CICEET Progress Report for the period 9/02/07 Through 2/29/08

Project Title: Vegetation, Impervious Surfaces, Soils, and Topographic Analysis Tools: Geospatial Technology to Promote Coastal Water Quality
Principal Investigator(s): Thomas R. Allen
Project Start Date: 1 September 2007

Figures

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Tasks to meet objectives
a. Spatial Database Development
The project has developed an extensive spatial database for the two intensive study areas: 1) Town of Southern Shores, Dare County, NC and 2) Camden County, NC (See Figure 1). Acquired data include LIDAR DEMs, orthophotography, multiple imagery, and GPS data for water quality, wells, and field surveys. In addition, our team has expanded to include a hydrologist and university colleague of the PI working in the City of Greenville, Pitt County, NC. For each jurisdiction, we have compiled all the necessary spatial data for further analysis. Derivative layers and model output, however, will require further field work and validation (e.g., impervious surface extent, runoff, flooding.)

b. Collaboration Team
The project has created a dedicated team at ECU, including graduate research assistants enrolled in the Geography MA program at East Carolina University (Ms. Carrie Tragert and Naomy Perez-Sanchez), a collaborating hydrologist (Dr. Michael O’Driscoll, ECU Department of Geological Sciences), and an undergraduate research assistant environmental biology major (Ms. Haley Cleckner.) The external participants in the project include Dean Carpenter (Science Coordinator, US EPA Albemarle-Pamlico National Estuary Program), Mr. Robert McClendon (UNC Coastal Studies Institute, Manteo, NC), Mr. Dan Porter and Ms. Amy Durden (Director of Planning and GIS Technician, respectively, County of Camden, NC), and Mr. Mike Hejduk (Inspections and Permits Officer, Town of Southern Shores.)

c. Project Management and Design
The first phase of the project emphasized data acquisition and integration and the production of impervious surface data from disparate and cost-effective sources for high-resolution planning. The hiring of the graduate students was underway at the beginning of the funding arriving at ECU, which was too late to recruit a student with the ideal set of technical skills. However, hiring two students (alternating in funding on the assistantship) has allowed sufficient progress. The addition of the expanded members of the collaboration team required some adjustment to the project design, opportunistically allowing the inclusion of a hydrologist and corresponding field data, which would strengthen the geospatial modeling of storm water runoff, water quality, and flooding assessment.

Progress on Tasks
a. Spatial Database Development
All requisite geospatial data have been acquired and co-registered to enable further spatial analyses. The primary data included those already in use by the partners, parcels, topography, streets, and limited extent of ditches or storm water areas (most notably spatial data for problem areas.) The ALOS PALSAR satellite data have been integrated and analyzed for the production of impervious surface maps. Results from this image classification highlighted the need to incorporate the SAR imagery with the multispectral imagery. Accuracy assessment is not complete, but preliminary maps show improvement of impervious area estimates for some infrastructure (especially residential areas.) In addition, to gauge the success of integrated analysis, separate classifications are being finalized using off-the-shelf sources (digital orthophotography) and two purchased data sources (ASTER and IKONOS). The IKONOS imagery was ordered and provided to specifications from GeoEye for Southern Shores. Impervious surface mapping will be completed for Southern Shores by mid-March and underway for Camden through May. Prior to our analysis, neither community has a high-resolution impervious surface map nor been able to analyze the impacts of impervious surface on local-level runoff problems (other than field-based investigation.) Southern Shores’ problems are severe and extensive and elicited external study by engineering contractors (at substantial expense to the town.) Our approach is to leverage the value of A) North Carolina’s excellent floodplain mapping program LIDAR-based DEMs and B) available digital terrain and hydrologic modeling tools. The Town has only used the Lidar DEMs for cartographic use and no analyses were ever conducted by the town for runoff. However, monitoring well data have been acquired. We also conducted a field mapping campaign to assess the depth to the water table using resistivity and conductivity measurements. These results confirm the effect of subsurface peat layers that may limit infiltration or exacerbate flooding from upslope impervious areas and throughflow.

b. Collaboration Team
Vigorous collaboration is ongoing by the PI, including field work for accuracy assessment of the impervious classifications and the hydrologic modeling output. The PI has met with each collaborative group, and a concept for a workshop has been proposed (before June). In addition, Jacquie Ott, NCNERR GIS analyst, has provided suggestions and identified opportunity to work in the NCNERR Currituck Banks and Kitty Hawk Woods sites.

c. Project Management and Design
The basic project design has been expanded to include the collaboration of hydrologist Michael O’Driscoll and his field techniques and modeling expertise. The collaboration comes with no-cost and brings quantitative understanding of subsurface peat and groundwater processes into the project. Otherwise, the project design remains fundamentally unchanged from the proposal. The PI has also met individually with each participant, and the ECU collaborators have jointly conducted field work and meetings twice with each locality to date.

Difficulties
a. Processing PALSAR Synthetic Aperture Radar Data
The exploratory analysis of the ALOS PALSAR satellite data presented a series of moderately difficult processing challenges. The technical issues have been resolved, as the image provider (in the US, the Alaska Satellite Facility at U. of Alaska Fairbanks) has provided a streamlined tool for search and retrieval. Data format issues and metadata have been resolved (whereas the project outset was confronted with independently resolving these, with the source imagery having from directly from the Japanese Space Agency, JAXA). From a processing standpoint, the PALSAR data appear to have difficulty accurately producing impervious surface layers without additional multispectral or other sources of information. The PALSAR data, for example, will backscatter intensely off developed features such as metal rooftops or powerlines, but they will be diffusely backscattered if a forest canopy covers such features. Hence, the PALSAR data are being relegated to a distinctly subordinate role. Figure 3 illustrates this effect and a partial resolution to it. For this area in Camden County, the rooftops are clearly evident in the 5 meter resampled images. This was achieved for this area by enhancing the dynamic range of the visualized image. The next step is to automate the PALSAR image enhancement to augment the spectral classification of the impervious surface from other imagery (ASTER or IKONOS).

b. Town Collaboration
A very strong collaboration with the Town of Southern Shores was affected by abrupt departure of the town code enforcement officer who was assisting with field data collection and knowledge of storm water problems. After some weeks, collaboration was reestablished with the town engineer and a memorandum of understanding reached to cooperate on sharing data and providing access to collect data from wells.

Data Generated to date
The project has developed new high-resolution maps of impervious surface area specific to the intensive study areas. These maps are undergoing accuracy assessment and evaluation for input into a hydrologic model in the GIS. Overland runoff and flooding problem areas were identified by the Town of Southern Shores and a contractor prior to our project startup. These chronic problem areas are also important for storm water pollution contamination of the shallow aquifer and fresh groundwater lens as well as the ocean and estuary. We digitized the problem areas provided by the Town and conducted a first cut hydrologic runoff model within ArcGIS using TauDEM software. The derived flow accumulation potential strongly coincides with most problem runoff areas (see Figure 2) The channelized runoff from small basins of impervious surfaces dominates the predicted runoff paths and drainage networks. Roads (not shown in the image) coincide spatially with the predicted runoff in the model. Since the site has virtually no stormwater prevention measures (swales, culverts, and curb and gutter infrastructure is rare), the roads are functionally channelizing overland flow and delivering the excess precipitation and accumulated surfactants to lower topography.

The production of impervious surfaces has attracted significant attention in the remote sensing community, and our analyses have adjusted to incorporate the latest results in the literature since proposal preparation. These changes include the adoption of two widely available remote sensing datasets, high-resolution ASTER multispectral imagery (affordable from NASA and USGS Eros Data Center at 15m spatial resolution), very high resolution IKONOS imagery (GeoEye with 0.8m pixel resolution), and updated orthophotography for Dare County (aerial orthophotography is the de facto standard for impervious surface mapping but typically requires labor-intensive heads-up digitizing or analysis with costly and prohibitively complex and costly object-based analysis software, for local GIS and planners).

Further results are shown in Figure 3, which highlights the use of low-cost PALSAR imagery for mapping residential imperviousness by contrast enhancement of radar backscatter for individual homes. This low-cost imagery is also moderately high resolution (5 m pixel resampling shown) and may be acquired during all weather using the active sensor. These may assist accurate mapping of highly reflective microwave signals from urban infrastructure, even below the resolution of the actual pixel size.

In contrast, Figure 4 shows the standard result of unsupervised classification using a high-resolution multispectral satellite image (IKONOS.) This image was rapidly acquired and produced, and the desktop GIS software is capable of mapping. However, the classification tool can bear processing overhead, and the discrimination between impervious surfaces and bare sand (note the beaches) is of concern.

Even more specific user targeting of sample sites (referred to as “training sites”) using supervised classification can blend the benefits of rapid and cost-efficient mapping and reduced ambiguity and error for impervious vs. bare cover types (see Figure 5) This image shows the same IKONOS scene classified after user specification of impervious training sites. This process can be automated and is easily transferrable.

Project Objectives for Next Reporting Period

Objectives
The project will have completed the impervious surface classifications and integrated the multispectral high resolution imagery and active PALSAR synthetic aperture radar imagery into a unified model and toolbox. The topographic hydrologic modeling will have produced a complete model using the impervious surface and LIDAR derived DEM for the Southern Shores study area. In the Camden study area, the DEM will be hydrologically corrected for ditches, and this process of “stream burning” automated into an ArcGIS modelbuilder tool. The protocol for the subsequent integrative analysis and modeling effort (the focus of year 2) will also be developed. The modeling technique is still under consideration (Python scripting, VBA, or suite of application extensions for ArcGIS). Towards this, the PI and research assistant may undertake additional programming training and assemble documentation in preparation. To gauge our progress toward meeting end-user requirements, we will also host a workshop with collaborators in the field in late spring or early summer. Separate meetings with individual collaborators will assess the need for training in tool use.

Work Plan to Meet Objectives
The work plan will involve three overlapping efforts:

• Site visits to assess the accuracy of the automated high resolution impervious mapping, the topographic runoff indices, and the associated vegetation and soils layers that will be integrated in year 2.
• Additional community stakeholder and model end-user engagement. Two workshop activities are planned. The first workshop will be in Kinston, NC, in association with the “Innovative Planning Tools for Communities” effort of the Southeast Regional Partnership for Planning and Sustainability (SERPPAS). The first project specific workshop is under planning for delivery by the end of June. The PI and assistant will prepare the new datasets (Camden County hydro-corrected DEMs) for analysis in Camden County. This workshop will present preliminary results, data products, and seek to collect further needs assessment and suggestions for the tools and supportive training and materials.
• Modeling and tool development will require the majority of year 2 work. The algorithm development and planning in this phase will necessarily precede this.

Expenditures
Expenses are in the range expected, with one exception. ECU changed its policy on the distribution of student tuition and now requires grants not pay tuition but instead pay the student a proportionally higher salary, where applicable. The result will effect a change in the indirect cost calculation and distributions. This will require reallocation within the grant, and this is requested. Since the policy changed after the grant was awarded, the PI is negotiating with the Sponsored Programs office to not effect any net change on the grant. The graduate assistants on the project are not affected in the first year. The PI requests approval by CICEET for reallocation of the tuition funds toward summer salary and year two personnel support.

Errata
This project has attracted the interest of the NC Sea Grant, NC State University Cooperative Extension/WEMO (formerly NC NOAA NEMO), the SERPPAS sustainability initiative, and the NC ArcUser Specialty Group through meetings, presentations, and word-of-mouth. The development of geospatial tools that could hold promise for dually addressing water quality and water quantity (i.e., flooding) has been a particularly noteworthy commentary. The highest priority is given to coastal natural hazards in NC, particularly hurricanes. The potential for this tool to simultaneously address the prevention of stormwater flooding from small, rapidly urbanizing catchments as well as the mitigation and restoration of water quality is exciting to people. Further, there has been an intriguing issue of scale revealed in this project’s early phase. Several professionals and practitioners in communities have stated to the PI and team that it is refreshing to have tools developed for micro-level, grassroots users (e.g., individual planners, GIS technicians, or code enforcement officers), as opposed to broad, coarse-scale tools that utilize low-resolution data and have uncertain applicability to local problems.

Field work with the community collaborators has yielded in additional revelations. It was understood that the study area was developed atop a relict barrier island spit. The extent to which this affects the surface runoff and groundwater was astounding. It is widely believed that sandy barrier islands would be unlikely to have stormwater runoff problems, but the pattern of development in ridge-and-swale topography has led housing to prioritize higher elevation ridges (runoff source areas) and roads to occupy linear, swales. The result is that impervious surfaces have come to dominate the upper catchments, with driveways and roads functioning as conduits for runoff, and swales collecting the runoff. Since swales are also shallow to depth of the water table, any antecedent conditions leading to a higher water table can rapidly induce flooding. In some areas, the swales are apparently nearly impermeable, so any runoff collecting has extremely slow percolation as a result of buried peat (marshes) (see Figure 6). This was anecodotally revealed to us by a homeowner digging a swimming pool and having excavated 6 feet of peat before reaching sand. This pattern of development continues, yet there are community participants in our project that understand the underlying, insurmountable geologic problem. Fortunately, the combination of our geographic, hydrologic, and community involvement will be communicating these issues to the residents and decision-makers. The trajectory being laid out is thus- runoff problems are not going away of themselves, there are solutions to spatially target and retrofit the stormwater runoff (this tool will dramatically aid that targeting), and this will by and large be more cost-effective, sustainable, and environmentally friendly than alternative massively engineered solutions.