Forest change detection over a decadal time frame was conducted for the Maya Biosphere Reserve in northern Guatemala. A simple and logical method of visualizing and quantifying forest change is presented. Analysis of time- series Landsat-Thematic Mapper imagery provided estimates of forest change at three time periods; prior to 1990, 1990 to 1993 and 1993 to 1995. Four dates of Landsat imagery were pre-processed, co-registered to a UTM projection and the normalized difference vegetation index was computed for each date. An unsupervised classification was performed and cluster classes were grouped into time-series change/no change categories. A color coded image was generated which resembled the RBG-NDVI color composite of the 1990, 1993, and 1995 imagery. Land cover information and Geographic Information System (GIS) editing techniques were applied to resolve some confusions between forest change and change in non-forest types. Results indicated that forest clearing rates in the reserve were less than 0.5 percent per year in the early to mid 1990s but the buffer zone clearing rates, at over two percent, were much higher.

Terrestrial spectroscopy in the 3 - 5 micrometer wavelength region offers unique applications, such as detection of carbon and sulfur compounds, and subpixel temperature determination of ground fires and volcanic lava flows. Two challenges in this region are the similar levels of solar reflected and terrestrial emitted fluxes, and the low values of both these fluxes making measurements of high quality more difficult than in the short-wave or long-wave IR regions. We collected airborne hyperspectral images with a Fourier transform spectrometer over seawater in a tropical environment combined with radiosonde data and MODTRAN atmospheric modeling, to distinguish ground radiances, atmospheric contributions, and spurious instrumental effects. Within the range of our radiosonde data (3 km altitude), MODTRAN very closely reproduces the sensor- detected mid-IR fluxes. Clear separation between solar reflected and terrestrial emitted components is achieved, and different atmospheric compositional spectral effects are analyzed. We applied the results from this well-truthed experiment, to spectra collected of the same locations at different times without supporting radiosonde data, to assess surface radiance accuracy and atmospheric variability.

Earth observation data is being acquired in increasing quantities, spurring a need for the ability to provide well- documented, organized, and accessible information databases. The collection and conjunctive use of multiple data types and formats of differing levels of complexity obtained from a variety of sources requires a sophisticated management system. Users desire a fast, efficient, easy-to-use system with the capacity to archive, associate, and access very different types of data. TRW's InfoWeb^TM multimedia data storage and dissemination system is being used to organize and maintain not only remotely sensed images, but also associated text, photo, audio, and video information. Pertinent examples include anything from journal articles to videos of biological processes. Information acquired can be tagged with latitude and longitude in the archiving process for subsequent geospatial search and retrieval. The system supports insertion of hypertext links, and database browse and query capability. In addition, access to external sources of information can be provided over the.Internet. A prototype reference database has been developed that illustrates the advantages of the system in facilitating the use of large amounts of multimedia ecosystem data.

Effective management of renewable resources requires accurate and current resource data bases on which to base decisions. The detection, identification, and mapping of change lies at the center of maintaining geographic data bases. The synoptic and repetitive acquisition capabilities of satellite imaging in conjunction with its ability to deliver data from bands of the EMR spectrum, which are sensitive to biophysical characteristics of vegetation, make satellite born sensors a potentially valuable tool in resource monitoring. It is the purpose to this research to determine quantitative classes of green biomass change in aspen dominated ecosystems in the upper Great Lakes. The different effects of the atmosphere on each image must be minimized to make the pixel values represent the same level of recorded EMR in both images. Both digital number matching and exoatmospheric reflectance matching will be compared in effectiveness of minimizing atmospheric effects. Both approaches use a variety of radiometrically stable targets to normalize the pixel values of one image to those of the other. Once the pixel values of both images 'represent' the same characteristic on the ground, they may be readily compared. Tasseled cap indices will then be calculated for each image and entered in a change detection algorithm yielding a change indicator. A relationship will be determined between the values of the change indicator and ground based estimates of green biomass change. This relationship will then be used to classify the image into quantitative classes of green biomass change.

The Nemadji River contributes 525,000 metric tons of sediment per year to Lake Superior. This sediment results in a number of environmental impacts to the Nemadji/Superior fluvial system. Basin hydrology has been changed through logging, burning, clearing for agriculture and road construction. These activities are hypothesized to have reduced evapotranspiration rates and to have decreased the soil's resistance to shear forces. A GIS was established to analyze landscape data for nine representative subwatersheds within the Nemadji basin. Predictor variables were examined in buffer zones on each side of the stream in each watershed. Output from the GIS was used as input to linear regression models used to quantify the relationships between the frequency of soil mass movement and landscape characteristics. Results show no significant relationships between watershed area, percent coniferous cover, percent of stream channel protected in coniferous cover, channel gradient, time of concentration, road density, mean weighted watershed slope, or number of beaver dams and frequency of mass soil movement. Significant relationships were found between slump frequency and total forested area and percent non-forested land. The results of this investigation have important implications for the maintenance and restoration of forest land in the red clay area. Keywords: mass soil wasting, forest cover, geographic information system

Early detection of plant stress has always been a priority in forestry and agriculture. Identifying and treating problems increases both food and fiber production as well as making the most of scarce resources. In the past, farmers, foresters, and gardeners have relied upon expensive electronic and/or photographic systems to detect plant stress problems before they devastate plants. This paper highlights some of the early findings on a study to create a simple detection system using specially tinted glasses and human observation.

Site-specific farming technology offers promising new farming methods that can increase the efficiency of agricultural chemicals while resulting in a safer environment. The premise of site-specific farming is that appreciable within field soil and landscape variability exists, which exerts a strong influence on factors that control yield. Recent technology is available to producers that allows for the management of these spatially variable factors. This technology includes geographical information systems mapping software to draw management unit boundaries within fields in combination with global positioning system coordinates obtained from satellite signals accepted by receivers, and variable rate application equipment. Acquisition of accurate, usable information is the core of the site-specific farming (Vanden Heuvel, 1996). A major problem with the adoption of site-specific farming technology is the cost of obtaining enough reliable fertility requirement information to base variable rate fertilization application upon. Multi-spectral satellite image analysis holds promise as a reliable, cost effective alternative to more widely used methods of delineating soil variability for site-specific application of fertilizer.

The Palouse region of eastern Washington is characterized by complex rolling hills with high erosion susceptibility. Various aspect and slope classes along with different soil types also create complex patterns in soil fertility and crop productivity. Division of fields into different units and addressing each unit as a separate management zone has been gaining importance in recent years. Landscape modeling is one of the tools that helps define management zones based on the spatial variability of the soil and topographic characteristics. In addition to comprehensive models, there is an increasing demand for simpler techniques to assist planners with field scale, day-to-day land management. The objective of this study was to develop a simple landscape model within a geographical information systems (GIS) framework to evaluate the effects of spatial variability of topographic factors on soil genesis. For this purpose, a commercial wheat farm was chosen as the research site and a digital elevation model (DEM) of the site was prepared. Landscape parameters such as slope, aspect and tangential curvature were calculated. GIS overlay of these values were georeferenced and combined with other data layers such as soil maps and air photos. Soil samples were collected on three different transects and representative pits were opened for further evaluation of soil properties. Depth to E horizon was measured for all sampling locations. Results indicate that spatial distribution of E horizon can be estimated by surface curvature, slope and aspect. Study also shows that contrasting soils that are in close proximity to each other, too close to be separated on conventional soil maps, can be detected with the help of landscape parameters. Big map units that extend over several hillslope positions can be further divided into smaller units to receive separate agricultural management based on soil, water relationships defined by these landscape parameters.

In site-specific crop management, treatments (e.g., fertilizer and herbicides) are applied precisely where they are needed. Global positioning system receivers allow accurate navigation of field implements and creation of crop yield maps. Remote sensing products help producers explain the wide range of yields shown on these maps and become the basis for digitized field management maps. Previous sources of remote sensing products for agriculture did not provide services that generated a sustained demand by crop producers, often because data were not delivered quickly enough. Public Access Resource Centers could provide a nearly uninterrupted electronic flow of data from NASA's MODIS and other sensors that could help producers and their advisors monitor crop conditions. This early warning/opportunity system would provide a low-cost way to discover conditions that merit examination on the ground. High-spatial-resolution digital aerial photographs or data from new commercial satellite companies would provide the basis for site-specific treatments. These detailed data are too expensive to acquire often and must be timed so as to represent differences in water supply characteristics and crop yield potentials. Remote sensing products must be linked to specific prescriptions that crop produces use to control operations and improve outcomes.

Researchers are expending considerable effort to develop the technology and methodology needed to identify and map within-field management zones for site-specific farming. Much of the research has focused on the use of either a high-density geographically referenced grid of soil samples or mechanical yield sensor measurements that record geographic positions and production levels. In either case, complex spatial models are generally used to extrapolate the various soil variables and production level information across the entire field. Both procedures produce a wealth of information, however, the analysis of soil samples tend to be quite expensive and the accuracy of mechanical yield measurements does vary. This study represents an ongoing effort designed to evaluate remote sensing as a tool for determining within field management zones. Color-infrared aerial photography and multispectral videography were used in concert to map and stratify two grain sorghum fields into regions or zones of homogeneous spectral response. A limited number of soil and plant samples were acquired to characterize the biotic and edaphic conditions within each zone. Results obtained during the first year of the study indicated that multispectral video can be used to develop within field management zones. Simple univariate analysis indicated that soil pH, Ca, and Fe were important variables affecting yield. Analysis of the yield data indicated that the economic returns from 17% of the first field and 20% of the second field were insufficient to recoup planting costs. Multispectral video also proved instrumental in modeling the spatial variability of yield. A significant negative correlation (r² greater than 0.90) was obtained between the red spectral band and crop yields for both fields. Stratification, in this case using image data, reduces the number of samples required to characterize a field by reducing the variance associated within each stratum. Image data also provided a comprehensive view of each field that maintained the spatial connectivity between sites, thus reducing the need for complex spatial modeling.

Center pivot (CP) irrigation systems account for more than half of all sprinkler irrigated lands in the U.S. In the Pacific Northwest, CP systems allow production of potatoes and other crops on sandy soils with variable topography. Prescription CP systems are being developed to reduce recharge to aquifers while maintaining or increasing crop yields. The hardware, software, and communication components for CP systems has been developed, installed and field tested. What is needed is the prescription for site specific applications of water, nitrogen, and pesticides. This paper discusses the role that remote sensing may have in providing some of the necessary spatial inputs to drive the irrigation models, and describes a CP prescription irrigation system that has been developed.

The utility of commercial satellite multispectral imagery for agricultural applications has traditionally been compromised by inadequate revisit times, time delays between image collection and delivery, and the imagery's spatial resolution. With the launch, in December 1997, of Space Imaging's remote sensing satellite, four-band multispectral imagery with resolutions as high as 4 meters, revisit times as frequent as once per day, and routine delivery times of as little as 48 hours will be commercially available. This new imagery, with it improved timeliness of information to the grower, promises to provide a valuable new information source for crop monitoring and distribution mapping.

Crop/weed/soil discrimination by optical reflectance on brassicas proved to be reliable for recognizing leaves of crops from leaves of weeds and soil, although differences on crop varieties, weed species population and type of soil. Radiometers using 10 nm bandwidth filters were used in earlier experiments, but higher reflected radiance could only be collected with the use of broader filtering. The discrimination integration index (DII), presented in this paper, is a broadband discriminator which allows us to study the effect of broader spectral bands on corp/weed/soil discrimination. Some results on cabbage/weed-soil discrimination using DII's for several wavebands are presented.

Plant species discrimination by optical reflectance requires us to analyze reflectance spectral differences between two sampled broad band signatures making use of the discriminative integration index, DII. Crop/weed discrimination analyses of these indexes using a crop threshold introduces the discrimination efficiency of the entire vegetation population consisting of crop and weeds. Best discriminative efficiencies are spectral dependant and results for leek, cabbage, turnip and potato crops having competence with ten different weed species are presented. With the reflectance of an unique wavelength found iteratively, it was possible to reproduce the NIR spectrum of the crop or weed. This reflectance was multiplied by the crop reference spectrum, producing a pattern similar to the real crop signature and for the four crops tested the control wavelength corresponded to the one having the peak reflectance at the red region. This algorithm is the heart of a new model used to distinguish crops from weeds and the results achieved per group are presented.

Precision agriculture, an information and technology based agricultural system, will advance the application of remote sensing techniques to agricultural management. There are specific and unique technical and managerial barriers and requirements for the application of remote sensing to soil and corp management. The four principal requirements relate to: spatial resolution, timeliness, coverage frequency, and imagery management infrastructure.

Modern agriculture management is an extraordinarily complex task. The most complex tasks are management for environmental benefits. Chemical, physical and biological characteristics are known to vary over short distances in a field. However, most fields are treated as uniform, leading to over application and environmental pollution, or under application and suboptimal yields. Affordable navigation and positioning systems linked to sensing technologies and integrated into a geographic information system (GIS) are revolutionizing the way agriculture can address environmental variabilities. One challenge to better management of within field variability is the establishment of management zones for various inputs. Our research and development group is currently using aerial acquired images to help establish management zones for nutrients, pest scouting, and to monitor crop growth and development. These images are ground truthed and coupled with additional information layers such as maps of yield, disease, insect and weed pests, soil properties, topography to help establish relationships between the various components affecting crop growth and to help improve management decisions during the growing season.

The main idea of pan sharpening is to combine registered high spatial resolution panchromatic (pan) and lower spatial resolution multispectral imagery to synthesize higher resolution multispectral imagery. The degree of misregistration between bands and the correlation of intensity values are key factors in generating radiometrically accurate and visually crisp output images. There are two components to the projective pan sharpening algorithm. The first component determines the linear radiometric relationship between the pan and multispectral bands and the second processes pixels to produce the sharpened product. The projective pan sharpening algorithm will be ideally suited for generating multispectral products using Lockheed Martin Commercial Remote Sensing System (CRSS) imagery.

During the standard geo-rectification processing of MISR imagery, all four spectral bands belonging to each of the nine MISR cameras are required to be geolocated and co- registered automatically to approximately one pixel accuracy. Two steps of processing are designed to accomplish this goal: (1) a complex multi-camera geolocation and co- registration of the red spectral band data for all nine cameras, and (2) the co-registration of the other three spectral bands of MISR imagery of each camera using their relationship with the already geolocated red band imagery. This paper addresses the second processing step. The geometry of the satellite orbit, the ellipsoid rotating earth, and the separation of the view angles between different spectral bands are combined in a mathematical model which describes the band-to-band line and sample parallaxes. The sensitivity study of this model to numerous error sources, such as variations in the orbit and earth radius, orbit perturbation, and navigation errors, leads to a practical polynomial band-to-band transform solution, and the decision on the usage of either a static or dynamic band-to-band transform as well as the application range of the transform.

Road intersections are very good ground control points in a SAR image because they can be obtained from intersecting road segments even when the intersections themselves are not observed. In a previous study, a Hough transform-based approach was developed for detecting intersections from a single-look ERS-1 SAR image. A similar Hough transform-based approach has been applied to detect road intersections in a 6-look ERS-1 SAR image. In the case of multi-look images, however, there are no pure road pixels because roads are thinner than the resolution size and the pixels that include road information are greatly affected by surrounding objects. More spatial information and even approximate road map information are required to improve the capability to detect road intersections. It is found that broad intersections ca be automatically determined from a multi- look image and the accuracy of the location of road intersections may be estimated from the heights of sharp peaks in Hough space.

The results of a study of Railroad Valley Playa, located 100 miles east of Tonopah, Nevada are presented. The primary objective of this study has been to determine its suitability as a satellite calibration target. We present the search algorithm that was used initially to identify the site from satellite imagery. The search program mainly focuses on locating areas in imagery that have large spatial extent, spatial uniformity, and high surface reflectance. We present the results of several field campaigns to the site to further examine the playa's surface characteristics. These studies include spectral surface reflectance over the 0.4 to 2.0 micron range, sample bi-directional reflectance characteristics, and the spatial uniformity of the overall playa. The target is intended to be used for post-launch radiometric calibration using the reflectance-based approach for small footprint senors. It also will be used for a cross-calibration approach where one sensor's calibration is transferred to another sensor. We present preliminary error estimates in these two methods from using the Railroad Valley site.

We present a practical approach for detecting and localizing clouds in satellite remote sensing images. Cloud detection is useful in improving the accuracy of land cover classification when there are clouds present in the images. After detection and removal of clouds we can selectively merge classification results from two temporally separate images of the same area to minimize the cloud effect. We emphasize the ease of implementation of the algorithm so that practitioners can easily adapt the method for their own use

With the advent of one or two dimensional focal plane arrays operating in 8 - 12 micrometer band, advanced second generation IR imaging systems are becoming possible. These systems can generally provide higher spatial resolutions and higher temperature sensitivity and are likely to be quite compact and light weight. A high performance thermal imager has been designed as a part of dual use night vision technology keeping in view medical and other applications. The present paper describes the trends and advances in imaging systems using second generation focal plane arrays (FPAs) operating in longwave infrared region. The high performance thermal imager has been designed at our laboratory using 288 multiplied by 4 linear MCT focal plane array received from SOFRADIR, France. The system has a twin magnification giving FOV of 8 degrees multiplied by 4.2 degrees and 2.7 degrees multiplied by 1.4 degrees using diffraction limited f/2 system having maximum optics aperture of 165 mm, yielding a systems NETD of better than 0.1 degree Celsius. The paper focuses attention on the optimization of variable afocal telescopes, optics aperture, FOVs and the MRTD. An experimental scheme of fixed pattern noise compensation in real time using two servo controlled standard reference sources, has been worked out. The system has been designed on modular approach so that, just by changing the front afocal telescope, system could be used for both civilian as well as military applications.

The interpretation of airphotos relies heavily on the identification of textures and spatial patterns. The 2-D wavelet transform can be used to quantify simple patterns for automated classification of pixels in an image. The transform generates a set of images similar in format to a multispectral image deck, but based on spatial information localized about each pixel. The images in this 'wavelet' deck each correspond to an analysis of patterns in the original image at different spatial resolutions. For example, where one image in the wavelet deck depends on spatial variations within a 32 by 32 pixel area, the next image in the deck contains information on spatial features within a 16 by 16 pixel area. Individual pixels in the wavelet deck can be classified using the same classification and pattern recognition algorithms used with multispectral images. Classifications based on the wavelet deck of four airphoto samples using a minimum-distance to means algorithm and an artificial neural network are presented.

Image-intensification has been improved constantly by enhancing the gains of image-intensification more and more, to achieve more sensitive imaging at low light-level. When the sensitivity needs for night vision have been satisfied, further improvement should be the enhancement of night vision image information quantity. On the basis of analysis of information transfer of an image-intensification system, this paper shows that the image transfer from the output screen of the intensifier to eye makes up limitation to the whole system information quantity, and proposes that color image-intensification utilizing spectral gate and color assignment, in the case of that the gain redundancy of image intensifier can compensate adequately spectral gate attenuation, will realize larger information quantity than conventional image-intensification. Then, give out a color image-intensifier scheme combining time-shared photocathode and spatial multiplexed screen.

In this paper, we propose two detail-preserving segmentation algorithms for multiband images and evaluate their performance using real multi-frequency multi-polarization SAR complex data. Adaptive neighborhood structures are selected for modeling the polarimetric complex amplitudes and the region labels, and for achieving detail- preservation. Experimental results obtained from multi-band and multi-polarization SIR-C data, selected for archaeological applications studies, show that the novel schemes produce significant visual improvements for detail preservation, and exhibit equivalent or higher classification performance with respect to the classical classification schemes.

The integration of global positioning system (GPS) technology in earth observation activities has become a fact of technological life. Not only is georeferencing of remotely sensed imagery greatly facilitated, but ground reference data delineation for classification and accuracy assessment purposes has been precision-enhanced to previously unheard-of levels. However, the acquisition of GPS locational information is sensitive to both instrument specifications and procedural variations. The goal of this study was to compare the efficiency and accuracy of two mapping-grade GPS systems under different data acquisition scenarios. The Trimble Basic Plus and ProXL receivers were tested with their respective base stations in conjunction with a range of receiver data collection time intervals. For each set of operational parameters, precision of the acquired position is represented by the standard deviation, while the root mean square error (RMS) depicts its accuracy. We conclude that, even with the data collection session reduced to one minute, the Trimble ProXL is truly a sub- meter GPS system. In the case of the Trimble Basic Plus, a session of three minutes is recommended to obtain or even slightly exceed an accuracy of 2.5 meters. The easting, northing, and altitude precision ratio obtained during the tests in Indiana, central USA, is of a magnitude of 1.0:1.5:3.0 in both cases. The effect of distance between base station and portable receives and of environmental factors (specifically in the forestry sector) is the subject of continuing research.

Improved methods for calibrating and characterizing the CCD array based off-nadir tiltable advanced solid-state array spectroradiometer (ASAS) were developed and applied. Sensor characteristics such as radiometric sensitivity, polarization sensitivity, signal-to-noise ratio, temperature sensitivity, spectral bandpass, spectral distortion, and spatial resolution were measured. Radiometric sensitivity, array temperature sensitivity, and signal-to-noise were measured using a barium sulfate coated integrating hemisphere whose output calibration is traceable to NIST. Polarization sensitivity was measured for 62 spectral bands across all 512 spatial pixels (31 K pixels). Spectral bandpass and spectral distortion were measured using a 0.5 meter doublepass monochromator. Spatial resolution (given as the modulation-transfer-function -- MTF) and distortion were measured using a combination of monochromatic collimated light to directly measure the point-spread-function (PSF) and the edge spread function (ESF) derived from actual image data. The MTF obtain using the two techniques are compared. Potential improvements to the test setups and methods are described.