Lidar and sub-meter satellite stereo imagery for Lake Erie shoreline mapping
Project Number: R/CE-010, Progress Report
Start Date: 3/31/2007
Completion Date: 2/28/2010
Revision Date: 10/20/2009
| Principal Investigator(s) | 1. | Rongxing Li, * |
| This shows the current affiliation and may not match affiliation at time of participation. * | ||
Funding Record
| Source: Ohio Sea Grant College Program | |||
| Source Fund | State Match | Pass Through | |
| Total | $ 179,952.00 | $ 90,148.00 | $ 0.00 |
Objectives
This project will conduct an investigation of mapping the Great Lakes shorelines by integrating airborne LIDAR data and the newly available 0.6-m QuickBird images for the improvement of cost effectiveness, accuracy, and efficiency. The objectives of this project are:
- In Year I, to improve and apply techniques for processing LIDAR data (higher vertical and lower horizontal accuracy) and QuickBird sub-meter satellite images (lower vertical and higher horizontal accuracy);
- In Year II, to develop methods for integration of LIDAR data and QuickBird images; and
- In Year III, to use the new integrated method for Lake Erie shoreline mapping with an accuracy of 10 to 30 cm in the vertical and 30 cm in the horizontal directions, and to assess benefits over the traditional aerial photogrammetric shoreline mapping method for the Great Lakes region.
This improved shoreline mapping technology will directly benefit ODNR’s project at the state level for designation of Lake Erie Coastal Erosion Areas using the resulting long- and short-term changes detected from the integrated data sets. The 3-D shorelines that will be produced will provide useful information for monitoring and research programs at Old Woman Creek NERR. Mapping results of this project will also support the Lake County GIS Department in their regulatory coastal management activities (see attached support letters from ODNR, Old Woman Creek NERR, and the Lake County GIS Department).
Abstract
This project will conduct an investigation of mapping the Great Lakes shorelines by integrating airborne LIDAR data and the newly available 0.6-m QuickBird images for the improvement of cost effectiveness, accuracy, and efficiency. The quality of the shoreline derived from the integration of the two data sources will be greatly enhanced. Bluff edges and shoreline protection structures are sharply represented by the LIDAR data. On the other hand, shoreline segments in areas with a small slope, such as beaches, are difficult to delineate using LIDAR data only. QuickBird images with a high resolution can be used in this situation. As a result, an accuracy of 10 to 30 cm in the vertical and 30 cm in the horizontal directions can be achieved for the extracted shoreline, a significant enhancement over the IKONOS satellite imagery mapping capability (2-3 m) and a close approximation of that of aerial photogrammetry (tens of centimeters). Ground control points surveyed by DGPS will be used to verify the achieved shoreline accuracy.
Using integrated method, we will produce new, high-quality shoreline maps along Lake Erie shore at two study sites: Painesville, OH and Old Woman Creek NERR. Because of the advantages of this integrated approach, bluff lines and the coastal protection structures will be mapped at a very high accuracy. Other coastal mapping products such as shoreline maps, DEMs and orthoimages will also be generated. The methods can also be applied to other areas of the Great Lakes region.
This improved shoreline mapping technology will directly benefit ODNR’s project at the state level for designation of Lake Erie Coastal Erosion Areas using the resulting long- and short-term changes detected from the integrated data sets. The 3-D shorelines that will be produced will provide useful information for monitoring and research programs at Old Woman Creek NERR. Mapping results of this project will also support the Lake County GIS Department in their regulatory coastal management activities.
Rationale
Erosion along the Ohio shore of Lake Erie is a serious problem. 95% of Ohio lakeshore is eroding. In some areas, erosion rates are as high as 110 feet in one year. Each year, nearly 1.6 million tons of material is eroded along Ohio’s lakeshore, threatening public safety, health, and welfare. Record-high lake levels in the early 1970s and again in the mid 1980s and 1990s caused extensive damage to residential, commercial, industrial, and agricultural property. Economic losses caused by coastal erosion in the Great Lakes region were estimated at $290 million in 1985 and 1986 and at $9 million in 1985 in Lake County, Ohio (NOAA and ODNR, 1999).
The Ohio Department of Natural Resources (ODNR) identifies Coastal Erosion and Flooding as one of several priority coastal management issues (NOAA and ODNR, 1999). To minimize coastal erosion damages, ODNR was directed to identify coastal erosion areas along Lake Erie shore. The delineation of coastal erosion is conducted by Lake Erie Geology Group under the ODNR’s Division of Geological Survey. Currently, they are updating maps of the Lake Erie Coastal Erosion Areas using 2004 orthophotography. Accurate shoreline positions and variation information are crucial to identification of erosion areas as well as to many other coastal applications, including coastal development, coastal environmental protection, and coastal resource management, and decision making.
Shorelines have never been stable in terms of their long-term and short-term positions. Changes in shorelines are usually caused by either natural processes or human activities. An investigation of the interrelationships between various causes and impacts of shoreline change is necessary before any objective/scientific decisions related to coastal zone policies, engineering projects, and coastal management can be made. Advanced tools such as GIS and modeling/forecasting systems may be used to support decision making. However, these tools depend on the actual data that provide the driving force of such systems. Shoreline mapping and monitoring are key technologies for collecting this vital data.
Shoreline mapping is traditionally carried out using aerial photography and field surveying methods. Despite the high level of accuracy achievable by field surveying and aerial surveys, these methods are expensive and need special logistics and complicated processing procedures. Therefore, the Great Lakes shorelines have not been mapped as frequently as needed. In Ohio, only every ten years, Lake Erie coastal erosion areas are reviewed and may be revised (NOAA and ODNR, 1999). Alternative technologies for shoreline mapping that are more cost effective and of the same or similar accuracy must be investigated.
Through a National Sea Grant-NOAA Partnership project, the OSU Mapping and GIS Laboratory, performed research on the potential of using one-meter IKONOS satellite imagery for shoreline change detection (1999-2001) in collaboration with the National Geodetic Survey (NGS)/NOAA. The project demonstrated that the IKONOS imagery is useful in determining shoreline position at a horizontal accuracy of 1-2 meters. However, vertical position error reached 2-4 m, and the derived shorelines sometimes exhibited an inconsistent vertical profile along the coast. This vertical deficiency is particularly significant when applied to high, eroding bluffs where both horizontal and vertical accuracies are important for coastal change detection, for example, for monitoring farmland discharge into the lakes. Also, the achieved level of accuracy does not satisfy many coastal geological and engineering applications at the state level. Therefore, investigation of an improved, cost effective method, one that is not based on aerial photography, is necessary for shoreline mapping in Great Lakes region.
Recent advances in LIDAR (LIght Detection And Ranging) remote sensing allow for mapping of ground objects at a vertical accuracy of 5 cm-30 cm, although the horizontal accuracy is 2-5 times lower than the vertical accuracy (Ackermann, 1999). On the other hand, the newly available 0.6-meter-resolution QuickBird satellite imagery can be used to improve the horizontal accuracy to 0.3 m (one-half pixel) if both LIDAR data and the QuickBird images are used in an integrated manner. LIDAR data and QuickBird imagery have complementary information useful for coastal applications (Lee and Shan, 2003). LIDAR data are geometric range measurements. Bluff edges and shoreline protection structures are sharply represented by the LIDAR data. Shoreline segments in areas with small slope, such as beaches, are difficult to delineate using LIDAR data only. This problem can be solved with the help of spectral reflectance of the ground measured by QuickBird imagery. Therefore, the combination of these two measurements will provide accurate geometric and spectral information about the coastal ground, which can then be integrated to produce an improved high-quality shoreline.
This project proposes an investigation of mapping the Great Lakes shorelines by integrating airborne LIDAR data and the newly available 0.6-m-resolution QuickBird images for the improvement of cost effectiveness, accuracy, and efficiency. Highly accurate 3-D shorelines will be produced using the resulting long- and short-term changes detected from the integrated data sets. This will directly benefit ODNR (Ohio Department of Natural Resource) in achievement of its goal for designation of Lake Erie coastal erosion areas (see the attached support letter from ODNR). The proposed 3-D shoreline mapping techniques will also provide useful information for monitoring and research programs at Old Woman Creek National Estuarine Research Reserve (NERR) (see the attached support letter from Old Woman Creek NERR). The mapping results of this project will also provide support for the Lake County Planning Commission and its GIS department in their regulatory coastal management activities.
Benefits & Accomplishments
We will produce new, high-quality shoreline maps along Lake Erie shore at the two study sites using integrated QuickBird satellite images and LIDAR data. Because of the advantages of this integrated approach, bluff lines and the coastal protection structures will be mapped at a very high accuracy. Other coastal mapping products such as shoreline maps, DEMs and orthoimages will also be generated. We will publish our methods and results on our project web page and in conference and journal papers. The methods can also be applied to other areas of the Great Lakes region.
The quality of the shoreline derived from the two integrated data sources will be greatly enhanced. Bluff edges and shoreline protection structures are sharply represented by the LIDAR data. On the other hand, shoreline segments in areas with a small slope, such as beaches, are difficult to delineate using LIDAR data only. QuickBird images with a high resolution can be used in this situation. As a result, an accuracy of 10 to 30 cm in the vertical and 30 cm in the horizontal directions can be achieved for the extracted shoreline, a significant enhancement over the IKONOS satellite imagery mapping capability (2-3 m) and a close approximation of that of aerial photogrammetry (tens of centimeters). We will use ground control points surveyed by DGPS to verify the achieved shoreline accuracy.
The research results will directly benefit the ODNR project for designation of Lake Erie Coastal Erosion Areas using the resulting long- and short-term shoreline changes.
Outreach and public education will also be a task of this research project. The PI will collaborate with the Old Woman Creek NERR to provide current, accurate, and research-based shoreline change information in Lake Erie to Ohio's citizens and others interested in Lake Erie and Great Lakes issues. We will take advantage of the tour site and education activities at the Old Woman Creek Reserve and focus on increasing public awareness of the importance of coastal erosion and shoreline change, as well as understanding of the evolution of shoreline mapping techniques and the basic principles of photogrammetry, remote sensing and GIS. Our research results will be integrated in the forms of audio-visual presentations, interpretive field trips and guided tours, guest lectures, educator workshops and technical training seminars.
Accomplishments and future tasks:
1. We have studied the physical and mathematical aspects of an area near Lake Erie that includes Lucas, Ottawa, Sandusky, Erie, Lorain, Cuyahoga, Lake, and Ashtabula counties using a dataset generated from LiDAR data and aerial orthophotos2. We have developed a method for shoreline extraction using LiDAR point cloud data and aerial orthophotos. Firstly, a Mean-Shift Classification algorithm is used for LiDAR point segmentation. The horizontal position and elevation of each LiDAR point along with color information from the orthophoto are used as the point features in the mean shift algorithm. Due to the homogeneous natures of water surface elevation and of color distribution of the water surface, LiDAR points classified as either distributed on the water surface or on the ground using the Mean-Shift Classification algorithm in a semi-supervised manner. Then, a modified Convex Hull algorithm is used to determine the boundary of each classified LiDAR point. The shoreline is then defined as the separation boundary between LiDAR points that belong to the water surface and those that do not. A preliminary experiment in the Lake Erie area using LiDAR data and orthophotos acquired at the same time shows that the accuracy of the derived shoreline is better than LiDAR point spacing.
3. Based on the Mean-Shift Classification algorithm that we previously established, we have added more data types and data sources into the classification procedure in order to improve the accuracy of shoreline extraction. For example, the near -infrared band from the QuickBird satellite imagery is used to determine the water and land separation. Other data types such as geometry of the terrain and point density also can be integrated into the classification procedure. An algorithm to refine the shoreline is also under development. This refinement process is based on using image processing algorithms to find the edges in between the LiDAR point spacing in order to minimize any error and to improve the overall accuracy to be smaller than LiDAR point spacing.
4. In this Mean-Shift Classification algorithm there are three different data sources and more than seven data types. When we try to accommodate all these types and/or sources of data, a normalizing procedure needs to be done. This is the learning procedure in our algorithm. We have developed a systematic way to normalize the scales between data and also to obtain a reasonable parameter for each classification step. Currently this procedure requires human interaction. In the future, however, a statistics-based test procedure will be added in an effort to achieve a fully automatically learning procedure.
5. Blufflines are also an important feature in shoreline mapping. There is a distinct bluffline along Lake Erie in the research area. Currently erosion along this bluffline is a critical problem in Lake County. We have also developed a new algorithm for bluff tops and bluff toes extraction from LIDAR data only. By connecting the LiDAR points using Delaunay triangulation, the surface of the terrain can be formed. Then, the terrain surface is analyzed and the distinct break lines in the terrain model are detected by a geometry-based method. Finally, a continuous bluffline is generated by connecting all these break lines together.
6. In the third year, we will further improve the robustness of the algorithms and test them in a larger area. In addition, we will collaborate with the Old Woman Creek NERR to strengthen public awareness of environmental issues of Lake Erie and the Great Lakes area.
Publications & Media
| Peer-reviewed Publications | |
| Li, R., S. Deshpande, X. Niu, F. Zhou, K. Di, and B. Wu 2008, Geometric Integration of Aerial and High-Resolution Satellite Imagery and Application in Shoreline Mapping Journal of Marine Geodesy. Made available by Ohio Sea Grant as OHSU-RS-406 | |
| Presentations | |
| Lee, I-C., S. Deshpande, X. Niu and R. Li 2008, Coastal Change Analysis Supported by Multi-dimensional Geospatial Data ASPRS Annual Conference, Portland, Oregon | |
| Li, R., S. Deshpande, X. Niu, I-C. Lee, and B. Wu 2008, Multi-dimensional Geospatial Data Integration for Coastal Change Analysis The XXIth ISPRS Congress, Beijing, China | |
| Lee, I-Chieh, Bo Wu, and Rongxing Li 2009, Shoreline Extraction from the Integration of LiDAR Point Cloud Data and Aerial Orthophotos Using Mean Shift Segmentation. ASPRS 2009 Annual Conference | |
| Theses/dissertations | |
| Yunjae Choung 2009, Extraction of Blufflines from 2.5D Delaunay Triangle Mesh using LiDAR data The Ohio State University Master Thesis | |
Supported Students
| I-Chieh Lee (Graduate, M.S.) The Ohio State University |
