent to the acquisition and analysis of the satellite data, several "untested" areas have been drilled and successfully completed, suggesting that the anomalies are associated with hydrocarbon microseepage. The geobotanical and structural characteristics of Michigan are strongly affected by the last stage of Pleistocene glaciation. In moving through this portion of Michigan, the ice was deflected by a series of topographic features related to subsurface highs. This is indicated by the concave bends in the Port Huron and Manistee moraines, as well as the glacial lake terraces that formed along the edges of the northern topographic high as the ice retreated. The structural interpretation suggests that a large patch reef system may have existed in the northern portion of the study area during deposition of the Devonian Traverse Formation and the Silurian Niagara Group. The Traverse Formation underlies the Antrim Shale that outcrops below the glacial drift of the area. A large patch reef system would account for the differential movement of the ice since the associated limestones would be more resistant to erosion than the surrounding shales. In comparing the location of the geobotanical anomalies with the glacial deposits, an interesting interaction is revealed: over 75% of the anomalies are located in the more porous and permeable outwash deposits. Furthermore, every vegetation anomaly is associated with a lineament or lineament intersection on the structural interpretation. The geobotanical anomalies are also very transient. They appear only during the initial "leaf-out" in the spring and as areas of pre-senesence in the fall. During the summer months they are not visible. Apparently the microseepage visibly affects the vegetation only at those times of the year when the vegetation is in a natural state of flux or stress. This vegetation "stressing" also seems to be limited to the larger species of trees that have extensive root systems. This suggests that only the larger trees sample the deeper, more highly affected soil horizons.

The Geosat oil and gas test site program stimulated interest in the interaction between surface hydrocarbon concentrations and interpretation of remote sensing data. The test case results suggested that lineaments correspond to avenues of preferential hydrocarbon seepage and that this seepage affects vegetation health and populations at Patrick Draw field in Wyoming and potentially at Lost River field, West Virginia. These two areas were selected for additional surface hydrocarbon surveys in order to test these hypotheses. The Patrick Draw study shows that a zone of stressed vegetation, visible on thematic mapper data, definitely coincides with an area of marked leakage of hydrocarbons and that the composition of these gases would predict an intermediate type oil and gas reservoir such as exists in the area. The study further indicates that the leakage is in large part controlled by the presence of fractures/faults recognized as lineaments on the remote sensing images. The Lost River study specifically investigated the possible existence of hydrocarbon leakage causing anomalous populations of maple trees in a climax oak forest. These maples were first recognized by study of thematic mapper simulator data. The soil gas hydrocarbon concentrations are above average in several of the maple anomalies over the field. This supports the inference that the maples are present because they are more tolerant of soil conditions where hydrocarbon seepage is active. The crest of the field has low soil gas magnitudes, but high values occur to the updip eastern edge of the field along a fault/fracture that was detected in the seismic data. The conclusion that preferential pathways of hydrocarbon leakage are recognized in spectral and textural analysis of remote sensing data is supported by other studies and integrated into a suggested exploration/hydrocarbon migration model.

In this paper a technique is proposed for analytical solving the problems arised to apply for civil space remote sensing, namely: determination of the detection thresholds of signals, remote measurement of the thickness of surface cover. These problems can be perceived as from the point of accepted decisions in the uncertain conditions. For their solving the main principles are applied from the theory of statistical decisions, the theory of estimation and the theory of optimization. The gynqral procedures are based on the usage of the idea of statistical invariant coupling method''', which one was taken into consideration for synthesis of statistical decision algorithms. As it is known to be rather clearly for having new results and the previous ones, to be effective on small samples of observations. The ideas of synthesis of statistical decision algorithms are being illustrated at easy examples and they are not overloaded by additional details. Practical meanings of results having widely received in the many areas: detection anomalies on the earth surface (monitoring problems), resource exploration, fisheries guidance, ship routing during winter navigation, etc).

The Centralized Storm Information System (CSIS) has been in place at the National Severe Storms Forecast Center (NSSFC) in Kansas City for two years. CSIS represents a major step toward providing the operational meteorologist a truly interactive, information handling and display system. The system consists of a GOES receiving antenna system, four Harris/6 computers, three interactive_ terminals, FAA "604" teletype input, two autodialers and are interfaced to the NSSFC computer. CSIS has had a dramatic and positive impact on the operations of NSSFC. The continued tailoring of the system to meet the operational needs of NSSFC has produced a high degree of forecaster acceptance and dependence on CSIS. Verification of tornado watches has shown a significant increase in skill since the implementation of CSIS.

The purpose of this paper is to present satellite-GOES based methodologies for analyzing and forecasting (short range) precipitation from convective systems, extratropical cyclones and tropical cyclones. Currently, satellite-derived precipitation estimates and 3-hour precipitation trends for convective systems, extratropical cyclones and tropical cyclones are computed on the NESDIS Interactive Flash Flood Analyzer and transmitted via AFOS to Weather Service Forecast Offices, Weather Service Offices and River Forecast Centers. These estimates and trends aid hydrologists and meteorologists in their evaluation of heavy precipitation events.

The presentation is intended as a contribution to greater understanding between data producers and data users. First, the historical evolution the data base for operational numerical weather prediction will be reviewed. Secondly, the characteristics of the various remotely-sensed observation systems will be outlined. Third, the objective analysis section of the data assimilation system will be described, with emphasis on the treatment of remotely sensed data. The discussion will conclude with some speculation concerning the challenges presented by new observing systems.

From the beginning of December 1983 through mid-February 1984 the Cooperative Institute for Meteorological Satellite Studies (CIMSS) carried out an exercise to deliver temperature and moisture profiles, derived from the GOES-6 VISSR Atmospheric Sounder (VAS), to the National Meteorological Center (NMC) in time for input to the operational forecast at 1330 GMT. The purpose was to provide meteorological data coverage over the data sparse eastern Pacific (FPAC) where timely polar orbiting satellite data are not available. Although a product was delivered only 40 percent of the time, the experiment successfully demonstrated the feasibility of a totally automated VAS retrieval procedure. Data reliability achieved at the EPAC scale appears to be good, though lack of independent verification data requires that forecast impact studies delineate their ultimate value.

A brief review is presented of simulation studies and real data experiments which were conducted to assess the impact of satellite observations on numerical weather prediction. These experiments show that while there has been some redundancy between observing systems, satellite data has made significant contributions toward improving global forecasting.

VAS [VISSR Atmospheric Sounder (Visible Infrared Spin Scan Radiometer)] data from the GOES-west are input to analyses used as inifial conditions for 48-hour numerical integra-tions of the Limited Fine-mesh Model (LFM) used operationally at the National Meteorological Center (NMC). Of four cases examined using VAS data, three produced improved forecasts of 500-mb height, while one produced a degraded forecast. The results are considered promising in regard to future use of such data in models such as the LFM.

At the Goddard Laboratory for Atmospheric Sciences (GLAS) we have developed a physi-cally based satellite temperature sounding retrieval system, involving the simultaneous analysis of HIRS2 and MSU sounding data, for determining atmospheric and surface conditions which are consistent with the observed radiances. In addition to determining accurate atmospheric temperature profiles even in the presence of cloud contamination, the system provides global estimates of day and night sea or land surface temperatures, snow and ice cover, and parameters related to cloud cover. Details of the system are described elsewhere. Here, a brief overview of the system is presented as well as recent improvements and previously unpublished results, relating to the sea-surface intercomparison workshop, the diurnal variation of ground temperatures, and forecast impact tests.

A detailed computer simulation of the WINDSAT global wind measuring process has been developed and used to establish error limits as a function of design parameters. Studies were conducted for a WINDSAT research system in a 300 km and an 800 km orbit. Wind measuring errors were less than 2 m s-1 in the troposphere for the recommended set of para-meters. Our study results indicate the feasibility of measuring global winds from a space platform using a Doppler laser radar.

In 1978 the United States launched the oceanographic satellite SEASAT. Although the satellite operated successfully for only one hundred days in space, it was a monumental event. In the years that followed, atmospheric and oceanographic scientists revealed the true potential of such a system. Following SEASAT there was a NASA proposal for SEASAT B, and the interagency (NASA, NOAA, NAVY) oceanographic satellite NOSS. Unfortunately for those of us who rely on accurate forecasts of ocean conditions, neither system became a reality. However, reiterating the operational requirement for such a system, in April 1981 the Director, Naval Oceanography Division, Office of the Chief of Naval Operations formulated the proposal for N-ROSS, the Navy Remote Ocean Sensing System. From the beginning, one thing was obvious, Navy could not afford to go it alone. It was true that the Defense budget was larger, but money needed to build N-ROSS was also required to build ships, aircraft and weapons systems. We approached NASA and NOAA informally with the N-ROSS proposal. Could they help? Although neither agency was in a position to make a firm commitment, we were encouraged to determine the Navy's willingness to commit itself to such an undertaking. In addition we asked the Air Force to help with the launch, command and control phases. They too gave us the encouragement to proceed.

The analysis of the requirements for satellite-derived oceanic data to meet the needs of the civil marine community has been a continuing effort by the National Oceanic and Atmospheric Administration (NOAA). The priorities established for the most critical data are, in decreasing order: winds, sea surface temperature, waves, sea ice, ocean color, and circulation and currents. The general needs of academic, commercial, and governmental users are considered, with more detailed requirements for wave data presented as an example. The consideration of these needs are presented within the context of planned satellites that have been designed to meet national and international scientific and operational requirements.

The global view of the oceans seen by Seasat during its 1978 flight demonstrated the feasibility of ocean remote sensing. These first-ever global data sets of sea surface topography (altimeter) and marine winds (scatterometer) laid the foundation for two satellite missions planned for the late 1980's. The future missions are the next generation of altimeter and scatterometer to be flown aboard TOPEX (Topography Experiment) and NROSS (Navy Remote Ocean Sensing System), respectively. The data from these satellites will be coordinated with measurements made at sea to determine the driving forces of ocean circulation and to study the oceans role in climate variability. Sea surface winds (calculated from scatterometer measurements) are the fundamental driving force for ocean waves and currents (estimated from altimeter measurements). On a global scale, the winds and currents are approximately equal partners in redistributing the excess heat gained in the tropics from solar radiation to the cooler polar regions. Small perturbations in this system can dramatically alter global weather, such as the El Niho event of 1982-83. During an El Ni?io event, global wind patterns and ocean currents are perturbed causing unusual ocean warming in the tropical Pacfic Ocean. These ocean events are coupled to complex fluctuations in global weather. Only with satellites will we be able to collect the global data sets needed to study events such as El Ni?o. When TOPEX and NROSS fly, oceanographers will have the equivalent of meteorological high and low pressure charts of ocean topography as well as the surface winds to study ocean "weather." This ability to measure ocean circulation and its driving forces is a critical element in understanding the influence of oceans on society. Climatic changes, fisheries, commerce, waste disposal, and national defense are all involved.

Polar ice has a significant impact on world climate and on ocean characteristics. Transfer of heat from tropical oceans to the polar regions is regulated by sea ice, which locally insulates the ocean from the cold atmosphere. The continental ice sheets of Greenland and Antarctica represent vast reservoirs of fresh water which can significantly impact sea level if the ice sheets are changing in size. Satellite remote sensing gives information on many aspects of the ice cover: sea-ice extent and physical characteristics; detailed images of ice floes and open-water leads within the ice pack; sea-ice movement; zones of summer melting and snow-accumulation rates on the continental ice sheets; accurate estimates of ice-surface elevation, and detection of zones on the ice sheet that are either thickening or thinning; accurate, all-weather mapping of ice coastlines and large crevasses, and estimates of ice discharge rates from the ice sheets.

The evolution of satellite altimetry from SKYLAB-1, GEOS-3, and SEASAT-1 to the current U.S. Navy program, GEOSAT-A, has been accompanied by a continuing improvement in measurement precision along with a better understanding of those factors internal and external to the instrument that act to limit precision. The precision goal of 2 centimeters imposed by the TOPEX mission will be realized by incremental enhancements to existing designs whose effectiveness has been demonstrated. The TOPEX altimeter will use a high pulse-rate, burst waveform and dual frequency (13.6/5.3 Ghz) operation to minimize height measurement noise and remove ionospheric bias.

About three years ago, NASA's Oceanic Processes Branch requested that we evaluate alternate tracking concepts for sub-decimeter positioning of earth orbiting spacecraft such as TOPEX. Current high precision techniques, specifically TRANET and ground based laser ranging, yielded at best about 40 cms orbit accuracies on Seasat which flew at an 800 km altitude. Improved versions of these techniques, when coupled with projected improvements in knowledge of the gravitational field of the earth and with the higher altitude (1300 km) of 'IOPEX, appear capable of achieving about a factor of 3 improvement in orbit accuracy. Even at this level orbit errors dominate the overall system error budget. Our study led us to adopt a new tracking system concept based on utilization of the constellation of Naystar satellites in the Global Positioning System (GPS). In simplest terms, this concept involves simultaneous and continuous metric tracking of the signals from all visible Naystar satellites by approximately six globally distributed ground terminals and by the TOPEX spacecraft (see Figure 1). Error studies indicate that this system could be capable of obtaining decimeter position accuracies and, most importantly, around 5 cm in the radial component which is key to exploiting the full accuracy potential of the altimetic measurements for ocean topography. A series of proof-of-concept demonstrations has been completed with a pair of recently developed precision GPS receivers. This paper provides brief background discussions of the GPS, the precision mode for utilization of the system, past JPL research for using the GPS in precision applications, the present tracking system concept for high accuracy satellite positioning, and results from the demonstration.

In 1983, a study for a scatterometer to be flown on the Navy Remote Ocean Sensing System was initiated. This mission will be launched in the late 1980's and will operate for a period of 3 years. This paper briefly describes the design of the scatterometer instrument and discusses some of the technical issues involved.

The all-weather, global determination of sea surface temperature (SST) has been identified as a requirement needed to support naval operations. The target SST accuracy is +1.0 K with a surface resolution of 10 km. Investigations of the phenomenology and technology of remote passive microwave sensing of the ocean environment over the past decade beginning with the Navy specification of the Remote Ocean-surface Measurement System (ROMS), through the NASA launched Scanning Multichannel Microwave Radiometer (SMMR) flown on both SEASAT and NIMBUS-7, to the planning by NASA of the Large Antenna Multichannel Microwave Radiometer (LAMMR), and development of the Mission Sensor Microwave/Imager (SSM/I) to be flown in 1985 by the Navy/Air Force, have demonstrated that this objective is presently attainable. Preliminary specification and trade-off studies have been conducted to define the frequency, polarization, scan geometry, antenna size, and other essential parameters, as well as the retrieval algorithms and spacecraft interface requirements, of the Low Frequency Microwave Radiometer (LFMR). As presently planned, the LFMR will be a stan6 alone system completely independent of the SSM/I but with a 30 rpm conical scan at 53.1 incidence angle identical to the SSM/I. It will be a dual-polarized, dual-frequency system at 5.2 and 10.4 GHz using a 5.9 meter deployable mesh surface antenna. It is to be flown on the Navy-Remote Ocean Sensing System (N-ROSS) satellite scheduled to be launched in late 1988.

The Coastal Zone Color Scanner (CZCS) was launched on Nimbus-7 in October of 1978 as a research tool to determine if the biological and non-biological content of the ocean could be determined by remote sensing to a degree of accuracy useful for oceanographers. By the end of the first year of the sensor's life, the algorithms for calculating such things as pigment concentration and diffuse attenuation coefficient had been well developed, and ship measurements showed that the agreement with surface truth was better than the original goal set for the sensor data product in the open ocean. Near shore, where high sediment levels prevented the atmospheric correction algorithm from working properly, gradients could be observed, but quantitative accuracy was not as good as in offshore in waters ranging from pigment concentrations from near 0 to 2 milligrams per cubic meter. Since the sensor and spacecraft designed to last for one year have now operated for over five years, schemes have been devised to use the data in a practical manner for both monitoring long term effects such as pigment concentration throughout the year along the East Coast, and direct assistance to fishermen in the Gulf of Mexico and on the West Coast. The pigment concentration derived from the CZCS imagery has been shown to agree very well with the maps produced by ships at sea, and to cover a much larger area than the ships could possibly cover in a very short period of time. This added capability allows the ships to be utilized on more specific tasks rather than simply making grid measurements over large areas such as the Chesapeake to Cape Cod, and requiring one month of ship time to produce an image. The West Coast Fisheries Demonstration Project utilizing the CZCS uses real time imagery, collected at the Scripps Institution, processed overnight, and transmitted to tuna fishermen the next day from Monterey, California. Comparison of tuna catch with the color data has shown that the tuna are not sensitive to temperature as originally thought, but are, in fact, hunting by staying in the clear water on the edges of the cool more turbid water where their prey normally live. Since the tuna hunt by eye, they must stay in the clear water in order to be able to see the other fish that they prey upon. Comparison of fish catch with the imagery shows that the largest fish catch per day is quite clearly correlated with this water mass boundary between the clear and the turbid water. In the Gulf of Mexico, a serious hypoxic condition was monitored using the CZCS data. A large body of hypoxic water preventing the migration of shrimp from the shore out into the Gulf was clearly defined using the satellite data, and the magnitude of the problem could be quickly analyzed by personnel of the National Marine Fisheries Service in Bay St. Louis, Mississippi. These applications, far from the original intent of the instrument, are ongoing, however, the wisdom of starting new projects is debatable since the CZCS has now exceeded its original expected lifetime by a factor of five, and the United States has no plan to fly a replacement instrument.

As a proof of concept, the Seasat oceanographic satellite successfully demonstrated the role of microwave sensors to synoptically measure geophysical parameters of interest to the oceanographic and meteorological communities. Nevertheless, problems surfaced, which can be corrected by improved system design concepts. Indeed, the next generation of satellite systems will incorporate modest engineering improvement which will greatly ease data reduction problems. Of more importance; however, the Seasat activity has generated ideas for new classes of measurements for the future. This paper explores a possible long term system growth in microwave scatterometry and radiometry.

A project is underway at the National Environmental Satellite, Data, and Information Service to develop products derived from satellite data that will be useful to USDA for agricultural monitoring. The products fall into two categories: meteorological observations and direct observations of vegetation. The meteorological-type measurements are to supplement conventional ground-based weather observations from data sparse areas of the world. These include estimation of precipitation, maximum/minimum temperature, solar radiation, and snowcover. The algorithms to retrieve these quantities were developed to work from operationally available data and designed to be implemented in a near real-time operation. Direct observations of vegetation are interpreted by means of indexes consisting of mathematical combinations of observations in visible and near infrared bands. This paper briefly describes the algorithms used to estimate these quantities from operational data, discusses the expected accuracy of the products, and reviews current problems.

Global agricultural production information is the key to many economic decisions. National level planners use it to plan imports or to assess balance of payments, farmers use it to make planting decisions, lending and aid institutions use it to plan loans and aid needs, commodity buyers use it to plan purchases. Traditional information systems are slow, offer little confidence and may be inaccurate; systems based on the use of space remote sensor systems are, on the other hand, fast, provide good confidence and are demonstrating improving accuracies. The system structure for remote sensor assisted agricultural information systems is centered on a geobased structure, mapped outputs pinpoint locations where plant stress is impacting yields. Meteorological satellite assessments pinpoint where rainfall and significant solar radiation is impacting the plant environment. The CROPCAST Agricultural Information System offers an opportunity to examine an operating system which contains characteristics essential to all future systems. CROPCAST's use of a grid/cell geobased structure provides a mechanism to effectively use remote-sensor derived data of all types, i.e., Landsats, metsats, aircraft and human eyeball derived data. Predictive models operating in CROPCAST provide updated agricultural assessments in the time intervals when no Landsat or other field observation data are available. Economic models provide the opportunity to merge CROPCAST diagnostic and predictive output with the market place at both the cash and futures level. This presentation will examine the CROPCAST structure as a model for future uses of remote sensing data from civil remote sensing systems in assessing global agricultural production. A review of the future direction to be taken by the CROPCAST System will be included to identify new avenues for remote sensor-based agricultural information system growth over the coming decade of change in remote sensor systems.

An integrated climate impact monitoring system was designed using all available data sources, including satellite data. Rather than attempt to use the complex processes that have been developed, the project began with the basic premises of image interpretation and introduced the simplest and most widely accepted procedures into an existing operational process. The intent was to improve the accuracy and timeliness of the impact assessment reports without a large increase in resources. To do this the human mind was used to do what it does best and the computer to do what it does best. The human mind is used to look at color images for pattern recognition and to do all of the final analysis. The computer does all of the computation and presentation but does none of the analysis.

While we do not have detailed understanding of the interactions and feedbacks between snow cover and the atmosphere, it is clear from even simple energy balance models and preliminary observational studies l that snow area is potentially a very important climatic variable. The areal distribution of snow is central in determining the global albedo as well as prescribing the surface boundary` temperature, moisture, and to some extent, the low level atmospheric static stability. Late season (i.e., spring) snow cover has been related to soil moisture which is in itself an important component of the climate system. While several climatic variables are adequately sampled through a network of surface observation stations, other variables, among them the snow cover, can only be sufficiently monitored by satellite. The National Oceanic and Atmospheric Administration (NOAA) has been analyzing satellite imagery on a weekly basis for operational estimates of snow cover for over fifteen years. This paper presents a synopses of recent climate research and climate diagnostics studies using these data at the National Weather Service's Climate Analysis Center. Currently available satellite products are evaluated in the light of these studies and a set of desired characteristics for future satellite products are presented.

Because of the large contrast between the dielectric constant of liquid water and that of dry soil at microwave wavelength, there is a strong dependence of the thermal emission and radar backscatter from the soil on its moisture content. This dependence provides a means for the remote sensing of the moisture content in a surface layer approximately 5 cm thick. The feasibility of these techniques has been demonstrated from field, aircraft and spacecraft platforms. The soil texture, surface roughness, and vegetative cover affect the sensitivity of the microwave response to moisture variations with vegetation being the most important. It serves as an attenuating layer which can totally obscure the surface. Research has indicated that it is possible to obtain 5 or more levels of moisture discrimination and that a mature corn crop is the limiting vegetation situation.

In cloud free areas the NOAA Advanced Very High Resolution Radiometer acquires data which relate to agricultural conditions. The visible and near infrared channels respond to the spectral signature of vegetation, providing information on the extent and vigor of crop cover, the thermal infrared (temperature) data are indicative of surface moisture condi-tions, as the cooling effect of evapotranspiration is readily apparent in the AVHRR data. Previous studies of Carlson and Boland,' Price,2 and Rosema et al.,3 have developed the formalism for reducing satellite thermal infrared data to quantitative estimates of surface moisture conditions. The results are generally reasonable, but have not been tested by comparison with surface observations. This problem of verification is chiefly one of spatial scale, e.g., the 1100 m scale of AVHRR data falls at the geometric mean of the scale of weather observations (100's of kilometers) and that of typical experimental ground truth (10's of meters). Yet precipitation and soil moisture at the scale of AVHRR are dominant variables affecting agricultural productivity. Data sets collected through the Statistical Reporting Service of the USDA represent an initial effort to permit tests the utility of AVHRR data for soil moisture and crop yield assessment. The physical and statistical formulations are discussed for dealing with these complex data sets, and the framework of a hypothetical operational system is addressed. The observing system of choice combines daily AVHRR passes with the infrequent but high spatial resolution data from Landsat or SPOT sensors.