SAMPLE PROJECT OVERVIEWS

Land Use and Impervious Surface Mapping

Anne Arundel County, Maryland

Project Background
Sanborn was contracted by Anne Arundel County, Maryland to provide a land use map and impervious surfaces map of the county from remotely-sensed imagery. The county had an existing land use map that had been produced by hand in the mid-1990s. They also had an updated land use map for the Severn River Watershed (a portion of the county) from a later date. The classification scheme from the later map was to be used for the new classification.

Project Scope
The objectives of the project were twofold: to use automated methods to produce the land use and impervious surfaces classifications and to prove that it could be done both efficiently and cost-effectively. Historically, the practitioners of remote sensing have held that the technology could be used to automatically map land cover, but not land use, because context, one of the factors that determine land use, is difficult to measure using automated methods.

Sanborn used IKONOS imagery (collected April of 2004, pan-sharpened to 1-meter, 4-band) for the classification. The imagery was mosaiced and registered to the county’s DOQQ database using Sanborn’s proprietary Autowarp routine. Impervious surfaces including buildings, houses, parking lots, driveways, and roads were extracted to form the impervious surface database. The imagery was then segmented using the eCognition software package, and models were built to examine the density of those surfaces. With this information and existing information from the 1995 land use classification, the new land use layer could begin to take shape.

Anne Arundel County is home to Annapolis, Glen Burnie, and many smaller towns and other land use types such as agriculture, heavy industry, and several airports. Many areas had changed since the previous land use layer was generated, and the impervious surfaces layer helped to determine the classification of new subdivisions and man-made features. Using the segments generated in eCognition and the patterns and density of impervious surfaces within those segments, new classifications could be made. Buffers were applied to the different land use classes: residential, commercial, industrial, to define the land parcels to which the adjacent land would belong.

Natural land covers, such as forest types, grasslands, agriculture and water, were classified in eCognition as well, using training data developed from the imagery. Agricultural types, Pasture/Hay and Cropland, could often be discriminated by field patterns. The ability to separate agricultural types will depend upon the timing of the imagery collects for any new project.

After the land use layer was complete, the impervious surfaces layer was summarized by the individual land use polygons. This provided an attribute in the land use database with a percent impervious value. This attribute can be used in watershed models to calculate runoff and groundwater recharge values.

Project Technical Specifications

Satellite Imagery: 1 meter IKONOS 4-band (accuracy of c.e. 90, 10 meters)
Land Use/Land Cover: Impervious, vector format

Digital Orthoimagery Database Update

Maricopa County, Arizona

Project Background
In 2005, Sanborn was selected by Maricopa County to update their Digital Ortho Imagery database with the Vexcel UltraCam-D multi-spectral mapping sensor.

Project Scope
The multi-year contract is currently finishing the third year of the program. In 2005, Sanborn completed approximately 5,000 square miles of imagery acquisition and processing at 0.32-foot resolution and 0.8-foot resolutions. The second year, the area was approximately 8,800 square miles at 0.32 and 0.8 foot resolution. The third year, the project consisted of 4,320 square miles at 0.32, 0.5 and 0.8 foot resolution. For the first and third year, the time from acquisition to delivery was less than five months. In all cases, the image count was greater than 25,000 images with the second year exceeding 50,000 images. Since its inception, the project is one of the largest, digitally captured imagery projects directed by local governments. During the final scoping meeting for the first year of the contract, the requirement with regard to resolution and area of interest increased by about five times what was originally proposed. Sanborn was able to meet the County’s needs by mobilizing multiple aircraft for the acquisition phase. The accuracy requirement for the project was +/-5 feet; however, the County’s accuracy review documented an accuracy of about +/-2.5 feet. For the third year, Sanborn deployed SanbornQA, an online QC tool, to allow the 30+ consortium agencies to perform quality control of the imagery for their jurisdictions from one common platform and share comments immediately with consortium team members and Sanborn. The SanbornQA system allowed for quality control of each individual agency’s area without the hassle of distributing interim products prior to final acceptance.

Impervious Surface and Land Cover Mapping

Oahu, Hawaii

Project Background
Sanborn’s role in the Oahu project is to create a highly detailed and accurate impervious surface layer for the entire island of Oahu. Additionally, Sanborn is creating a high resolution land cover layer for the island based on image segmentation. The result is a CCAP map based on higher resolution imagery compared to the medium resolution nationwide land cover. This new and detailed information will be used in assisting coastal resource managers, on Oahu, make more informed, site-specific management decisions.

The mission of the NOAA Coastal Services Center is to support the environmental, social, and economic well-being of the coast by linking people, information, and technology. The Center has been developing moderate resolution land cover products, including impervious surface layers, for coastal areas of the U.S for several years, which Sanborn has had the opportunity to support.

Lidar

Oakland County Michigan

Project Background
Sanborn was contracted to collect and process high accuracy Lidar data for Oakland County Michigan which is located approximately 20 miles NE of Detroit. The County offices located in Pontiac, MI had determined the need for a Lidar survey that encompassed the entire County (~930 sq. mi) as part of a Strategic Plan to “leverage accurate and current location-based data to support decision making in the County”.

During April of 2008, Sanborn flew Lidar missions covering the entire County during the “leaf-off” period. Sanborn also acquired 100 ground truth points in different terrain classes throughout the County. Final data was submitted November 2008.

Project Scope
The project included the Lidar data acquisition, processing and quality control surveys of approximately 930 square miles covering all of Oakland County. The data were collected, processed and delivered in accordance with FEMA’s Guidelines and Specifications for Flood Hazard Mapping Partners – Appendix A. Additional deliverables included hydro-enforced breaklines.

The Lidar data were collected at nominal 1.4 m spacing. These Lidar data were then filtered to remove bridges, buildings and vegetation in order to provide a bare earth data set in the industry standard LAS format. Other deliverables include bare earth data in gridded DEM format, and all returns in .LAS format. Intensity images as well as FEMA-compliant ground truth reports were also provided. All delivered data easily met the horizontal accuracy requirement of 1 meter RMSE and the vertical accuracy requirement of 0.185 meters RMSE (37 cm RMSE in vegetated areas). The bare earth DTM, along with hydro-enforced breaklines collected from stereo intensity imagery was provided in ESRI’s Geodatabase format.

In addition to providing the data Sanborn also provide the County with GeoCue LP360 software for data analysis and visualization. Data demonstrations and software demonstrations were also performed.

Lidar

State of Iowa

Project Background
Understanding the multiple benefits of the a Statewide Lidar project in terms of watershed management and hydrologic analysis, the State Department of Natural Resources set about developing a consortium to fund the program. When a commitment for funding support was received from the USGS, NRCS, State Department of Transportation and other state and federal agencies, the DNR contracted with Sanborn through the USGS GPS-C contract vehicle. The program started in earnest in spring of 2007 and approximately 20,000 square miles were captured that year despite difficult weather and flooding conditions. Five sensors were deployed in the spring of 2008, and almost 15,000 square miles were completed even with the weather conditions and flooding. In addition, Sanborn has acquired two-foot digital imagery for 17 counties in the northwest part of the state. Despite three seasons of extreme weather and flooding, Sanborn managed to maintain the overall schedule and deliver high quality data within the State’s budget.

Project Scope
This program is providing Lidar data for the entire state of Iowa (55,869 square miles). Lidar is to be collected with an average point spacing of 1.4 meters for two distinct specifications: FEMA compliant for 11,050 sq miles and Standard specifications data accuracy for the remaining 44,819 square miles of the state. Deliverables for both products include LAS and Bare-earth ASCII X-Y-Z-i formatted Lidar files. Data collected under this scope of work commensurate with the following:

Accuracy Requirements:

Vertical Bare Earth Accuracy: 18.5cm RMS or better
Vertical in Vegetation Accuracy: 37cm RMS or better
Horizontal Accuracy: 1m RMS
(An independent survey of more than 1,200 points by the NRCS demonstrated Sanborn data to be accurate to 6.5cm)
Standard Product / FEMA Compliant Product
89% of artifacts removed 90% of artifacts removed
90% of outliers removed 95% of outliers removed
90% of vegetation removed 95% of vegetation removed
95% of buildings removed 98% of buildings removed

Land Cover Mapping

State of Rhode Island

Project Scope
Sanborn was contracted by the State of Rhode Island to use a semi-automated approach to update the existing land use map from 1995. The update involved a digital land use and land cover using scanned, color, 2” pixel-resolution and leaf-off orthophotos from 2003/04. This project involved full redelineation of the entire state. As part of the project, Sanborn also decided to provide an impervious layer that delineated all impervious areas in the State. The minimum mapping unit was ½ acre.

The semi-automated approach was customized for the State of Rhode Island by Sanborn. The Rhode Island staff participated actively in the development of the final approach – on decisions regarding the buffer distances, rules, exceptions, and incorporation of wetlands, etc. A pilot encompassing one square mile was first completed and accepted by the Client. Two large pilots were then completed for two diverse landscapes – Block Island (very rural and coastal) and East Providence (representing a variety of high-intensity urban uses). The purpose of the pilots was to incrementally determine and validate rules, and improve the production processes for various types of geographies and land use mix.

Any changes to the rules were made and full production took place from April 2007 – October 2007. Both internal and external qc were performed and any changes based on local knowledge provided by the Rhode Island staff were incorporated and delivered. The entire area of the state was broken up into production areas for processing and production. Edge matching was performed to deliver a statewide land use/land cover dataset and a statewide impervious dataset.