Search

Many countries have adopted large-scale tree planting programmes as a climate mitigation strategy and to support local livelihoods. We evaluate a series of large-scale tree planting programmes using data collected from historical Landsat imagery in the state of Himachal Pradesh in Northern India. Using this panel dataset, we use an event study design to estimate the socioeconomic and biophysical impacts over decades of these programmes. We find that tree plantings have not, on average, increased the proportion of forest canopy cover and have modestly shifted forest composition away from the broadleaf varieties valued by local people. Further cross-sectional analysis, from a household livelihood survey, shows that tree planting supports little direct use by local people. We conclude that decades of expensive tree planting programmes in this region have not proved effective. This result suggests that large-scale tree planting may sometimes fail to achieve its climate mitigation and livelihood goals. Large-scale tree planting programmes have been implemented or planned for areas around the world suffering from deforestation, but this study presents evidence that such efforts may not necessarily deliver the desired environmental and economic outcomes.

Classification is a critical, widely employed type of hyperspectral image (HSI) processing. Recently, deep learning models have been attracting more attention within the hyperspectral remote-sensing community due to their improved classification performance. Among them, recurrent neural networks (RNNs), which were initially used to handle sequential data, have been applied to HSI classification with promising results. The key point for such RNN-based models in a classification context is the extraction of a sequential feature for each individual pixel in a HSI. One popular strategy is to first extract similar pixels compared with a target pixel from the HSI, and then use those similar pixels to encode its sequential feature. However, the computational cost is tremendous, especially if such similarity-calculation search is done on the whole image. In this article, inspired by our previous work regarding similarity measurement-based sequential feature construction, a faster sequential feature extraction framework for long short-term memory (LSTM)-based HSI classification is proposed, where object-based segmentation method is employed for the purpose of imposing spatial constraints and computational acceleration. Within the proposed framework, both the local segment containing the target pixel and nonlocal segments are considered. For a target pixel, similar segments are selected first based on segment-based features, and then similar pixels from selected segments are extracted to construct a sequential feature. During pixel-wise similarity measurement, both spectral and spatial information are considered in such computation. Experimental results on three benchmark HSI data sets illustrate that the proposed methods achieve promising classification performance with markedly lower computation-time cost.

Modern precision weed management relies on site-specific management tactics to maximize resource use efficiency and yield, while reducing unintended environmental impacts caused by herbicides. Scouting for weeds is an important activity to assist weed management decision making, and has been carried out by trained specialists through extensive and routine visual examination of the fields. Recent advancements in Unmanned Aircraft Systems (UAS)-based tools and geospatial information technology have created enormous applications for efficient and economical assessment of weed infestations as well as site-specific weed management. The utilization of UAS-based technologies for weed management applications is currently in its infancy, but this field has witnessed rapid growth in recent times in terms of aerial data acquisition and analysis. Challenges exist in UAS platform reliability, sensor capability and integration, image pre-processing, quantitative assessment and prediction, final product development, and product delivery. This review summarizes current knowledge on the utility of UAS platforms and remote sensing tools for weed scouting and precision weed management. Further, it critically examines potential opportunities and limitations to current UAS technologies, with particular emphasis on the lessons learned from UAS-based weed management research conducted at Texas A&M University.

A fundamental premise of the Millennium Ecosystem Assessment is that biodiversity and ecosystem services are key determinants of long-term sustainability of social-ecological systems. With a continuing decline in local and global biodiversity and ecosystem services, it is crucial to understand how biodiversity and various ecosystem services interact and how land change may modify these interactions over time. However, few studies have been conducted to quantify these relationships. In this study, we present the first empirical comparative results to analyze how spatial associations between biodiversity and ecosystem services (BES) changed at multiple scales between 1984 and 2010 in the rapidly urbanizing San Antonio River Basin (SARB), Texas, USA. We found statistically significant positive spatial associations among biodiversity, carbon storage, and sediment retention both in the entire SARB and the urban watersheds in Bexar County. Overall, biodiversity and carbon storage declined across the SARB, while sediment retention remained relatively stable. Moreover, the rates of biodiversity loss and carbon storage degradation were negatively related to the urban expansion and have accelerated since the inception of the North American Free Trade Agreement (NAFTA) in 1994. During the pre- and post-NAFTA periods (1984-1995 and 1995-2010, respectively) the rates of biodiversity loss increased from 0.7% to 0.9%, and the rates of carbon-storage loss increased from 0.1% to 1.4% per annum in the urban watersheds. Our hotspot analyses indicate that the upstream watersheds in the Basin, which supply water to the critically important Edwards Aquifer, should be targeted for priority conservation to mitigate the adverse impacts of land change on BES. Our results suggest the strong need for green infrastructure policies that integrate biodiversity conservation and sustainable use of multiple ecosystem services to address the environmentally deleterious impacts of the extensive land change under the NAFTA and to ensure the long-term social-ecological sustainability of the rapidly urbanizing SARB.

The San Antonio River Basin (SARB) is an ecologically diverse region in South Texas. The city of San Antonio is located within the basin and is the hub of the North American Free Trade Agreement (NAFTA). San Antonio, together with other major metropolitan centers in Texas, has experienced rapid population and economic growth over the last thirty years, which accelerated after the implementation of NAFTA in 1994. To assess the environmental implications of this growth in the SARB, we first conducted a land-change analysis using Landsat images from 1984, 1995, and 2010. Then, we analyzed spatiotemporal changes in ecosystem services across the SARB and within three watersheds in Bexar County where the city of San Antonio is located. To estimate changes in ecosystem service values (ESV) during this period, we combined the results of the land-change analysis with a benefit transfer approach using two sets of widely cited ecosystem-service valuation coefficients published in 1997 and 2014 but we modified the urban coefficient from the 2014 publication for low-density and high-density urban areas. When 1997 coefficients were applied, the ESV in the SARB decreased, on average, by $1.2 million/year during 1984–1995 and by $1.8 million/year during 1995–2010. The ESV in Bexar County decreased, on average, by $0.5 million/year and $0.7 million/year during the first and second periods, respectively. When the 2014 coefficients and modified urban value coefficients were applied, the ESV in the SARB decreased, on average, by a 27% more during the first period than when the 1997 coefficients were applied, while, ESV increased during the second period by an average of $2.2 million/year. This temporally opposite trend in ESV change did not occur in Bexar County, however. Using the 2014 coefficients, ESV in Bexar County decreased 5 times more during the first period and decreased 2.5 times more during the second period than when 1997 coefficients were applied. The differences in ESV trends resulting from the two sets of coefficients can be explained primarily by the different coefficients assigned to urban spaces ($0/ha/year in the 1997 study and $7005/ha/year in the 2014 study). Our results suggest that the value placed on urban areas in the 2014 publication, taken from a single case study and intended primarily for large urban parks, substantially overestimates the ESV of urban space. In our study areas, applying this value, even only to urban green space, led to the improbable conclusion that urbanization had a positive overall effect on the delivery of ecosystem services. While open spaces in urban areas do provide valuable ecosystem services, it is highly unlikely that their value exceeds those provided by less modified ecosystems. The ability to confidently use value coefficients when applying benefit transfer methods to estimate ESVs demands rigorous assessments of their broad applicability.

A simple approach to enable water-management agencies employing free data to create a single set of water quality predictive equations with satisfactory accuracy is proposed. Multiple regression-derived equations based on surface reflectance, band ratios, and environmental factors as predictor variables for concentrations of Total Suspended Solids (TSS) and Total Nitrogen (TN) were derived using a hybrid forward-selection method that considers both p-value and Variance Inflation Factor (VIF) in the forward-selection process. Landsat TM, ETM+, and OLI/TIRS images were jointly utilized with environmental factors, such as wind speed and water surface temperature, to derive the single set of equations. Through splitting data into calibration and validation groups, the coefficients of determination are 0.73 for TSS calibration and 0.70 for TSS validation, respectively. The coefficients of determination for TN calibration and validation are 0.64 and 0.37, respectively. Among all chosen predictor variables, ratio of reflectance of visible red (Band 3 for Landsat TM and ETM+, or Band 4 for Landsat OLI/TIRS) to visible blue (Band 1 for Landsat TM and ETM+, or Band 2 for Landsat OLI/TIRS) has a strong influence on the predictive power for TSS retrieval. Environmental factors including wind speed, remote sensing-derived water surface temperature, and time difference (in days) between the image acquisition and water sampling were found to be important in water-quality quantity estimation. The hybrid forward-selection method consistently yielded higher validation accuracy than that of the conventional forward-selection approach.

Floodplain forests serve a critical function in the global carbon cycle because floodplains constitute an important carbon sink compared with other terrestrial ecosystems. Forests on dynamic floodplain landscapes, such as those created by river meandering processes, are characterized by uneven-aged trees and exhibit high spatial variability, reflecting the influence of interacting fluvial, hydrological, and ecological processes. Detailed and accurate mapping of aboveground biomass (AGB) on floodplain landscapes characterized by uneven-aged forests is critical for improving estimates of floodplain-forest carbon pools, which is useful for greenhouse gas (GHG) life cycle assessment. It would also help improve our process understanding of biomorphodynamics of river-floodplain systems, as well as planning and monitoring of conservation, restoration, and management of riverine ecosystems. Using stochastic gradient boosting (SGB), multivariate adaptive regression splines (MARS), and Cubist, we remotely estimate AGB of a bottomland hardwood forest on a meander bend of a dynamic lowland river. As predictors, we use 30-m and 10-m multispectral image bands (Landsat 7 ETM+ and SPOT 5, respectively) and ancillary data. Our findings show that SGB and MARS significantly outperform Cubist, which is used for U.S. national-scale forest biomass mapping. Across all data-experiments and algorithms, at 10-m spatial resolution, SGB yields the best estimates (RMSE = 22.49 tonnes/ha; coefficient of determination ( R 2 ) = 0.96) when geomorphometric data are also included. On the other hand, at 30-m spatial resolution, MARS yields the best estimates (RMSE = 29.2 tonnes/ha; R 2 = 0.94) when image-derived data are also included. By enabling more accurate AGB mapping of floodplains characterized by uneven-aged forests, SGB and MARS provide an avenue for improving operational estimates of AGB and carbon at local, regional/continental, and global scales.

A simple approach to enable water-management agencies employing free data to achieve the goal of using a single set of predictive equations for water-quality retrievals with satisfactory accuracy is proposed. Multiple regression-derived equations based on surface reflectance, band ratios, and environmental factors as predictor variables for concentrations of Total Suspended Solids (TSS), Total Nitrogen (TN), and Total Phosphorus (TP) were derived using a hybrid forward-selection method that considers Variance Inflation Factor (VIF) in the forward-selection process. Landsat TM, ETM+, and OLI/TIRS images were jointly utilized with environmental factors, such as wind speed and water surface temperature, to derive the single set of equations. The coefficients of determination of the best-fitting resultant equations varied from 0.62 to 0.79. Among all chosen predictor variables, ratio of reflectance of visible red (Band 3 for Landsat TM and ETM+, or Band 4 for Landsat OLI/TIRS) to visible blue (Band 1 for Landsat TM and ETM+, or Band 2 for Landsat OLI/TIRS) has a strong influence on the predictive power for TSS retrieval. Environmental factors including wind speed, remote sensing-derived water surface temperature, solar altitude, and time difference (in days) between the image acquisition and water sampling were found important in water-quality parameter estimation.

Research on aboveground biomass (AGB) retrieval via remote sensing in floodplain forests, in particular, is urgently needed for improved understanding of carbon cycling in such areas. AGB estimation is particularly challenging in floodplain forests, which are characterized by high spatial variability in AGB resulting from biogeomorphodynamic processes. In this study, we perform remote AGB retrieval for a deciduous riparian forest on a river meander bend based on hyperspectral/high-dimensional Hyperion bands and other input variables. We compare multivariate adaptive regression splines (MARS)-, stochastic gradient boosting (SGB)- and Cubist-based AGB estimates. Results show that MARS- and SGB-derived estimates are significantly more accurate than Cubist-based AGB. The most accurate MARS and SGB estimates have a coefficient of determination, R2, of 0.97 and 0.95, respectively, whereas the Cubist estimate with the lowest error has an R2 of 0.85. MARS and SGB AGB are not significantly different, however. Thes...

Accurate mapping of water surface boundaries in rivers is an important step for monitoring water stages, estimating discharge, flood extent, and geomorphic response to changing hydrologic conditions, and assessing riverine habitat. Nonetheless, it is a challenging task in spatially and spectrally heterogeneous river environments, commonly characterized by high spatiotemporal variations in morphology, bed material, and bank cover. In this study, we investigate the influence of channel morphology and bank characteristics on the delineation of water surface boundaries in rivers using high spatial resolution passive remote sensing and a template-matching (object-based) algorithm, and compare its efficacy with that of Support Vector Machine (SVM) (pixel-based) algorithm. We perform a detailed quantitative evaluation of boundary-delineation accuracy using spatially explicit error maps in tandem with the spatial maps of geomorphic and bank classes. Results show that template matching is more successful than SVM in delineating water surface boundaries in river sections with spatially challenging geomorphic landforms (e.g. sediment bar structures, partially submerged sediment deposits) and shallow water conditions. However, overall delineation accuracy by SVM is higher than that of template matching (without iterative hierarchical learning). Vegetation and water indices, especially when combined with texture information, improve the accuracy of template matching, for example, in river sections with overhanging trees and shadows – the two most problematic conditions in water surface boundary delineation. By identifying the influence of channel morphology and bank characteristics on water surface boundary mapping, this study helps determine river sections with higher uncertainty in delineation. In turn, the most suitable methods and data sets can be selectively utilized to improve geomorphic/hydraulic characterization. The methodology developed here can also be applied to similar studies on other geomorphic landforms including floodplains, wetlands, lakes, and coastlines. Copyright © 2014 John Wiley & Sons, Ltd.

Optical remote sensing imagery offers a cost-effective alternative to echo sounding and bathymetric light detection and ranging surveys for deriving high density bottom depth estimates for coastal and inland water bodies. The common practice of previous studies has been to calibrate a single global bathymetric inversion model for an entire image scene. The performance of conventional global models is limited when the bottom type and water quality vary spatially within the scene. To address the inadequacy of the conventional global models, this paper presents a geographically adaptive inversion model to better estimate bottom depth. Although the general mathematical form of the geographically adaptive model is the same, model parameters are optimally determined within a geographical region or a local area, in contrast to the entire scene in the global inversion model. By using high-resolution IKONOS and moderate-resolution Landsat satellite images, we demonstrated that regionally and locally calibrated inversion models can effectively address the problems introduced by spatial heterogeneity in water quality and bottom type, and provide significantly improved bathymetric estimates for more complex coastal waters.

Hyperspectral images (HSIs) contain hundreds of spectral bands, providing high-resolution spectral information pertaining to the Earth’s surface. Additionally, abundant spatial contextual information can also be obtained simultaneously from a HSI. To characterize the properties of ground objects, classification is the most widely-used technology in the field of remote sensing, where each pixel in a HSI is assigned to a pre-defined class. Over the past decade, deep learning has attracted increasing attention in the machine-learning and computer-vision domains, due to its favourable performances for various types of tasks, and it has been successfully introduced to the remote-sensing community. Instead of utilizing the shallow features within in a given image, which is the approach that is generally adopted in other conventional classification methods, deep-learning algorithms can extract hierarchical features from raw HSI data. Within the deep-learning framework, recurrent neural networks (RNNs), which are able to encode sequential features, have exhibited promising capabilities and have achieved encouraging performances, especially for the natural-language processing and speech-recognition communities. As multi-temporal remote-sensing images can be readily obtained from increasing numbers of satellite and unmanned aircraft systems, and since analysis of such multi-temporal data comprises a critical issue within numerous research subfields, including land-cover and land-change analyses, and land-resource management, RNNs have been applied in recent studies in order to extract temporal sequential features from multi-temporal remote-sensing images for the purpose of image classification. Apart from using multi-temporal image datasets, RNNs can also be utilized on a single image, where the spectral feature/band of each individual pixel can be taken as a sequential feature for the input layer of RNNs. However, the application of such sequential feature extraction that relies on a single image still needs to be further investigated since applying RNNs to spectral bands will directly introduce more parameters that need to be optimized, consequently increasing the total training time.In this study, we propose a novel RNN-based HSI classification framework. In this framework, unlabelled pixels obtained from a single image are considered when constructing sequential features. Two spatial similarity measurements, referred to as pixel-matching and block-matching, respectively, are employed to extract pixels that are “similar” to the target pixel. Then, the sequential feature of the target pixel is constructed by exploiting several of the most “similar” pixels and ordering them based on their similarities to the target pixel. The aforementioned two schemes are advantageous, as unlabelled pixels within the given HSI are taken into consideration for similarity measurement and sequential feature construction for the RNN model. Moreover, the block-matching scheme also takes advantage of spatial contextual information, which has been widely utilized in spatial-spectral-based HSI classification methods. To evaluate the proposed methods, two benchmark HSIs are used, including a HSI collected over Pavia University, Italy by the airborne Reflective Optics System Imaging Spectrometer (ROSIS) sensor, and an image acquired over the Salinas Valley, California, USA via the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) sensor. Spatio-temporally coincident ground-reference data accompanies each of these respective HSIs. In addition, the proposed methods are compared with three state-of-the-art algorithms, including support vector machine (SVM), the 1-dimensional convolutional neural network (1DCNN), and the 1-dimensional RNN (1DRNN).Experimental results indicate that our proposed methods achieve markedly better classification performance compared with the baseline algorithms on both datasets. For example, for the Pavia University image, the block-matching based RNN achieves the highest overall classification accuracy, with 94.32% accuracy, which is 9.87% higher than the next most accurate algorithm of the aforementioned three baseline methods, which in this case is the 1DCNN, with 84.45% overall accuracy. More specifically, the block-matching method performs better than the pixel-matching method in terms of both quantitative and qualitative assessments. Based on visual assessment/interpretation of the classification maps, it is apparent that “salt-and-pepper” noise is markedly alleviated; with block-matching, smoother classified images are generated compared with pixel-matching-based methods and the three baseline algorithms. Such results demonstrate the effectiveness of utilizing spatial contextual information in the similarity measurement.

Classification is a common objective when analyzing hyperspectral images, where each pixel is assigned to a predefined label. Deep learning-based algorithms have been introduced in the remote-sensing community successfully in the past decade and have achieved significant performance improvements compared with conventional models. However, research on the extraction of sequential features utilizing a single image, instead of multi-temporal images still needs to be further investigated. In this paper, a novel strategy for constructing sequential features from a single image in long short-term memory (LSTM) is proposed. Two pixel-wise-based similarity measurements, including pixel-matching (PM) and block-matching (BM), are employed for the selection of sequence candidates from the whole image. Then, the sequential structure of a given pixel can be constructed as the input of LSTM by utilizing the first several matching pixels with high similarities. The resulting PM-based LSTM and BM-based LSTM are appealing, as all pixels in the whole image are taken into consideration when calculating the similarity. In addition, BM-based LSTM also utilizes local spectral-spatial information that has already shown its effectiveness in hyperspectral image classification. Two common distance measures, Euclidean distance and spectral angle mapping, are also investigated in this paper. Experiments with two benchmark hyperspectral images demonstrate that the proposed methods achieve marked improvements in classification performance relative to the other state-of-the-art methods considered. For instance, the highest overall accuracy achieved on the Pavia University image is 96.20% (using both BM-based LSTM and spectral angle mapping), which is an improvement compared with 84.45% overall accuracy generated by 1D convolutional neural networks.

Delineation of river-flow boundaries constitutes an important step in various river-related studies, including river hydraulic modeling, flow-width estimations, and river and floodplain habitat mapping and assessment. Increasing the level of automation of delineation of flow boundaries from synoptic remote-sensing images provides great potential, by reducing the labor cost, especially for studies focusing on long river reaches and those examining flow changes over time. This article investigates the boundary delineation of river channel flow from aerial photographs using Support Vector Machine (SVM) and image-derived ancillary data layers. It also includes a quantitative evaluation of delineation accuracy. The findings show that SVM performs satisfactory delineations of the boundaries, and the ancillary data layers generated using edge detectors and spatial domain texture statistics particularly increase delineation accuracy. Moreover, a multiscale evaluation scheme allows for examining the performance of...

The Mission-Aransas coastal region (MACR) in Texas is home to settlements vulnerable to coastal hazards. The region also contains significant biodiversity including several endangered species. The habitats in the bays and estuaries of MACR are especially sensitive to changes in land use/land cover (LULC) within the drainage basins upstream. We examine LULC change in the MACR from 1990 to 2010 and its implications for coastal vulnerability of the built environment and for the biodiversity in the region. Our findings show that, from 1990 to 2010, about a quarter of the MACR experienced LULC change. Developed land increased 71% (from 118 km 2 in 1990 to 203 km 2 in 2010), by far the greatest proportional change among all land cover classes. The rate of increase of developed land was slightly higher along the coast, 75% (from 65 km 2 in 1990 to 114 km 2 in 2010). Almost 90% of all developed land was within 50 km of the protected areas in both years. Overall, our findings point to increased exposure of the people and infrastructure to coastal hazards. Given the high social vulnerability in the study area, our study can inform formulation of sustainable management options that minimize both the coastal vulnerability of people and infrastructure and the pressure on the protected areas that are critical for conservation of biodiversity in the region.

For aquatic studies, radiative transfer (RT) modeling can be used to compute hyperspectral above-surface remote sensing reflectance that can be utilized for inverse model development. Inverse models can provide bathymetry and inherent-and bottom-optical property estimation. Because measured oceanic field/organic datasets are often spatio-temporally sparse, synthetic data generation is useful in yielding sufficiently large datasets for inversion model development; however, these forward-modeled data are computationally expensive and time-consuming to generate. This study establishes the magnitude of wall-clock-time savings achieved for performing large, aquatic RT batch-runs using parallel computing versus a sequential approach. Given 2,600 simulations and identical compute-node characteristics, sequential architecture required ~100 hours until termination, whereas a parallel approach required only ~2.5 hours (42 compute nodes)—a 40x speed-up. Tools developed for this parallel execution are discussed.

Computational trends toward shared services suggest the need to automatically manage spatial scale for overlapping applications. In three experiments using high-spatial-resolution optical imagery and LIDAR data to extract impervious, forest, and herbaceous classes, this study optimized C5.0 rule sets according to: (1) spatial scale within an image tile; (2) spatial scale within spectral clusters; and (3) stability of predicted accuracies based on cross validation. Alteration of the image segmentation scale parameter affected accuracy as did synergy with LIDAR derivatives. Within the tile examined, forest and herbaceous areas benefited more from optical and LIDAR synergy than did impervious surfaces.

We examine how access regimes, defined by a set of interrelated institutions and organizations, facilitated the flow of production resources and benefits that resulted in patterns of land change in the Brazilian Cerrado. We analyzed remotely sensed data (Landsat MSS, TM, +ETM) for three time periods (1972/1973–1986, 1986–1992, 1992–2002) with qualitative and archival data for eastern Mato Grosso state. Overall we found that land privately colonized was cleared more rapidly and extensively than lands under no colonization scheme. We also identified significant spatial variability in Cerrado conversion within and outside colonization areas and variability of annual rates of Cerrado conversion during each period. We explain that farmers in the Cerrado engaged in land-leasing and production contracts and worked through cooperatives and firms to marshal resources, including credit, technology, and inputs, which, in turn, influenced land-use decisions and regional patterns of land-cover change. We conclude that...

Introduced in this paper is the utilization of support vector machines (SVMs) to semiautomatically perform endmember extraction from hyperspectral data. The strengths of SVM are exploited to provide a fast and accurate calculated representation of high-dimensional data sets that may consist of multiple distributions. Once this representation is computed, the number of distributions can be determined without prior knowledge. For each distribution, an optimal transform can be determined that preserves informational content while reducing the data dimensionality and, hence, the computational cost. Finally, endmember extraction for the whole data set is accomplished. Results indicate that this SVM-based endmember extraction algorithm has the capability of semiautonomously determining endmembers from multiple clusters with computational speed and accuracy while maintaining a robust tolerance to noise.

The Brazilian Cerrado is threatened by agricultural land use conversion. Accurate quantification of overall and subtype Cerrado distributions is essential for regional monitoring. In this research, unsupervised fuzzy ARTMAP was compared against conventional k-means classification of Cerrado and agriculture, based on Hyperion satellite data. We systematically tested a range of fuzzy ARTMAP parameters, determining the best parameter combinations. The effect of an additional surface liquid-water input vector was also tested. Similar results were obtained when only Hyperion apparent surface reflectance data were used; fuzzy ARTMAP, however, was generally markedly more accurate than k-means when the additional surface liquid-water input was included.

The Brazilian Cerrado, a biodiverse savanna ecoregion covering ∼1.8 million km2 south and east of the Amazon rainforest, is in rapid decline because of the expansion of modern agriculture. Previous studies of Cerrado land-use and land-cover (LULC) change imply spatial homogeneity, report widely varying rates of land conversion, use ambiguous LULC categories, and generally do not attempt to validate results. This study addresses this gap in the literature by analyzing moderate-resolution, multi-spectral satellite remote sensing data from 1986 to 2002 in two regions with identical underlying drivers. Unsupervised classification by the ISODATA algorithm indicates that Cerrado was converted to agro-pastoral land covers in 31% (3646 km2) of the study region in western Bahia and 24% (3011 km2) of the eastern Mato Grosso study region, while nearly 40% (4688 km2 and 5217 km2, respectively) of each study region remained unchanged. Although aggregate land change is similar, large and contiguous fragments persist in western Bahia, while smaller fragments remain in eastern Mato Grosso. These findings are considered in the current context of Cerrado land-use policy, which is dominated by the conservation set-aside and command-control policy models. The spatial characteristics of Cerrado remnants create considerable obstacles to implement the models; an alternative approach, informed by countryside biogeography, may encourage collaboration between state officials and farmer-landowners toward conservation land-use policies.

Remote classification of land-use/land-cover (LULC) types in Brazil's Cerrado ecoregion is necessary because knowledge of Cerrado LULC is incomplete, sources of inaccuracy are unknown, and high-resolution data are required for the validation of moderate-resolution LULC maps. The aim of this research is to discriminate between Cerrado and agriculture using high-resolution Landsat 7 ETM+ imagery for the western region of Bahia state in northeastern Brazil. The Maximum Likelihood Classification (MLC) and Spectral Angle Mapper (SAM) algorithms were applied to a ∼3000 km2 subset, yielding comparable classification accuracies. The panchromatic band was reserved for validation. User's and producer's accuracies were highest for non-irrigated agriculture (∼94%) but lower for Cerrado Lato Sensu (89%). Classification errors likely resulted from spatial and spectral characteristics of particular classes (e.g. riparian forest and burned) and overestimation of other classes (e.g. Eucalyptus and water). Manual misinterp...

Continuum removal (CR) is often used for geologic mapping; however, more research is needed to better establish the utility of CR for vegetation classification, particularly when used with artificial neural networks (ANNs). In this paper, fuzzy learning vector quantization (FLVQ) was applied to hyperspectral Airborne Visible/Infrared Imaging Spectrometer imagery for coastal vegetation classification. FLVQ performance was compared with that of a multilayer perceptron (MLP), a self-organizing map (SOM), and an endmember-based algorithm [spectral feature fitting (SFF)]. The objective was to assess the effect of CR as an input vector-preprocessing step for ANN model development on classification accuracy. Compared with a related study, continuum intact (CI) reflectance data generally yielded higher classification accuracies than those based on CR. Thus, CR may not be a preferred preprocessing method for coastal vegetation mapping over broad wavelength ranges. MLP slightly outperformed FLVQ when applied to CI data, but FLVQ yielded higher accuracy than MLP with CR. However, there was no significant difference between them for both data treatments at the 95% confidence level. All ANNs tested yielded significantly higher classification accuracies than SFF. For model development, the 588-neuron FLVQ required only 8.2% of MLP training time, 27.8% of the 400-neuron SOM time, and 8.8% of the 729-neuron 3-D SOM time

Bathymetry is an important variable in scientific and operational applications. The research objectives in this study were to estimate bathymetry based on derivative reflectance spectra used as input to a multilayer perceptron artificial neural network (ANN) and to evaluate the efficacy of field and simulated training/testing data. ANNs were used to invert reflectance field data acquired in optically shallow coastal waters. Results indicate that for the simulation-based models, nonderivative spectra yielded more accurate bathymetry retrievals than the derivative spectra used as ANN input. However, for the empirical field-based models, derivative spectra were superior to nonderivative spectra as ANN input. This study identifies circumstances under which derivative spectra are useful in bathymetry estimation, and thus increases the likelihood of obtaining accurate inversions.

Artificial neural networks (ANNs) may be of significant value in extracting vegetation type information in complex vegetation mapping problems, particularly in coastal wetland environments. Unsupervised, self-organizing ANNs have not been employed as frequently as supervised ANNs for vegetation mapping tasks, and further remote sensing research involving fuzzy ANNs is also needed. In this research, the utility of a fuzzy unsupervised ANN, specifically a fuzzy learning vector quantization (FLVQ) ANN, was investigated in the context of hyperspectral AVIRIS image classification. One key feature of the neural approach is that unlike conventional hyperspectral data processing methods, endmembers for a given scene, which can be difficult to determine with confidence, are not required for neural analysis. The classification accuracy of FLVQ was comparable to a conventional supervised multi-layer perceptron, trained with backpropagation (MLP) (KHAT ( Kˆ ) accuracy: 82.82% and 84.66%, respectively; normalized accuracy: 74.60% and 75.85%, respectively), with no significant difference at the 95% confidence level. All neural algorithms in the experiment yielded significantly higher classification accuracies than the conventional endmember-based hyperspectral mapping method assessed (i.e., matched filtering, where Kˆ accuracy = 61.00% and normalized accuracy = 57.96%). FLVQ was also dramatically more computationally efficient than the baseline supervised and unsupervised ANN algorithms tested, including the MLP and the Kohonen self-organizing map (SOM), respectively. The 400-neuron FLVQ network required only 3.6% of the computation time used by the MLP network, and only 5.9% of the MLP time was used by the 588-neuron FLVQ network. In addition, the 400-neuron FLVQ used only 16.7% of the time used by the 400-neuron SOM for model development.

Accurate characterization of the direction of land change is a neglected aspect of land dynamics. Knowledge on direction of historical land change can be useful information when understanding relative influence of different land-change drivers is of interest. In this study, we present a novel perspective on land-change analysis by focusing on directionality of change. To this end, we employed Maximum Cross-Correlation (MCC) approach to estimate the directional change in land cover in a dynamic river floodplain environment using Landsat 5 Thematic Mapper (TM) images. This approach has previously been used for detecting and measuring fluid and ice motions but not to study directional changes in land cover. We applied the MCC approach on land-cover class membership layers derived from fuzzy remote-sensing image classification. We tested the sensitivity of the resulting displacement vectors to three user-defined parameters—template size, search window size, and a threshold parameter to determine valid (non-noisy) displacement vectors—that directly affect the generation of change, or displacement, vectors; this has not previously been thoroughly investigated in any application domain. The results demonstrate that it is possible to quantitatively measure the rate of directional change in land cover in this floodplain environment using this particular approach. Sensitivity analyses indicate that template size and MCC threshold parameter are more influential on the displacement vectors than search window size. The results vary by land-cover class, suggesting that spatial configuration of land-cover classes should be taken into consideration in the implementation of the method.

Shadows in remote-sensor images can yield marked errors in classification of riverine environments. We propose use of a modified shadow-removal algorithm as a preprocessing step for remote-sensing image classification of riverine landscapes. To accommodate characterization of spatially complex river features in the image, we investigate an illumination suppression-based shadow-removal algorithm, modified to include a user-defined tiling approach. We quantitatively evaluate the influence of shadow removal from aerial photography on classification accuracy as such studies are currently lacking. Experimental results demonstrate that this modified shadow-removal method significantly increases classification accuracy and improves detection of small river channels partially obscured by shadow.

Automatic feature extraction from satellite imagery is cost effective and fast. An essential issue in this context is the degree of accuracy for thematic correctness obtainable through common pixel-based and object-oriented classification algorithms. By applying two classification algorithms to Landsat 5 TM imagery for the extraction of different morphological river features the thematic correctness of the resulting raster images and the separability of the river features is evaluated. River features of meandering rivers evolve through dynamic avulsion, erosion and deposition processes. Although many studies focus on the analysis of these river environments, diverse methods of GIS and remote sensing based river feature classification methods have not been evaluated and assessed yet. In the literature several techniques to monitor spatio-temporal changes such as lateral river channel migration are already mentioned but the tendency there is to identify the changes by examining time spans rather than a point in time. Besides that the semiautomatic river feature methods described in related studies mainly focus on the identification of a river channel itself and do not consider additional features such as oxbows, scars, relic channels, etc. that in fact are significant characters in riverine environments. Therefore, this paper evaluates the application of a supervised classification using ENVI’s Support Vector Machine and an object based classification using the ArcGIS extension Feature Analyst to extract river features from Landsat 5 TM images including ancillary data files. Furthermore, the results of the classification methods are evaluated with regard to thematic correctness and separability of the various classified river features using accuracy assessment as presented in the specialist literature. Finally the long-time changes in the riverine environments are traced by interpreting the distribution of the classified river features. Accordingly, the approach of this work contributes to on-going research concerning semiautomatic or automatic river feature extraction.

Remote sensing involves the collection of information about an object from a distance. Often remote-sensing instruments are mounted onboard an airor space-borne platform and typically record electromagnetic energy in specific wavelength intervals, or bands. The electromagnetic energy recorded over a given area contains information about surfaces reflecting or emitting energy. This information can be used for a variety of applications; for example, remote-sensing image analysis can extract thematic information such as land-cover types (Jensen, 2005). Artificial neural network (ANN) techniques have increasingly been employed in the analysis of remotely-sensed images. ANNs can be advantageous in digital image processing in that no assumption is made about the statistical properties of the images, and they are thus widely applicable to a variety of dataset types. In addition, ANNs learn adaptively through examples and have a high tolerance to noisy or incomplete data (Jensen, 2005). ANN model development can proceed via either supervised or unsupervised means, and if adequate training data are available, supervised training may be readily performed. However, obtaining reliable training data in remote-sensing applications is often problematic (Congalton and Green, 1999), as a remote sensor image typically covers a large area, and only a limited number of training locations can be sampled in the field due to cost, time, personnel requirements, and various other logistical constraints, including potential restrictions on access to the study area. Unsupervised image-processing methods— including unsupervised ANNs—can be of significant utility in such circumstances (Filippi et al., 2009). Unsupervised ANNs are used in situations where the correct outputs may not be known, or if it is desired that the network discover or categorize regularities or features in the training data on its own. There is no teacher signal (Hassoun, 1995). The unsupervised Kohonen self-organizing map (SOM) is a two-layer network, with an input fan-out layer, and an output layer (known as the Kohonen or competitive layer), and the method is based upon competitive learning. The Kohonen layer is comprised of a physical net of neurons located at fixed positions (i.e., intersections in a grid of square meshes). Adjacent neurons are assumed to have a Euclidean distance of unity. The input 14

Extracting endmembers from remotely-sensed images of vegetated areas can present difficulties. In this research, we applied a recently-developed endmember-extraction algorithm based on Support Vector Machines to the problem of semi-autonomous estimation of vegetation endmembers from a hyperspectral image. This algorithm, referred to as Support Vector Machine-Based Endmember Extraction (SVM-BEE), accurately and rapidly yields a computed representation of hyperspectral data that can accommodate multiple distributions. The number of distributions is identified without prior knowledge, based upon this representation. Prior work established that SVM-BEE is robustly noise-tolerant and can semi-automatically estimate endmembers; synthetic data and a geologic scene were previously analyzed. Here we compared the efficacies of SVM-BEE, N-FINDR, and SMACC algorithms in extracting endmembers from a real, predominantly-agricultural scene. SVM-BEE estimated vegetation and other endmembers for all classes in the image, which N-FINDR and SMACC failed to do. SVM-BEE was consistent in the endmembers that it estimated across replicate trials. Spectral angle mapper (SAM) classifications based on SVM-BEE-estimated endmembers were significantly more accurate compared with those based on N-FINDR- and (in general) SMACC-endmembers. Linear spectral unmixing accrued overall accuracies similar to those of SAM.

[1] An optimization-based, shallow water remote-sensing inversion algorithm was recently developed by Lee et al. (1999) that simultaneously derives bottom depth and water column inherent optical properties. Only measured remote-sensing reflectance is required as input, which is a noted advantage; however, the algorithm is sensitive to noise. Given the observation that bottom depth and optical properties generally change slowly over the spatial domain, we applied a smoothness assumption by modifying the Lee algorithm to accommodate a spatial smoothness constraint. Spatial constraints were introduced through a linear diffusion process. The new spatial constraint model retrievals were compared with those from the original Lee model, as well as from the original model plus smoothing performed as postprocessing. Synthetic and real field data experiments were performed. For the synthetic experiments the Lee method was the most sensitive to noise and posted the largest absolute and standard errors. For the field experiments, Portable Hyperspectral Imager for Low Light Spectroscopy Sensor (PHILLS) imagery was acquired over optically shallow water. An in situ acoustic sensor mounted on a towed Hyperspectral Tethered Spectral Radiometer Buoy (HTSRB) provided measured bottom depth transects, and remote-sensing reflectance was HTSRB derived. Bathymetry estimates were validated in this research. Lee method bottom depth retrievals were more erratic than those generated via the other methods. Unlike postsmoothing the spatial constraint method corrected large deviations of the Lee method spanning several transect points. Overall, our spatial constraint method yielded the most accurate estimates and represents a significant improvement upon the Lee method. Mean absolute estimation errors for the Lee, postprocessing, and spatial constraint methods were 0.412, 0.389, and 0.335 m, respectively.

For natural or anthropogenic disasters, rapid assessment is critical for an appropriate and effective emergency response. Remote sensing has served—and will continue to serve—a vital function in disaster damage-assessment activities. This includes disaster-mapping of natural and agricultural ecosystems and human settlements, which may involve assessments of structural damage, contamination, and affected populations. Single- and multi-date (change detection) analyses can be employed, and a need to exploit both spectral and spatial information in order to delineate damage regions from remote sensor imagery is identified. This chapter provides a brief overview of some of the remote-sensing damage-assessment applications that are of utility in the realm of homeland security. Specific attention is given to remote sensing-based detection of vegetation damage and soil contamination, including a discussion of the remote-sensing implications of artificial radionuclide contamination, as well as damage to urbanized areas and other human settlements.

The Cerrado is a savanna ecoregion with grassland and woodland subtypes covering ~one-quarter of Brazil and is considered to be a biodiversity hotspot, threatened by land-use conversion. Hyperspectral remote sensing enables spatio-temporal monitoring, while providing the possibility of vegetation-mapping at a high level of specificity. However, because imaging spectrometer data availability/coverage is currently limited, a need exists for effective exploitation of multispectral satellite imagery with broad-area spatial coverage. The objective was to assess the utility of hyperspectral Hyperion and multispectral CBERS-2 satellite imagery in discriminating among Cerrado subtypes and agricultural classes. Temporally-coincident field-transect data for Cerrado physiognomies and agricultural sites were collected, including biophysical metrics. Nonmetric multidimensional scaling and hierarchical cluster analysis were used to identify potential environmental gradients of biophysical groupings. Four Cerrado subclasses were identified: Campo Limpo (Open Cerrado Grassland), Campo Sujo (Shrub Savanna), Cerrado Tipico (Wooded Cerrado), and Cerrado Denso (Cerrado Woodland). Subclasses were then merged, forming two Cerrado subclasses. To facilitate sensor intercomparison, image classification involved PCA transformations, followed by unsupervised clustering of the component images. Results indicate that both dimensionality-reduced Hyperion and CBERS datasets were sufficient in distinguishing between the two more general Cerrado subclasses and agriculture, but the Hyperion-derived classification was more accurate.

[1] Ocean color remote-sensing systems require highly accurate calibration (

The United States Navy has recently shifted focus from open-ocean warfare to joint operations in optically complex nearshore regions. Accurately estimating bathymetry and water column inherent optical properties (IOPs) from passive remotely sensed imagery can be an important facilitator of naval operations. Lee et al. developed a semianalytical model that describes the relationship between shallow-water bottom depth, IOPs and subsurface and above-surface reflectance. They also developed a nonlinear optimization-based technique that estimates bottom depth and IOPs, using only measured spectral remote sensing reflectance as input. While quite effective, inversion using noisy field data can limit its accuracy. In this research, the nonlinear optimization-based Lee et al. inversion algorithm was used as a baseline method, and it provided the framework for a proposed hybrid evolutionary/classical optimization approach to hyperspectral data processing. All aspects of the proposed implementation were held constant with that of Lee et al., except that a hybrid evolutionary/classical optimizer (HECO) was substituted for the nonlinear method. HECO required more computer-processing time. In addition, HECO is nondeterministic, and the termination strategy is heuristic. However, the HECO method makes no assumptions regarding the mathematical form of the problem functions. Also, whereas smooth nonlinear optimization is only guaranteed to find a locally optimal solution, HECO has a higher probability of finding a more globally optimal result. While the HECO-acquired results are not provably optimal, we have empirically found that for certain variables, HECO does provide estimates comparable to nonlinear optimization (e.g., bottom albedo at 550 nm).

Ratio-based bottom depth-retrieval algorithms are conceptually simple relative to other algorithms and can be effective. The objective of this study was to determine the utility of imposing a spatial-smoothing assumption on three ratio-based, feed-forward remote-sensing bathymetry algorithms: Polcyn et al., Stumpf et al., and Dierssen et al. We consider three smoothing operators: median, Savitzky-Golay, and linear diffusion with data fidelity, applied in three domains: spatial, spectral, and spectral-spatial. Thus, we consider nine smoothing methods. In addition, we consider two points at which smoothing is applied: one before the inversion process (pre-smoothing) and the other after the inversion process (post-smoothing). Our new formulations were tested with synthetic data, in situ remote-sensing reflectance, and simultaneous acoustic bathymetry, acquired in optically shallow waters. Analysis and results from the synthetic-data experiment indicate that pre-smoothing method is more effective than post-smoothing method. The field-data experiments indicate that spatial-domain smoothing is effective regardless of the type of smoothing operator, whereas spectral smoothing is not. Spectral-spatial-domain smoothing is as effective as spatial-domain smoothing, but is prone to over-segmentation. Effectiveness of spatial pre-smoothing was observed with every ratio-based inversion method, which suggests potential universal applicability of smoothing operators to ratio-based algorithms.