update coming soon
update coming soon
update coming soon
update coming soon
An unprecedented land conservation effort is presently underway in the Gulf of Mexico Coastal Region (GCR) due to an influx of funds from settlements related to the 2012 RESTORE Act. A complete understanding of the priorities of the states in the GCR is critical to ensure that land conservation planning efforts are implemented effectively and efficiently. This work reviews past, current, and future land conservation priorities in the GCR to inform strategic planning efforts. This review catalogs an extensive list of projects and plans proposed and implemented at federal, state, county and city levels with direct ties to land conservation during the past 20 years. Comprehensive restoration goals proposed by the Gulf Coast Ecosystem Restoration (Restore) Council were used as a framework for grouping priorities within conservation plans and projects. Plans were first compiled via internet searches and expert sources, then a series of eight stakeholder charrettes were held across the GCR states to validate the catalog and add missing projects and plans. A geospatial web tool was developed using the Restore Council goal framework to allow for the identification and exploration of plans in the GCR.
Conservation prioritization tool developed in this work (journal article in preparation) aims to help resource managers choose optimal land parcel for conservation by using openly available GIS/Remote sensing data (from federal and state agencies such as NOAA, EPA, USDA, etc) and stochastic multi-criteria decision analysis algorithm.
Wildfires can cause severe impacts to ecosystems, property, human health, and safety. Management, monitoring, and suppression of wildfires can cost billions of dollars. In the United States of America (USA) alone, it cost $13 billion for fire suppression and $5 billion for management from 2006–2015. Research suggests that this may get worse due to climate change as projections suggest an increase of 20% to 50% in the number of days conducive to wildfire events. Although wildfire has ecological benefits, such as increasing biodiversity, reducing biomass and fuel loads, releasing nutrients, and influencing plant stand composition and health, they can also cause significant financial and quality of life impacts for humans. Since wildfire events can create long-term ecosystem impacts, information on damage to impacted vegetation must be estimated in a timely manner for restoration, and prediction purposes. Burned areas have unique patterns based on burn severity, local topography, and vegetation type. Methods for identifying and mapping these areas include (a) traversing the boundary of a burned area with a handheld Global Positioning System (GPS) unit, (b) using imagery/data captured from a human-crewed aircraft, and/or (c) using satellite-collected imagery. Remote sensing of wildfires via multispectral sensors mounted on satellite or manned aircraft allows estimation of vegetation recovery rate and burn severity on various vegetation types over large areas, but satellite imagery often lacks the spatial resolution needed to provide accurate burn maps in smaller, localized areas and manned aircraft operation involves complex logistics such as clearance and nexus to an airport, etc. The UAS based approach proposed in this work alleviates many drawbacks of traditional approaches.
Manned aircraft have been traditionally used to conduct aerial surveys of wildlife in large areas where ground surveys would be too costly or impractical. Although these aerial surveys are useful, they have limitations in terms of the high costs associated with the purchase or lease of planes, and operational costs such as fuel, and staff and pilot labor. Also, considerable expertise and training are required of pilots and staff to meet operational needs and reduce the risk of injury or even death in one of the more hazardous endeavors in the wildlife profession. Manned aircraft are also well known for disturbing wildlife during low altitude surveys, and often result in biased estimates due to observer subjectivity. Unmanned aerial systems (UAS) are a rapidly advancing tool that may be used to address some of the issues associated with these legacy approaches to aerial survey methods.
High-resolution imagery collected from UAS platforms also provides the opportunity for computer-guided algorithms to identify target organisms, eliminating the time-consuming task of manual counting. This application is already showing promising results in wildlife monitoring. The benefits of low altitude sensing with high-resolution optical sensors and computer vision algorithms to precisely identify and measure ground targets make UAS a potentially useful wildlife monitoring tool. I employed a relatively small, inexpensive UAS platform capable of collecting geo-referenced high-resolution imagery to conduct surveys of fish-eating birds on selected catfish (Ictalurus spp.) aquaculture facilities in the primary aquaculture producing areas of Mississippi.
The considerable research effort has been expended on determining potential economic impacts of fish-eating birds on the catfish aquaculture industry. A key component in determining the extent of depredation and loss has been the distribution of fish-eating birds on farm ponds, the proportion of farm ponds utilized, and the type or condition of ponds utilized. Proportional use and count information are essential in determining the economic impact of fish-eating birds on the catfish aquaculture industry. Historically these surveys have been conducted from the ground or by air using certified pilots, typically in fixed-wing aircraft. Platforms such as UAS may be a useful alternative to assess damage to agricultural commodities from many sources, including wildlife. The goal was to evaluate the resolution and extent of coverage necessary to provide for UAS remotely sensed and pattern recognition based censuses of fish-eating birds.
Wild pig (Sus scrofa) population expansion and associated damage to crops, wildlife, and the environment is a growing concern in the United States. The destructive rooting behavior of wild pigs indicates where they have foraged and their general presence on the landscape. This study used a texture-based classification approach on aerial imagery collected with a small unmanned aerial system to assess the damage of corn (Zea mays) fields by wild pigs in the Mississippi Alluvial Valley of Mississippi, USA, during the 2016 growing season. Images were automatically classified using segmentation‐based fractal texture analysis and support vector machines. I assessed the accuracy of automated classification with 5,400 Global Positioning System ground reference points collected in the fields. Classification accuracies for identification of damaged and nondamaged areas were between 65% and 78%. In general, automated classification underestimated the area of damage present within fields. Kappa values ranged from 0.26 to 0.51, on a scale of 0.0–1.0. Small-unmanned aerial systems overcome limitations of existing methods because they can survey an entire field rapidly and without significant field labor
Timber theft is the intentional illegal harvesting of timber that belongs to someone else. Similarly, timber trespass is the unintentional harvest of another person’s timber; this usually occurs while legally logging a site adjacent to where the trespass occurs. Timber theft and trespass are common problems worldwide. Both carry a civil penalty in most US states and the thief/trespasser, if caught and convicted, is usually required to pay damages in excess of the market value of the illegally harvested timber at the time the harvest occurred. The annual estimation of loss to landowners and timber companies due to timber theft/trespass is in excess of US $20 billion globally. Aside from the economic loss, timber theft can also cause environmental damage from leftover slash piles as they can become fuel for wildfires. Timber theft/trespass is generally very difficult to detect in remote areas, as some landowners live far from the property they own or often travel away from the property they live on. Accurate assessment of the diameter and number of tree stumps due to timber theft or trespass is needed to quantify the financial value of illegally harvested timber.
Tree stump diameter can be used to estimate tree height and thus timber volume. Use of ground sampling techniques to quantify stump diameter and number following illegal timber harvest may not be cost effective and can be labor intensive. Initial detection of timber trespass or theft is often not found for months or years after the event. Traditionally when it is found, labor-intensive data collection and investigations must occur. If a rapid and cost-effective assessment of recently finished logging jobs could occur (as they are completed), landowners would know exactly what was cut and that no theft had occurred. Likewise, if there had been tree removals that were unknown at the time, these issues could be addressed immediately. Traditional data collection after timber theft places a financial value directly estimated from the number and diameter of the stumps removed. However, this process can be time-consuming and expensive as it requires a person walking over a site to collect needed data.
To overcome the limitations of traditional approaches, this study used a small unmanned aerial system (UAS) capable of collecting geo-referenceable, high spatial resolution, visible spectrum imagery, template matching, phase coding, fast marching and Hough circle transform, a computer vision algorithm for detecting circular objects in the imagery for counting and estimating parameters of cut tree stumps. The use of UAS has been shown to be an effective tool in forest monitoring and management.
Phragmites australis (Cav.) Trin. Ex Steud (or common reed - henceforth Phragmites) is a perennial grass found in marsh systems and wetlands across the US. Invasive Phragmites is a tall (up to 6 m) plant that forms large impenetrable monocultures that displace native plants by outcompeting them for resources. Leaves are lanceolate at an average of 30 cm long and 3 cm wide. Phragmites can spread rapidly by stout creeping rhizomes over short distances. Long distance dispersal is predominantly carried out via seeds carried on wind currents. However populations along the Gulf Coast were shown to have sterile seeds thus long-distance dispersal in these populations is most likely done through the dispersal of rhizome fragments. In the US, apart from native Phragmites, two invasive haplotypes have been introduced: haplotype M and haplotype I. Haplotype M is predominant in most of the US, except for the Gulf Coast region where Haplotype I is predominant. Native Phragmites was not historically found on the Gulf Coast of the US, thus any population in this region is considered non-native. In the delta of the Pearl River in southeastern Louisiana, Phragmites is a major navigation hazard to small boats, hindering visibility as it is the tallest grass species in this highly braided waterbody. The US Environmental Protection Agency (EPA) recently recognized invasive species such as Phragmites as a major threat to coastal ecosystems. Current control measures for Phragmites include mowing, grazing, burning, and herbicide application. The above control measures must usually be repeated to halt the spread of Phragmites.
The effectiveness of these control efforts is greatly influenced by the accuracy of mapped Phragmites locations. A location map of Phragmites is usually obtained using satellite imagery, imagery captured from a manned aircraft, and/or walking around or through a Phragmites stand with a Global Positioning System (GPS) unit. Satellite imagery, while readily available, usually has spatial resolutions above 5 m and revisit times can vary from weeks to months. The poor spatial resolution of satellite imagery restricts the ability to delineate and map small Phragmites patches (<5 m2) which can lead to the reestablishment of a stand after the completion of management efforts. The longer revisit times of satellites can hamper management efforts as Phragmites may have spread beyond the last known location by the time control measures are implemented. Collecting imagery from a manned aircraft may suffer none of these issues, but in some cases is not an economically viable option, or imagery of a site may be obstructed by cloud cover. Manually mapping Phragmites patches is time-consuming. The presence of wildlife and navigation over difficult terrain creates a dangerous working environment for field crews involved in the mapping process. Each of the above methods has drawbacks that make them infeasible for efficient and precise mapping of Phragmites. To overcome these issues, I propose an approach based on a small Unmanned Aerial System (UAS) capable of collecting geo-referenced high-resolution multispectral imagery from under any cloud layer. In particular, our team deployed a PrecisionHawk Lancaster (PHawk) UAS with a MicaSense RedEdge (MSRE) camera. The MSRE simultaneously captures five discrete spectral bands (Blue, Green, Red Edge and Near Infrared) with 8 cm ground sample distance (GSD) at 120 m altitude with a radiometric resolution of 12-bits per pixel per band. Digital Surface Models (DSMs) are produced from this 5-band imagery using photogrammetry techniques to get the elevation information of surface objects.
Traditional statistical classification approaches often fail to yield adequate results with HSI because of the high dimensional nature of the data, multimodal class distribution, and limited ground truth samples for training. Over the last decade, SVMs and Multi-Classifier Systems (MCS) have become popular tools for HSI analysis. Random Feature Selection (RFS) for MCS is a popular approach to produce higher classification accuracies. In this study, I have developed a Non-Uniform Random Feature Selection (NU-RFS) within an MCS framework using SVM as the base classifier. I proposed a method to fuse the output of individual classifiers using scores derived from kernel density estimation. This study demonstrated the improvement in classification accuracies by comparing the proposed approach to conventional analysis algorithms and by assessing the sensitivity of the proposed approach to the number of training samples. These results are compared with that of uniform RFS and regular SVM classifiers. The study demonstrated the superiority of Non-Uniform based RFS system with respect to overall accuracy, user accuracies, producer accuracies, and sensitivity to number of training samples.
The approach proposed in this work combines the following methods to create diversity within a pool of classifiers and to ensure that strengths and weaknesses of individual classifiers are incorporated into the final decision making: a) re-sampling features in the data through RFS; b) manipulation of input features through NU-RFS; c) manipulation of output classes through scores computed from Kernel Density Estimation. The approach uses a spectral band grouping to perform NU-RFS and a decision fusion strategy based on kernel density scores. To verify the effectiveness of this approach, we performed experiments to compare overall accuracies of SVM, RFS, NU-RFS, SVM with kernel density fusion, and NU-RFS with kernel density fusion. The sensitivity of the above-mentioned approaches to the number of samples required to train them is also studied in this work.
Jeff Carmical, Mississippi State University - Summer 2020
Ethan Worch, Mississippi State University – Spring 2019, Summer 2019, Fall 2019, Spring 2019
Allyson Espy, Mississippi State University – Summer 2019, Spring 2020
Meilun Zhou, Mississippi State University – Fall 2018, Summer 2019, Fall 2019, Spring 2020, Summer 2020, Fall 2020
Cary McCraine, Mississippi State University – Summer, Fall 2017, Spring, Summer and Fall 2018 (Currently a research engineer at GRI)
Donna Jaison, Mississippi State University – Spring and Summer 2016 (Currently a graduate student at the University of Texas – Arlington)
Preston Stinson, Honors Student in EE, Mississippi State University – 2015 (Currently a graduate student at the North Carolina State University)
Luan Carlos da Silva Casagrande, Universidade Federal de Santa Catarina, Brazil – 2015 (Honors student currently a Machine learning engineer at Agremo Inc)
Bharath Sridhar, Amrita University - 2010 (Currently working at Robert Bosch USA – Mooresville, NC as Embedded System Engineer)
Akram Sheriff, Amrita University – 2010 (Currently working as IoT engineer at Cisco, San Jose, CA)
Daniel McCraine, Dennison Lacomini, and Caleb Lott, “ivWatch Sensor Test Bench System” 2018 (Mississippi State University)
Luan Carlos da Silva Casagrande, “Comparative Study of Image Texture analysis And Machine Learning Methods For classification of phragmites Australis using True-Color High Resolution Images” 2017 (Universidade Federal de Santa Catarina, Brazil)
Sylvania Golla, et.al. “Extracting Cricket Game Summaries via Frame Clustering” 2008 (Amrita University) Project Bundle Presentation
Manoj Kumar, et.al. “Non-Chronological Dynamic Video Abstraction using Rack Through Method” 2009 (Amrita University) Presentation Report
Arunchander Kalyanasamy, et.al. “Video object based Content-Based Video Retrieval system” 2007 (Amrita University) Report Presentation
Kaushik Prakash, et.al. “Resolution Enhancement of Color Video Sequences” 2007 (Amrita University) Project Bundle