Recipient Organization
UNIVERSITY OF NEW HAMPSHIRE
51 COLLEGE RD SERVICE BLDG 107
DURHAM,NH 03824
Performing Department
Natural Resources and the Environment
Non Technical Summary
Riparian habitat and associated water bodies, because of their typically long, but narrow shape, have been difficult to map and monitor from remotely sensed imagery. However, recent technological improvements have provided imagery of greater detail (higher spatial resolution) including unmanned aerial systems (UAS), sometimes called drones, which have allowed geospatial scientists the potential to assess these habitats. This research project evaluates the use of high spatial resolution imagery, including UAS, for mapping and monitoring riparian habitat and its associated water bodies. Imagery recently available from some new satellites (US - Landsat 8 and Europe - Sentinel -2) will be tested to see if these improved spatial resolutions can produce high quality riparian habitat maps. In addition, the UAS imagery will be used to evaluate if mapping of invasive species within the riparian habitat is possible; to assess its potential for monitoring submerged aquatic vegetation (SAV), an indicator of stream health; and finally to monitor stream flow.
Animal Health Component
25%
Research Effort Categories
Basic
50%
Applied
25%
Developmental
25%
Goals / Objectives
The overall goal of this research is to evaluate mapping and monitoring of riparian habitat and stream characteristics from multi-resolution, remotely sensed imagery. There are three specific objectives to accomplish this work:(1) Investigate the level of detail (spatial resolution) required to map riparian habitat effectively and accurately by evaluating Landsat 8, Sentinel-2, and UAS imagery.(2) Evaluate if maps created from UAS imagery can be used to monitor stream characteristics.(3) Communicate the results of this research and engage the appropriate audiences that can most benefit from this information.
Project Methods
Methods for Objective 1:1A. Identify study areas for riparian area analysis.The study areas chosen will be riparian habitats and associated streams within New Hampshire that have available ground information including prior ground samples and stream gauges.1B. Imagery download and collection.The Landsat 8 Operational Land Imager and Sentinel-2 data will be downloaded for April - October for the last 3 years and evaluated to provide the best combination of multiple dates through the growing season to create the best map possible. UAS imagery will be collected to investigate the impact of flying height (altitude) and spatial resolution obtained from both a copter-based and a fixed wing system. Given the cameras on each system, flying height will determine the spatial resolution. Spring imagery and fall imagery will be acquired when phenology may help to better distinguish vegetation characteristics. Finally, the USDA NAIP imagery will be downloaded and used as reference data for comparison with the maps generated from the satellite imagery.1C. Image processing and classification to create satellite imagery mapsThe satellite imagery (Landsat and Sentinel) will be analyzed to create riparian habitat maps using image classification algorithms in the ERDAS Imagine and eCognition software. The Imagine software will be used to create traditional pixel-based maps while eCognition will be used to create object-based maps that have the advantage of grouping the image pixels into meaningful objects before the classification occurs. The results of both approaches will be evaluated using standard thematic map accuracy procedures (step 1E), and the best map selected for each image source.1D. Photogrammetric processing and image analysis to create the UAS imagery maps.Before classification (Step 1C) can be performed on the UAS imagery, the data must be photogrammetrically processed to create an orthomosaic image. This software (Agisoft Photoscan) uses a process called Structure for Motion (SfM). Once completed, the imagery can be classified using ERDAS Imagine and eCognition to create the best riparian map including the identification of any invasive species.1E. Accuracy Assessment.The final step in any image mapping project is accuracy assessment; a process of comparing the map created from the imagery to some reference data source of higher accuracy using an approach developed by Congalton (1991) called an error matrix. Evaluating the maps generated from the satellite imagery, the USDA NAIP imagery will be used for the reference data samples to create the error matrices. The UAS map will be evaluated based on ground-collected information collected by us or from reliable sources.Methods for Objective 22A. Identify study areas.The study areas chosen for objective 2 will be streams that can be readily accessed in New Hampshire. Stream reaches that have had issues with SAV/algae will be a priority but monitoring new outbreaks will also be possible. Finally, streams with a flow gauge are necessary to compare the UAS imagery estimates of flow with the reference stream gauges.2B. Collect UAS imagery of selected stream areasUAS imagery will be collected for the SAV/algae monitoring using both a fixed wing and a copter-based UAS to evaluate the advantages and disadvantages of each system. Transects will be flown at safe heights above all vegetation but at the highest possible spatial resolution that still allows adequate image collection for generating the required orthomosaics needed for the image analysis. An analysis will be conducted to determine this tradeoff between spatial resolution/flying height and UAS platform.Only the copter-based system will be used for the stream flow monitoring since the methodology works best when the sensor is hovering above the stream. Wood chips are tossed into the stream and then images are taken by the hovering UAS. A technique called particle image velocimetry (Lewis et al., 2018) is used to compute the flow. The results will be compared either to existing stream gauge data or a SonTek/YSI FlowTracker will be used to determine the stream flow when the UAS imagery is collected.2C. Collect corresponding ground data on SAV and stream flow.For every sample area collected with UAS imagery, corresponding ground/reference data will be collected or obtained if it already exists. We will conduct a traditional simple quadrant system method for assessing total SAV and collect water samples to quantify total phosphorus and chlorophyll-a concentrations at randomly selected locations corresponding to UAS imagery acquisition. Stream flow measures will be obtained for the dates of imagery collection either from existing stream gauges or by using a flow meter (SonTek FlowTracker).2D. Perform an analysis to compare the ground sample information with that collected by the UAS.Once all the imagery and ground data have been collected, a statistical analysis will be performed to evaluate the capabilities of the UAS imagery compared to the ground reference data. Various spatial resolutions/flying heights will be evaluated along with the capabilities of both the fixed wing and copter-based UAS. This analysis will be a combination of both the statistical results and the practical application of the technology. For example, if the same statistical accuracy can be achieved using either the fixed wing or copter-based UAS, then other considerations such as ease of use, cost to operate, and time in the air will be used to determine the optimal combination. However, if one system is more accurate than the other, that system will be selected as optimal. Once the optimal combination is determined, this configuration will be used for future evaluation of stream characteristics.Methods for Objective 3:3A. Develop materials to communicate relevant information to appropriate audiences.Throughout this research project, materials will be prepared to share important information with those interested in our work. It is important to capture photos, videos and other relevant information in an ongoing manner and not wait until the end of the project to develop these materials.3B. Disseminate this information.When the results are ready, they will be shared in the following ways: (1) as peer-reviewed papers to the academic community, (2) as presentations at local, regional, and national meetings to not only the academic community, but also to land managers, policy makers, and the public, (3) as reports to the NH AES and USDA NIFA as well as College of Agriculture and Life Science Communications, (4) as videos posted on our lab's website and on my YouTube Channel that can then be viewed by anyone who wants to see them, and (5) to individuals that we have interacted with throughout the project who supplied data or allowed land access. In this way, the results of work will be made available to the largest audience possible.