Timothy B. Minor

Hyperspectral mixture modeling for quantifying sparse vegetation cover in arid environments.

Abstract A linear mixture model based on calibrated, atmospherically corrected Probe-1 hyperspectral imagery was compared with three vegetation indices to test its relative ability to measure small differences in percent green vegetative cover for areas of sparse vegetation in arid environments. The goal of this research was to compare multispectral and hyperspectral remote sensing approaches for detecting human disturbance of arid environments. The normalized difference vegetation index (NDVI) was tested using both narrow and broad bandwidths. Broadband NDVI provided results r 2 =0.63 similar to NDVI derived from individual hyperspectral channels r 2 =0.60 . While the soil-adjusted vegetation index (SAVI) was designed as an improvement to NDVI for sparse vegetation, in this study SAVI performed significantly worse than NDVI r 2 =0.51 . The modified soil-adjusted vegetation index (MSAVI) provided an insignificant improvement over NDVI r 2 =0.64 . Linear mixture modeling provided significantly better results, r2 of 0.74. Cross-validation was used to test the significance of differences between the various methods and to determine the standard error associated with each method. Results suggest that any improvements provided by adjusted vegetation indices over NDVI may be strongly dependent on those adjustments being derived from local conditions. The use of a linear mixture model with multiple soil endmembers appears to provide the best method for quantifying sparse vegetative cover. Though present in small amounts, a single plant species, Krameria erecta, was strongly correlated with residuals of the mixture model. Inclusion of a spectral endmember for this species increased the r2 of the fit with percent green cover to 0.86. However, it is not clear if the explained variation was actually due to K. erecta or a correlated phenomena. Problems were also identified with the use of multiple vegetation endmembers.

Detecting and discriminating impervious cover with high-resolution IKONOS data using principal component analysis and morphological operators

The objective of this study was to directly detect impervious cover using new high-resolution IKONOS imagery in South Lake Tahoe, California, USA. The research presented was a pilot analysis to assess the ability of satellite imagery to derive accurate estimates of impervious cover, critical for assessing impacts to water quality, wildlife, and fish habitat. A combination of image processing methods based on principal component analysis and spatial morphological operators was developed for a 25 km2 urban area in the Lake Tahoe Basin. The methodology produced very accurate identification of both commercial and residential impervious cover in an area dominated by dense conifer canopy. Sub-canopy and sub-shadow surfaces were not only detectable, but also discernible with respect to the underlying substrate. An overall accuracy of 92.94% was obtained, with an even higher user accuracy of 95.83% based on 170 ground truth points. Impervious cover is a difficult feature to delineate accurately and efficiently using direct methods. For this application, spatial resolution proved a better operator than spectral resolution. Results from this analysis will be used to better understand the impacts of urban development on the ecology of the Lake Tahoe Basin.

Evaluating Change in Rangeland Condition using Multitemporal AVHRR Data and Geographic Information System Analysis

Coarse-scale, multitemporal satellite image data were evaluated as a tool for detecting variation in vegetation productivity, as a potential indicator of change in rangeland condition in the western U.S. The conterminous U.S. Advanced Very High Resolution Radiometer (AVHRR) biweekly composite data set was employed using the six-year time series 1989–1994. Normalized Difference Vegetation Index (NDVI) image bands for the state of New Mexico were imported into a Geographic Information System (GIS) for analysis with other spatial data sets. Averaged NDVI was calculated for each year, and a series of regression analyses were performed using one year as the baseline. Residuals from the regression line indicated 14 significant areas of NDVI change: two with lower NDVI, and 11 with higher NDVI. Rangeland management changes, cross-country military training activities, and increases in irrigated cropland were among the identified causes of change.

Classification of hyperspectral imagery with neural networks: comparison to conventional tools

Efficient exploitation of hyperspectral imagery is of great importance in remote sensing. Artificial intelligence approaches have been receiving favorable reviews for classification of hyperspectral data because the complexity of such data challenges the limitations of many conventional methods. Artificial neural networks (ANNs) were shown to outperform traditional classifiers in many situations. However, studies that use the full spectral dimensionality of hyperspectral images to classify a large number of surface covers are scarce if non-existent. We advocate the need for methods that can handle the full dimensionality and a large number of classes to retain the discovery potential and the ability to discriminate classes with subtle spectral differences. We demonstrate that such a method exists in the family of ANNs. We compare the maximum likelihood, Mahalonobis distance, minimum distance, spectral angle mapper, and a hybrid ANN classifier for real hyperspectral AVIRIS data, using the full spectral resolution to map 23 cover types and using a small training set. Rigorous evaluation of the classification accuracies shows that the ANN outperforms the other methods and achieves ≈90% accuracy on test data.

Predictive Soil Maps Based on Geomorphic Mapping, Remote Sensing, and Soil Databases in the Desert Southwest

We present an expert based system to rapidly predict the shallow soil attributes that control dust emissions in the arid southwest U.S. Our system’s framework integrates geomorphic mapping, remote sensing, and the assignment of soil properties to geomorphic map units using a soil database within a geographic information systems (GIS) framework. This expert based system is based on soil state factor-forming model parameters that include: (1) climate data, (2) landform, (3) parent material, and (4) soil age. The four soil-forming data layers are integrated together to query the soil database. To validate the accuracy of the expert based model and resultant predictive soil map, a blind test was performed at Cadiz Valley in the Mojave Desert, California. The desert terrain in Cadiz Valley consists of alluvial fans, fan remnants, sand dunes, and playa features. The test began with three users independently mapping an area of over 335 km2 using 1:40,000-scale base maps to rapidly create geomorphic and age class layers, and then integrating these with climate and parent material layers. The results of the four data layers were then queried in the soil data base and soil attributes assigned to map unit layers. The soil-forming model presented here is geomorphic-based, and considers soil age as a significant factor in accurately predicting soil conditions in hyper arid to mildly arid regions. This work comprises a successful first step in the development of an expert-based system to map shallow soil conditions in support of dust emission models in remote desert regions.

Progressive Discrimination: An Automatic Method for Mapping Individual Targets in Hyperspectral Imagery

This paper demonstrates a new method called progressive discrimination (PD) for mapping an individual spectral class within an image. Given training data for a target, PD iteratively samples nontarget image pixels using a collapsing distance threshold within the space of an evolving discriminant function. This has the effect of progressively isolating the target class from similar spectra in the image. PD was compared to Bayesian maximum likelihood classification, mixture-tuned matched filtering, spectral angle mapping, and support vector machine methods for mapping three different invasive species in two types of high-spatial-resolution airborne hyperspectral imagery, AVIRIS and AISA. When tested with 20 different randomly selected groups of training fields, PD classification accuracies for the two spectrally distinct plant species in these images had an average of 98% and a standard deviation of 1%. These randomized trials were capable of providing higher classification accuracies than the best results obtained by two expert analysts using existing methods. For the third species that was less distinct, PD results were comparable to the results obtained by experienced analysts with existing methods. Despite requiring less input from the user than many techniques, PD provided more consistent high mapping accuracy, making it an ideal tool for scientists and land use managers who are not trained in image processing.

Integrated Terrain Forecasting for Military Operations in Deserts: Geologic Basis for Rapid Predictive Mapping of Soils and Terrain Features

During the past three decades, the U.S. armed forces have been called on repeatedly to operate in the deserts of the Middle East and southwest Asia. Avoiding locations susceptible to extreme dust emissions and other terrain-related hazards requires the ability to predict soil and terrain conditions, often from limited information and under dynamic environmental conditions. This paper reports the approach used to develop an integrated, predictive tool for forecasting terrain conditions to support military operations in desert environments at strategic, operational, and tactical scales. The technical approach relies on the systematic integration of desert landform parameters in geomorphic models for predicting terrain conditions. This integrated effort is performed in a geographic information system (GIS) framework using expert-based analysis of airborne and spaceborne imagery to identify terrain elements. Advances in earth science research have established that unique, predictable relations exist among landscape position, soils, vegetation, and geology. Furthermore, new instrumentation allows the collection of a wide range of environmental information to characterize surface and subsurface conditions. By integrating models and methods from geomorphology, soil science, climatology, and atmospheric science with remote sensing and other technologies, a predictive model can be developed to support military operations.

Assessment of wind energy for Nevada using towers and mesoscale modeling

The main objectives of the study were to support wind energy assessment for all of Nevada by providing two annual cycles of high-resolution mesoscale modeling evaluated by data from surface stations and towers, estimating differences between these annual cycles and standard wind maps, and providing wind and wind power density statistics at elevations relevant to turbine operations. In addition to the 65 existing Remote Automated Weather Stations in Nevada, four 50-m-tall meteorological towers were deployed in western Nevada to capture long-term wind characteristics and provide database input to verify and improve modeling results. The modeling methodology using Mesoscale Model 5 (MM5) was developed to provide wind and wind power density estimates representing mesoscale effects that include actual synoptic forcing during the two annual cycles (horizontal resolution on the order of 2 and 3 km). The results from the two annual simulation cycles show similar wind statistics with an average difference of less than 100 W/m2 . The available TrueWind results for the wind power density at 50 m show greater values of wind power density compared to both MM5-simulated annual cycles for most of the area. However, mainly in the Sierras and the mountainous regions of southern and eastern Nevada, the MM5 simulations indicate greater values for wind power density. The results of this study suggest that the synthesis of the data from a network of tower observations and high-resolution mesoscale modeling is a crucial tool for assessing the wind power density in Nevada and, more generally, other topographically developed areas.© 2007 ASME

Parent Material Mapping of Geologic Surfaces Using ASTER in Support of Integrated Terrain Forecasting for Military Operations

Predicting soil physical and chemical properties for military operations requires knowledge of the geologic and lithologic component of the soil parent material. Geologic maps, a traditional source of geologic information, are often limited in coverage or inadequate for determining the basic characteristics of a soil parent material. We describe an approach for the rapid development of geologic surface maps that identify the lithologic composition of soil parent material generated from ASTER (Advanced Spaceborne Thermal Emission and Reflection Radiometer) data. Generated maps of parent material, in turn, provide key input parameters for a comprehensive terrain predictive model that forecasts key soil and surface cover characteristics in support of military operations. Parent material maps are generated using a multilayer approach where calibrated image data are mapped into lithologic units that best identify soil parent material and corresponding landform units (i.e. bedrock, fan, playa, dune, etc.). A unique and critical aspect of our approach is that expert-based analysis of spectral and geospatial information can produce a geologic map, covering 1000–5000 km2 of terrain, of soil parent material and surface cover in as little time as nine staff-hours. The approach was developed with a guiding principle that terrain predictions in military operations must be rapidly developed for areas where available ground information is limited. Results indicate that it is possible to quickly produce a realistic map of soil parent material using ASTER data without any additional geologic information or data. Results also indicate that analysts developing parent material maps require expert knowledge in both spectral analysis of remotely sensed data and the geologic and geomorphic processes that form desert landforms.

Final Report for the MANNRRSS II Program Management of Nevada's Natural Resources with Remote Sensing Systems, Beatty, NV

This document provides the Final Report on the Management of Nevada’s Natural Resources with Remote Sensing Systems (MANNRRSS) II program. This is a U.S. Department of Energy (DOE)-funded project tasked with utilizing hyperspectral and ancillary electro-optical instrumentation data to create an environmental characterization of an area directly adjacent to the Nevada Test Site (NTS).