Reginald S. Fletcher

Evaluating AISA+ hyperspectral imagery for mapping black mangrove along the South Texas Gulf coast.

Mangrove wetlands are economically and ecologically important ecosystems and accurate assessment of these wetlands with remote sensing can assist in their management and conservation. This study was conducted to evaluate airborne AISA hyperspectral imagery and image transformation and classification techniques for mapping black mangrove populations on the south Texas Gulf coast. AISA hyperspectral imagery was acquired from two study sites and both minimum noise fraction (MNF) and inverse MNF transforms were performed. Four classification methods, including minimum distance, Mahalanobis distance, maximum likelihood, and spectral angle mapper (SAM), were applied to the noise-reduced hyperspectral imagery and to the band-reduced MNF imagery for distinguishing black mangrove from associated plant species and other cover types. Accuracy assessment showed that overall accuracy varied from 84 percent to 95 percent for site 1 and from 69 percent to 91 percent for site 2 among the eight classifications for each site. The MNF images provided similar or better classification results compared with the hyperspectral images among the four classifiers. Kappa analysis showed that there were no significant differences among the four classifiers with the MNF imagery, though maximum likelihood provided excellent overall and class accuracies for both sites. Producer’s and user’s accuracies for black mangrove were 91 percent and 94 percent, respectively, for site 1 and both 91 percent for site 2 based on maximum likelihood applied to the MNF imagery. These results indicate that airborne hyperspectral imagery combined with image transformation and classification techniques can be a useful tool for monitoring and mapping black mangrove distributions in coastal environments.

Spectral analysis of coastal vegetation and land cover using AISA + hyperspectral data

This paper describes a spectral analysis of several coastal land cover types in South Padre Island, Texas using AISA+ hyperspectral remote sensing data. AISA+ hyperspectral data (1.5 metre) were acquired throughout the area on 9 March 2005. Data over mangrove areas were converted to percent reflectance using four 8×8 metre reflectance tarps (4%, 16%, 32% and 48%) and empirical line calibration. These data were then compared to percent reflectance values of other terrestrial features to determine the ability of AISA+ data to distinguish features in coastal environments. Results suggest that these data may be appropriate to discriminate coastal mangrove vegetation and provide researchers with high resolution spatial and spectral information to more effectively manage coastal ecosystems.

Evaluating high spatial resolution imagery for detecting citrus orchards affected by sooty mould

This study evaluated high spatial resolution colour‐infrared (CIR) (2.44 m) and pansharpened CIR imagery (0.61 m) for detecting citrus (Citrus spp.) orchards affected by sooty mould (Capnodium citri), an indicator of insect infestation of a citrus grove. These resolutions were chosen because they are equivalent to the spatial resolution of multispectral and pansharpened QuickBird imagery. Citrus groves north‐west of Mission, Texas, USA, were assessed. CIR photography and image processing software were used to develop the images. Sooty mould‐affected areas were readily detected on the CIR and pansharpened CIR images. The latter provided better detail, increasing image interpretation accuracy. Findings of this study support the theory that high spatial resolution satellite imagery may be used to detect sooty mould‐affected citrus orchards.

Using Aerial Color‐infrared Photography and QuickBird Satellite Imagery for Mapping Wetland Vegetation

Abstract Aerial color‐infrared (CIR) photography and QuickBird high resolution (2.8 m) false color satellite imagery were evaluated for differentiating among wetland vegetation in two freshwater lakes on the Welder Wildlife Refuge in south Texas. Field spectral measurements made on dominant vegetation types (plant species and vegetation mixtures) showed significant differences in visible and near‐infrared reflectance. Several plant species and mixtures of species could be distinguished in the aerial photos and satellite imagery. Accuracy assessments performed on computer classifications of the photos of the two lakes had overall accuracies of 84% and 87%; whereas, accuracy assessments performed on classifications of the satellite imagery of the lakes had overall accuracies of 69% and 76%. The lower accuracies of the satellite image classifications were attributed to their coarser spatial resolution.

Early Detection of Crop Injury from Glyphosate on Soybean and Cotton Using Plant Leaf Hyperspectral Data

In this paper, we aim to detect crop injury from glyphosate, a herbicide, by both traditionally used spectral indices and newly extracted features with leaf hyperspectral reflectance data for non-Glyphosate-Resistant (non-GR) soybean and non-GR cotton. The new features were extracted by canonical analysis technique, which could provide the largest separability to distinguish the injured leaves from the healthy ones. Spectral bands used for constructing these new features were selected based on the sensitivity analysis results of a physically-based leaf radiation transfer model (leaf optical PROperty SPECTra model, PROSPECT), which could help extend the effectiveness of these features to a wide range of leaf structures and growing conditions. This approach has been validated with greenhouse measured data acquired in glyphosate treatment experiments. Results indicated that glyphosate injury could be detected by NDVI (Normalized Difference Vegetation Index), RVI (Ratio Vegetation Index), SAVI (Soil Adjusted Vegetation Index), and DVI (Difference Vegetation Index) in 48 h After the Treatment (HAT) for soybean and in 72 HAT for cotton, but the other spectral indices either showed little use for separation, or did not show consistent separation for healthy and injured soybean and cotton. Compared with the traditional spectral indices, the new features were more feasible for the early detection of glyphosate injury, with leaves sprayed with a higher rate of glyphosate solution having larger feature values. This trend became more and more pronounced with time. Leaves sprayed with different glyphosate rates showed some separability 24 HAT using the new features and could be totally distinguished at and beyond 48 HAT for both soybean and cotton. These findings demonstrated the feasibility of applying leaf hyperspectral reflectance measurements for the early detection of glyphosate injury using these newly proposed features.

Using high resolution QuickBird imagery for crop identification and area estimation

Imagery from recently launched high spatial resolution satellite sensors offers new opportunities for crop assessment and monitoring. A 2.8-m multispectral QuickBird image covering an intensively cropped area in south Texas was evaluated for crop identification and area estimation. Three reduced-resolution images with pixel sizes of 11.2 m, 19.6 m, and 30.8 m were also generated from the original image to simulate coarser resolution imagery from other satellite systems. Supervised classification techniques were used to classify the original image and the three aggregated images into five crop classes (grain sorghum, cotton, citrus, sugarcane, and melons) and five non-crop cover types (mixed herbaceous species, mixed brush, water bodies, wet areas, and dry soil/roads). The five non-crop classes in the 10-category classification maps were then merged as one class. The classification maps were filtered to remove the small inclusions of other classes within the dominant class. For accuracy assessment of the classification maps, crop fields were ground verified and field boundaries were digitized from the original image to determine reference field areas for the five crops. Overall accuracy for the unfiltered 2.8-m, 11.2-m, 19.6-m, and 30.8-m classification maps were 71.4, 76.9, 77.1, and 78.0%, respectively, while overall accuracy for the respective filtered classification maps were 83.6, 82.3, 79.8, and 78.5%. Although increase in pixel size improved overall accuracy for the unfiltered classification maps, the filtered 2.8-m classification map provided the best overall accuracy. Percentage area estimates based on the filtered 2.8-m classification map (34.3, 16.4, 2.3, 2.2, 8.0, and 36.8% for grain sorghum, cotton, citrus, sugarcane, melons, and non-crop, respectively) agreed well with estimates from the digitized polygon map (35.0, 17.9, 2.4, 2.1, 8.0, and 34.6% for the respective categories). These results indicate that QuickBird imagery can be a useful data source for identifying crop types and estimating crop areas.

Glyphosate-resistant and glyphosate-susceptible Palmer amaranth (Amaranthus palmeri S. Wats.): hyperspectral reflectance properties of plants and potential for classification

BACKGROUND Palmer amaranth (Amaranthus palmeri S. Wats.) is a troublesome agronomic weed in the southern United States, and several populations have evolved resistance to glyphosate. This paper reports on spectral signatures of glyphosate-resistant (GR) and glyphosate-sensitive (GS) plants, and explores the potential of using hyperspectral sensors to distinguish GR from GS plants. RESULTS GS plants have higher light reflectance in the visible region and lower light reflectance in the infrared region of the spectrum compared with GR plants. The normalized reflectance spectrum of the GR and GS plants had best separability in the 400-500 nm, 650-690 nm, 730-740 nm and 800-900 nm spectral regions. Fourteen wavebands from within or near these four spectral regions provided a classification of unknown set of GR and GS plants, with a validation accuracy of 94% for greenhouse-grown plants and 96% for field-grown plants. CONCLUSIONS GR and GS Palmer amaranth plants have unique hyperspectral reflectance properties, and there are four distinct regions of the spectrum that can separate the GR from GS plants. These results demonstrate that hyperspectral imaging has potential application to distinguish GR from GS Palmer amaranth plants (without a glyphosate treatment), with future implications for glyphosate resistance management. Published 2014. This article is a U.S. Government work and is in the public domain in the USA.

Evaluating airborne hyperspectral imagery for mapping saltcedar infestations in west Texas

Abstract. The Rio Grande River of west Texas contains by far the largest infestation of saltcedar (Tamarix spp.) in Texas. The objective of this study was to evaluate airborne hyperspectral imagery and different classification techniques for mapping saltcedar infestations. Hyperspectral imagery with 102 usable bands covering a spectral range of 475 to 845 nm was acquired from two sites along the Rio Grande in west Texas in December 2003 and 2004 when saltcedar was undergoing color change. The imagery was transformed using minimum noise fraction and then classified using six classifiers: minimum distance, Mahalanobis distance, maximum likelihood, spectral angle mapper, mixture tuned matched filtering, and support vector machine (SVM). Accuracy assessment showed that overall accuracy varied from 71% to 86% in 2003 and from 80% to 90% in 2004 for site 1 and from 60% to 76% in 2003 and from 77% to 91% in 2004 for site 2. The SVM classifier produced the highest overall accuracy, as well as the best user’s and producer’s accuracies for saltcedar among the six classifiers. The imagery taken in early December 2004 provided better classification results than that in mid-December 2003. Change detection analysis based on the classification maps quantified the class changes between the two years. These results indicate that airborne hyperspectral imagery incorporated with image transformation and classification techniques can be a useful tool for mapping saltcedar infestations.

Evaluation of High-Resolution Satellite Imagery for Assessing Rangeland Resources in South Texas

Abstract QuickBird satellite imagery was evaluated for differentiating among rangeland cover types on the Welder Wildlife Refuge in south Texas. The satellite imagery had a spatial resolution of 2.8 m and contained 11-bit data. Four subsets of the satellite image were extracted and used as study sites. Field spectral measurements made among the dominant vegetation types showed significant differences in visible and near-infrared reflectance. Unsupervised classification techniques were used to classify false color composite (green, red, and near-infrared bands) images of each study site. Accuracy assessments performed on the classification maps of the 4 sites had overall accuracies ranging from 79% to 89%. These results indicate that QuickBird imagery can be a useful tool for identifying rangeland cover types at a regional level.

Random forest and leaf multispectral reflectance data to differentiate three soybean varieties from two pigweeds

We tested random forest to distinguish two pigweeds from three soybean varieties.We used leaf multispectral data as input variables for the random forest algorithm.Classification accuracies ranged from 93.8% to 100%.Random forest can use leaf multispectral data to distinguish pigweeds from soybean. Accurate weed identification is a prerequisite for implementing site-specific weed management in crop production. Palmer amaranth (Amaranthus palmeri S. Wats.) and redroot pigweed (Amaranthus retroflexus L.) are two common pigweeds that reduce soybean Glycine max (L.) Merr. yields in the southeastern United States. The objective of this study was to evaluate leaf multispectral reflectance data as input into the random forest machine learning algorithm to differentiate three soybean varieties (Progeny 4928, Progeny 5160, and Progeny 5460) from Palmer amaranth and redroot pigweed. Leaf reflectance measurements of soybean, Palmer amaranth, and redroot pigweed plants grown in a greenhouse were collected with a plant contact probe attached to a hyperspectral spectroradiometer. Data were obtained at the vegetative growth stage of the plants on two dates, June 30, 2014, and September 17, 2014. The hyperspectral data were aggregated to sixteen multispectral bands (viz. coastal, blue, green, yellow, red, red-edge, near-infrared 1 and 2, and shortwave-infrared 1-8) mimicking those recorded by the WorldView-3 satellite sensor. Classifications were binary, meaning one soybean variety versus one weed tested per classification. Random forest classification accuracies were determined with a confusion matrix, incorporating users, producers, and overall accuracies and the kappa coefficient. Users, producers, and overall accuracies of the soybean weed classifications ranged from 93.8% to 100%. Kappa results (values of 0.93-0.97) indicated an excellent agreement between the classes predicted by the models and the actual reference data. Shortwave-infrared bands were ranked the most important variables for distinguishing the pigweeds from the soybean varieties. These results suggest that random forest and leaf multispectral reflectance data could be used as tools to differentiate soybean from two pigweeds with a potential application of this technology in site-specific weed management programs.