SPATIAL DECISION SUPPORT: OPTIMIZING DRONE DATA ACQUISITION FOR NATURAL RESOURCE MANAGEMENT

• Sponsoring Institution: National Institute of Food and Agriculture
• Project Status: NEW
• Funding Source: HATCH
• Reporting Frequency: Annual
• Accession No.: 1015648
• Project No.: WVA00711
• Project Start Date: Mar 1, 2018
• Project End Date: Feb 28, 2023

Project Director
Strager, MI.

Recipient Organization
WEST VIRGINIA UNIVERSITY
886 CHESTNUT RIDGE RD RM 202
MORGANTOWN,WV 26505-2742

Performing Department
Agri Economics

Non Technical Summary
Consumer and commercial drones or UAVs have become very popular in the past 5 years as digital cameras have improved and the platforms for flying them have become less expensive. Beyond just the pictures or video that a UAV provides for hobbyist use, scientists have recognized the benefit of creating products from multiple overlapping images that point directly downward from the UAV. The ability to quickly fly an area at an altitude much lower than an aircraft enables the creation of a very high resolution product that can detect many features that were never possible before with aircraft or satellites. While the unique spatial and temporal benefits are well acknowledged, little is known about specific flight planning and operation decisions to be more efficient in using this tool for improved management and analysis of natural resources. This has created a barrier for entry into applied research by scientists which has prohibited them from integrating this technology into their specific expertise. Our research effort will enable more users to confidently and successfully integrate UAV technology into their management and analysis of resources by providing specific flight planning expertise and selection of equipment to complete mapping and analysis projects with UAV technology. We will develop optimal flight plans and protocols for UAV selection, camera system, battery source, flight paths, time requirement to capture data, number of targets, use of real time kinetic GPS, time of day for light display, post image processing, and image analysis to develop products to aid monitoring and management. The basic methods and approaches we will use to answer these implementation questions will be determined from research projects in two main areas. The first is a forest inventory and analysis at the WVU Research Forest in which we have started the collection of imagery over different seasons and in varying terrain. Second, is an application of UAVs in capturing watershed and riparian information. These two study areas will provide us with the base information from the UAV to compare with on the ground traditional sampling information on forest plots and stream/riparian hydrological parameters. The ultimate goal of our research will be to provide a cost effective plan for incorporating UAV technology to better measure, analyze, and manage two natural resource areas - forest inventory and watershed riparian and corridor extents. The impact from this work will help to grow the use of UAV technology for natural resource specialists and managers to more effectively and efficiently address the issues they encounter at their work specifically for the spatial data interactions.

Animal Health Component

0%

Research Effort Categories

Basic

40%

Applied

30%

Developmental

30%

Classification

Knowledge Area (KA)	Subject of Investigation (SOI)	Field of Science (FOS)	Percent
605	7210	3010	70%
123	7299	2050	10%
404	7410	2060	20%

Knowledge Area
605 - Natural Resource and Environmental Economics; 404 - Instrumentation and Control Systems; 123 - Management and Sustainability of Forest Resources;

Subject Of Investigation
7299 - Research equipment and methods, general/other; 7410 - General technology; 7210 - Remote sensing equipment and technology;

Field Of Science
2050 - Hydrology; 3010 - Economics; 2060 - Geography;

Keywords

unmanned aerial vehicles

resource economics and management

Goals / Objectives
The goal of our optimization approach will focus on the flight planning and design for natural resource applications with UAVs. There are multiple objectives in this process that include maximizing aerial coverage, minimizing number of photos, maximizing image resolution, minimizing processing time, maximizing flight time to battery constraints, choosing the correct UAV platform, minimizing control ground points, and choosing an optimal flight pattern. All of these objectives are inherently built into our optimization approach. The choice variables for these objectives will be derived from field work applications.Our two resource applications will be in forestry and water resource management and monitoring. These two contexts will be autonomous and independent even though a study site may have involve both contexts. Our conceptual model is to develop a procedural approach to guide the optimal use of drone use in natural resource applications. This will be a function of the application (forestry or water resources), platform, imagery sensor, study area extent, battery power constraints, atmospheric conditions and constraints, resolution or scale of output desired, lens optics, GPS requirements, distribution of targets and number, leaf on or off vegetation, line of sight constraints, flight speed, amount of image overlap, direction of flight lines, and processing needs.The project will involve the testing of drone imagery acquisition and analysis for two types of natural resource management with the overall goal to develop an optimization framework to most effectively and efficiently capture the information needed for implementation.Objective 1. The first will be to compare traditional forest field sampling and measurement techniques to that captured with drone imagery to determine the scale, extent, and process that is optimal. Forest field sampling is interested in using on the ground plots and measurements to determine species, heights, and volumes. Drone imagery can be acquired and processed to create orthophotos which are aerial photos capable of making accurate measurements. The crown extent, size, and species can be identified with appropriate analysis techniques such as segmentation classification. The main question our first objective will attempt to answer will involve optimal size of study area, ideal flying height to resolution ratio, and what ideal spectral band(s) for imagery collected in order to produce results comparable to those obtained from traditional field plot sampling by foresters on the ground.Objective 2. The second objective will have an aquatic focus on riparian areas and instream conditions. As noted by Michez et al. (2015) multi-temporal and hyperspectral imagery from drones can be used as a framework to explore and differentiate deciduous riparian forest species and health conditions. In addition, Dietrich (2014) demonstrated the ability to analyze a fluvial system for geomorphology and to help explain the downstream patterns in structure with high resolution imagery. Our approach will focus on riparian vegetation, stream sinuosity and structure (gradient, pool/riffle ratios), and water quality with band filters that use the Normalized Difference Water Index (NDWI) (Stevens et al., 2003). The NDWI has been shown to be strongly related to the plant water content and can be a good proxy for plant water stress (Gao, 1996). The main objective will be to determine the feasibility and applicability of using drone imagery and different band combinations to effectively characterize watershed conditions linked to riparian vegetation and channel morphology to predict water quality.

Project Methods
The following specific studies are planned with additional potential study areas located throughout the Mid-Atlantic Highlands. Our first study area site is the WVU Research forest (Monongalia and Preston Counties, WV). For 5 different plots that are 80 to 100 ha in size, we will fly the area with true color and near infrared over four different seasons to capture varying phenology. Traditional forest plots will be established and sampled to capture species, heights and volume estimates. The imagery will be processed to create varying resolutions for canopy delineation and subtracted from ground elevation to determine tree heights. Comparisons will be done across seasons and different imagery types to determine optimal capture and management.The second study area site is the Reedsville Farm (Preston County, WV). We plan to capture riparian vegetation during different phenology over an approximate 800m section of stream reach on the farm property using a DJI phantom drone with interchangeable camera systems. The imagery will be captured to produce an orthophoto capable of centimeter nominal pixel resolution. Image processing will be conducted with Pix4D and Agisoft software and classification with ENVI and Ecognition. Different band combinations will be explored to create true color, near infrared, normalized difference vegetation index, and the normalized difference water index. We will compare the UAV data to field data surveyed at the same sites to compare and analyze for accuracy, efficiency, and cost.From the collection of this imagery over different areas, seasons, camera systems, and UAV flying parameters, we will explore the question of what are the optimal operational parameters to achieve desired results for various resource management goals. As one example, we will determine the preferred parameters for creating stream corridor and riparian elevation models from the UAV imagery capable to the rapid Physical Habitat Assessment (rPHA) level. To answer this question, one approach could be to apply a value of information economic analysis to determine the marginal utilities for each of the parameters. Another may be to incorporate optimal control theory to find the flying parameters that minimize a cost function which for our stream riparian model data collection example could be flying time of the UAV. These are just two of the many applied economic and optimization tools available that we will explore toward our overall goal to provide spatial decision support to drone data acquisition for natural resource management. Once we build models which provide insights into these decisions, we will explore the creation of a software package or app which will allow the flight planner to key in variables and output flight parameters to optimize the mission. We have already had conversations with the Launch Lab at WVU regarding how to market and target our idea to aid which we feel will greatly aid the rapidly growing UAV industry.

Progress 10/01/19 to 09/30/20

Outputs
Target Audience:The US Fish and Wildlife Service National Conservation and Training Center was the recipient of a Waterhed Analysis and Hydrological Modeling online class I taught for them covering UAV drone based runoff analysis modeling aproaches and other watershed analysis. The class was taught from June 1-11, 2020. Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided?During this past year, I provided instruction for the US Fish and Wildlife Service National Conservation Training Center (USFWS NCTC). For the USFWS NCTC, I co-taught two online courses on Watershed Analysis and Hydrological Modeling to professionals from the USFWS and USGS across the country. This experience helps me in teaching and research to make sure I am providing relevant and timely instruction to those working with spatial and natural resource issues. How have the results been disseminated to communities of interest?Results have been disseminated to communities of interest in the publications and presentations at conferences as noted earlier in this report. What do you plan to do during the next reporting period to accomplish the goals?The next reporting period will focus on objective two regarding UAV use and applications for aquatic riparian areas and instream conditions. We have identified the Holly River as a site in which to pursue use of the technology to map habitat and provide guidance on instream structures to improve fish habitat conditions.

Impacts
What was accomplished under these goals? The goal of this project is to better understand the optimal spatial and procedural approaches for applying appropriate drone technologies under different natural resource management objectives to assure cost effective and sustainable operations. We proposed to investigate this main question with specific applications of drone technology in the areas of forest management and water resources. In the first year we focused mostly in the area of forest management. We determined optimal flying heights and amount of side and end lap coverage for a mixed hardwood stand to extract canopy cover and differentiate species. We also investigated the resolution necessary to detect tree insect damage to leaves and explored forest uniformity estimates using structure from motion point clouds generated from an aerial drone system. Highlights: We were able to produce an efficient and approachable work flow for producing forest stand board volume estimates from UAV imagery in a mixed hardwood stand of West Virginia. Using UAV imagery we were able to highlight the importance of combining spectral and 3D information to evaluate forest health features -specifically from Cicada impacts to deciduous trees. We applied 3D point clouds generated by images obtained from an UAV to evaluate the uniformity of young forest stands and found them close to values obtained in the field. In the second year, we demonstrated the use of UAV for energy extraction and landscape analysis. We showed how UAVs can be applied to collect valuable information during the construction of oil and gas well pads. The aerial photos from a UAV provide high spatial and temporal information and when built as orthomosaics and digital surface models, can map the sources and paths of runoff for better sediment management. The objective of this work was to use UAV imagery to predict total suspended solid runoff resulting from the construction of an oil and gas well pad. We collected true-color images with a Phantom 4 Pro and processed the imagery to create an orthomosaic and a digital surface model (DSM). The orthomosaic was classified using a maximum likelihood classifier to determine cover types and the DSM was processed to create a runoff grid. Annual total suspended solid (TSS) loading rates were assigned to the cover type classes and modeled with an annual runoff grid. We were able to identify three locations with elevated TSS loading values that could be detrimental to downstream aquatic systems and suggest the use of mitigation activities at these sites to prevent potential impacts. The use of UAVs in the oil and gas industry proved to be a viable option for better sediment management during pad construction.

Publications

• Type: Book Chapters Status: Published Year Published: 2019 Citation: Strager, M .P. 2019. GIS&T and Natural Resource Management. The Geographic Information Science & Technology Body of Knowledge (4th Quarter 2019 Edition), John P. Wilson (ed.). DOI: 10.22224/gistbok/2019.4.2
• Type: Journal Articles Status: Published Year Published: 2020 Citation: Pourmohammadi, P., D. Adjeroh, M. P.Strager, Y. Z. Farid. 2020. Predicting developed land expansion using deep convolutional neural network. Environmental Modeling and Software. Vol 34, 104751. https://doi.org/10.1016/j.envsoft.2020.104751
• Type: Journal Articles Status: Published Year Published: 2020 Citation: Strager, M. P., A. M. Klein Hentz, P. Kinder, S. Grushecky. 2020. Using unmanned aerial vehicles to model surface runoff during well pad development. Journal of the American Society of Mining and Reclamation. ISSN Number 2328-8744. Vol 9, No. 1.pp. 51-69.
• Type: Journal Articles Status: Awaiting Publication Year Published: 2021 Citation: Oliver, M., M. P. Strager. 2021. A Spatial Analysis of high and low farmer participation in the USDA NRCS Conservation Technical Assistance Program. Journal of Water and Soil Conservation. IN PRESS.
• Type: Journal Articles Status: Under Review Year Published: 2021 Citation: Pourmohammadi, P., M. P. Strager, M. Dougherty, D. Adjeroh. Analysis of Urban-Rural Land Development Drivers Using Multi-Source Spatial Data and Regression Models, Journal of Applied earth observation and geoinformation, IN REVIEW.
• Type: Journal Articles Status: Under Review Year Published: 2021 Citation: Ajanaku, B., M. P. Strager, A. R., Collins. GIS-based Multi-Criteria Decision Analysis of Utility-Scale Wind Farm Site Suitability in West Virginia. GeoJournal. IN REVIEW

Progress 10/01/18 to 09/30/19

Outputs
Target Audience:The target audience that will be most interested in our efforts during this reporting period are natural resource specialists who use UAV technology for forest management. Our delivery of the results are in peer-reviewed publications and presentations at technical conferences. Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided?As part of this project we were able to offer a class through WVU called Drones in Natural Resource Management. This was an intensive 3 week course introducing drone system flight operations, FAA compliance, and the remote sensing of natural resources. This course covered topics in safe and effective drone flight operations and hands-on flight experience, knowledge about FAA aviation rules and regulations for drones, drone-based sensors (cameras, thermal, hyperspectral, lidar, NIR, NDVI), drone applications in natural resources, agriculture, energy environments, flight planning, mission execution, and data processing. This course was taught to 13 students in the summer of 2019. How have the results been disseminated to communities of interest?Results have been disseminated to communities of interest in the publications and presentations at conferences as noted in this report. What do you plan to do during the next reporting period to accomplish the goals?The next reporting period will focus on objective two regarding UAV use and applications for aquatic riparian areas and instream conditions. We have planned to fly a pipeline as part of a recently awarded project with the US Department of Transportation Pipeline Safety and Management program. Our flights will include data on potential leaks as well as landscape surface conditions to identify slips and areas of pipe stress.

Impacts
What was accomplished under these goals? During this past reporting period, we applied UAV aerial imagery and LiDAR data to map and track overland runoff at a gas well pad in West Virginia. A maximum likelihood classifier was used to classify the imagery and highlight disturbed areas of runoff. Next, we incorporated a landscape-based runoff model for total suspended solid (TSS) estimates to help identify sediment management locations. We found this technology provided very site-specific high-resolution information in a timely manner to effectively identify runoff sources, extent, and water quality from the well pad. The predicted TSS estimates indicated a potential risk to downstream biological conditions highlighting the importance and utility of this approach to monitoring such pad sites for this industry.

Publications

• Type: Conference Papers and Presentations Status: Accepted Year Published: 2019 Citation: Strager, M. P., P. Kinder, J. M. Strager, S. Grushecky, J. Kimmet. Applying UAV Imagery to Minimize Impacts to Surface Water from Oil and Gas Development. American Society of Mining and Reclamation, Big Sky, MT, June 3-6, 2019. Pourmohammadi, P., D. Adjeroh, M. P. Strager. Predicting Impervious Land Expansion Using Deep Deconvolutional Neural Networks. International Geoscience and Remote Sensing Symposium, Yokohama, July 28-Aug2, 2019. Pourmohammadi, P., M. P. Strager. An Application of High Performance Computing (HPC) and Deep-Learning in Predicting Developed Land Expansion in an Appalachian Watershed. American Association of Geographers, Washington, DC, April 3-7, 2019. Cribari, V., M. P. Strager. Assessing Long Term Changes and Landscape Disturbances in the Headwaters of Big Coal River. Nature and Society Facing the Athropocene Challenges and Perspectives for Landscape Ecology, International Association for Landscape Ecology, Milan, Italy, July 1-5, 2019. Strager M. P., P. Kinder, S. Grushecky. Modeling Potential Runoff from a Converted Forest lot Using UAV Imagery. 12th Southern Forestry and Natural Resource Management GIS Conference, Athens, GA, Dec 9-10, 2019. Strager, M. P. 2019. High temporal and spatial resolution mapping of land cover to support water research in Appalachia. Geo EDF Stakeholder Workshop, Purdue University, West Lafayette, IN. Oct 6-8, 2019. Strager, M. P., A. E. Maxwell. Large-area, high spatial resolution land cover mapping using random forests, GEOBIA, and NAIP orthophotography: findings and recommendations. Appalachian Freshwater Initiative Meeting, Huntington, WV, June 11-13, 2019. Strager, M. P., N. Zegre. A Two-Scaled Approach for Flood Susceptibility Prediction in Appalachia. Appalachian Freshwater Initiative Meeting, Huntington, WV, June 11-13, 2019.
• Type: Journal Articles Status: Accepted Year Published: 2019 Citation: Strager, M .P. 2019. GIS&T and Natural Resource Management. The Geographic Information Science & Technology Body of Knowledge (4th Quarter 2019 Edition), John P. Wilson (ed.). DOI: 10.22224/gistbok/2019.4.2 Maxwell, A. E., M. P. Strager, T. A. Warner, C. A. Ramezan, A. N. Morgan, C. E. Pauley. 2019. Large-Area, High Spatial Resolution Land Cover Mapping using Random Forests, GEOBIA, and NAIP Orthophotography: Findings and Recommendations. Remote Sensing, Vol. 11, 1409, 1-27, https://doi.org/10.3390/rs11121409

Progress 03/01/18 to 09/30/18

Outputs
Target Audience:The target audience that will be most interested in our efforts during this reporting period are natural resource specialists who use UAV technology for forest management. Our devlivery of the results are in peer-reviewed publications and presentations at technical conferences. Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided?As part of this project we were able to offer a class through WVU called Drones in Natural Resource Management.This was an intensive 3 week course introducing drone system flight operations, FAA compliance, and the remote sensing of natural resources. This course covered topics in safe and effective drone flight operations and hands-on flight experience, knowledge about FAA aviation rules and regulations for drones, drone-based sensors (cameras, thermal, hyperspectral, lidar, NIR, NDVI), drone applications in natural resources, agriculture, energy environments, flight planning, mission execution, and data processing. This course was taught to 15 students in the summer of 2018. How have the results been disseminated to communities of interest?Results have been disseminated to communities of interest in the publications and presentations at conferences as noted earlier in this report. What do you plan to do during the next reporting period to accomplish the goals?The next reporting period will focus on objective two regarding UAV use and applications for aquatic riparian areas and instream conditions. We have planned to fly an oil and gas well pad to estimate impervious surface extent and assign loading values for total suspended solids. Next, we will use the UAV data to create a structure from motion elevation surface to track the runoff across the pad site to receiving stream to estimate pollutant load and concentration impacts to aquatic resources.

Impacts
What was accomplished under these goals? The goal of this project is to better understand the optimal spatial and procedural approaches for applying appropriate drone technologies under different natural resource management objectives to assure cost effective and sustainable operations. We proposed to investigate this main question with specific applications of drone technology in the areas of forest management and water resources. In this first year we focused mostly in the area of forest management. We determined optimal flying heights and amount of side and end lap coverage for a mixed hardwood stand to extract canopy cover and differentiate species. We also investigated the resolution necessary to detect tree insect damage to leaves and explored forest uniformity estimates using structure from motion point clouds generated from an aerial drone system. Highlights: We were able to produce an efficient and approachable work flow for producing forest stand board volume estimates from UAV imagery in a mixed hardwood stand of West Virginia. Using UAV imagery we were able to highlight the importance of combining spectral and 3D information to evaluate forest health features -specifically from Cicada impacts to deciduous trees. We applied 3D point clouds generated by images obtained from an UAV to evaluate the uniformity of young forest stands and found them close to values obtained in the field.

Publications

• Type: Journal Articles Status: Published Year Published: 2018 Citation: Hentz, A. M. K., C. A. Silva, A. P. Dalla Corte, S. P. Netto, M. P. Strager, C. Klauberg. 2018. Estimating forest uniformity in Eucalyptus spp. and Pinus taeda L. stands using field measurements and structure from motion point clouds generated from unmanned aerial vehicle (UAV) data collection. Forest Systems, Vol 27, Issue 2, e005. https://doi.org/10.5424/fs/2018272-11713
• Type: Journal Articles Status: Published Year Published: 2018 Citation: Hentz, A. M. K., M. P. Strager. 2018. Cicada (Magicicada) tree damage detection based on UAV spectral and 3D data. Natural Science, Vol. 10, (1). 31-44.
• Type: Journal Articles Status: Published Year Published: 2018 Citation: 45. Liebermann, H., J. L. Schuler, M. P. Strager, A. K. Hentz, A. E. Maxwell. 2018. Using unmanned aerial systems for deriving forest stand characteristics in mixed hardwoods of West Virginia. Journal of Geospatial Applications in Natural Resources. Vol 2, (1). 23-52.