High_resolution_GEC
- 1. MARCH 2012 GLOBAL ECOSYSTEM CENTER www.systemecology.org Remote Sensing & Classified Land Cover Essential Land Use Decision Support Tools Using High-Resolution Imagery
- 2. Remote Sensing & Classified Land Cover: Essential Land Use Decision Support Tools 2 GLOBAL ECOSYSTEM CENTER High-resolution imagery is collected by a“camera-like”device called a data collector. The data collector is a sophisticated technical instrument which captures four or more spectral bands of data for analysis by sensing the electromagnetic waves emitted by the landscape. The data collector can be carried on an aerial or satellite platform. Once the imagery has been collected, it is transferred to a ground station. Once on the ground and ready for distribution, the data can be downloaded over the internet by an image analyst for use. The data contained in the files can be extracted with scientific precision and classified into distinctive land cover categories. High-resolution imagery provides detailed information regarding a landscape including the size and location of objects as large as roads, houses and trees, or as small as cars and back- yard sheds. Once high-resolution imagery has been classified into land cover categories and incorporated into a Geographic Information System, it provides municipal managers and plan- ners the resources to immediately gather and analyze data. This process includes investigating complex growth and development scenarios that require ancillary data such as demographics, transportation, precipitation and soils. The collection of high-resolution imagery can be placed in two categories: • Leaf-on imagery for analyzing green infrastructure • Leaf-off imagery for analyzing gray infrastructure H igh-resolution imagery collected through remote sensing may visually resem- ble a photograph, but it is not. The remotely sensed image is a robust data file, not just a picture pleasing to our eyes. The image reveals intricate detail, often invisible to the human eye, about the objects in the landscape. Technically, a high-reso- lution, remotely sensed (RS) image is a digital, multispectral, geo-referenced file with a pixel size of less than four meters. The image also provides detailed data that documents the landscape and can be used to produce accurate decision-making models. “There is nothing insignificant in the world. It all depends on the point of view.” —Johann Wolfgang Von Goethe
- 3. 3MARCH 2012 Green Infrastructure from High-Resolution Imagery Green infrastructure data is produced by conducting a land cover classification of leaf-on high-resolution imagery. While this process has always been an accurate technique for mapping urban ecosystems, it has also been expensive. Developing green infrastructure data has recently become very affordable for two reasons, the imagery necessary for the analysis can now be obtained at no cost from the Na- tional Agriculture Inventory Program (NAIP), and the Global Ecosystem Center (GEC) has developed a methodology to produce accurate land cover classifications for five to six hundred dollars per square mile. The information derived from a green infrastructure analysis allows communities to dramatically reduce spend- ing on unnecessary maintenance and construction by utilizing the natural benefits provided by vegetation and soils. With the high-resolution land cover product, scenario modeling techniques can be conducted before decisions are made. The scenario models allow planners, managers and community leaders options to evaluate and maximize their resources. Technical reports demonstrating the value of green infrastructure analysis and scenario modeling can be reviewed at www.systemecology.org/pastprojects.html Figure 1 – An example of high-resolution sprectral imagery (left) and the image classified into land cover categories (right). High-Resolution Imagery Classified Land Cover - Green Infrastructure High-Resolution Imagery & Urban Infrastructure High-resolution imagery has been used for the man- agement of city infrastructures for over two decades. This imagery is used to map, analyze and manage urban infrastructure. City management has been revolutionized by Geographic Information Systems (GIS) and the robust data gathering technology emanating from the analysis of high-resolution imagery. Imagery used for measuring gray infrastructure (the things we build) is often less than one meter resolution and the imagery used for green infra- structure is between one and three meters resolution. Management of gray infrastructure is the dominate use for high-resolution imagery in urban areas. Almost all urban areas with full time staff utilize high-resolution imagery and GIS technology to plan for growth and development and to manage the built infrastructure, while few use it to manage its green counterpart. Both green and gray data are needed for decision support and the costs of creating the gray data layer is huge when com- pared to the green. The imagery needed for managing green infrastruc- ture can be obtained at no cost to the community, is only needed on a three year schedule, and a new methodology developed by the Global Ecosystem Center reduces the processing of the data to about $550 per square mile. For details on developing a low cost green infrastructure assessment go to www.systemecology.org/ greeninfrastructure.
- 4. Remote Sensing & Classified Land Cover: Essential Land Use Decision Support Tools 4 GLOBAL ECOSYSTEM CENTER Image Scale, Applications and Interaction There are three general scales of imagery used for remote sensing: high, moderate and low-res- olution. Each resolution is designed for a specific purpose; for example, low-resolution satellite data is appropriate for global or continental-scale issues such as global climate measurements or weather forecasts. Moderate-resolution imagery provides regional land cover data for planning, and high- resolution imagery supplies data required for urban infrastructure mapping and analysis. Generally speaking, high-resolution imagery pro- vides detail while moderate resolution imagery pro- vides perspective. The differences are obvious when you examine the pixel size of each type of imagery. There are 900 one-meter high-resolution pixels in a single 30 meter Landsat pixel. A single landsat pixel is slightly larger than two tennis courts, while a high-res- olution pixel is approximately the size of an office desk. The initial step in selecting imagery for analysis is determining the reasons for the analysis. When the pur- pose is to identify major land cover features (forest scale versus individual tree data), a Landsat analysis is logical. While both Landsat and high-resolution imagery can be used to identify landscape changes, Landsat provides a historic record extending several decades while the high- resolution comparisons are often limited to less than a decade. Figure 2 is an example of two map scales. Moder- ate-resolution Landsat imagery (left) provides a broad perspective (see publication listed on the back page for more explanation) but does not have the resolution needed to provide accurate descriptions of individual objects compared to high-resolution imagery. Figure 2 – A comparison of image resolution. 30-meter Landsat imagery (left) contrasted to 1-meter high-resolution imagery (right).
- 5. 5MARCH 2012 Ecosystem Services from Green Infrastructure In 1992, the staff at the Center developed the first meth- odology for mapping green infrastructure and calculating ecosystem services using remote sensing and GIS technol- ogy. The value of green infrastructure to a municipality for producing basic services, like managing stormwater or improving air and water quality, has been documented by hundreds of green infrastructure analysis since the mid 90s. Even though these analysis document huge financial benefits of green infrastructure to a city, it is rare to find a city seriously investigating the options to balance green and gray infrastructure. The Global Ecosystem Center (GEC) developed a low cost Urban Tree Canopy (UTC) assessment product that utilizes free imagery from the National Agricultural Im- agery Program (NAIP) in tandem with an efficient feature extraction method. This allows the production of a report, and development of GIS technical data, and presentation material for use by local experts. The cost for obtaining green infrastructure analyses is so inexpensive that discretionary spending limits within departments is often adequate to pay for the analysis. In addition, further cost saving can be obtained when local landscape ecologists or urban foresters work directly with the municipal officials to interpret the findings. Figure 3 -“Leaf-off”imagery is used to create greay infrastructure (left). ”Leaf- on”imagery is utilized for green infrastructure (right). Ecosystem Services People benefit from a multitude of resources and pro- cesses that are supplied by natural systems. Collectively, these benefits are known as ecosystem services and include things like air and water quality. The task of identifying an extensive list of ecosystem services is a scientific endeavor well beyond the scope of this publication. However, urban decision makers will likely make very different decisions about the management of green infrastructure if they calculate the value of just three eco- system services; stormwater management, water quality and air quality. Here is a short description of the technical models used by the GEC to calculate these ecosystem services: The stormwater analysis estimates the amount of stormwa- ter that runs off a land area during a 2 yr. 24 hour storm event. It calculates the volume of runoff using the TR-55 hydraulic engineering formula developed for small urban watersheds by the U.S. Natural Resource Conservation Service. The water quality assessment is based on the Long-Term Hydrological Impact Assessment (L-THIA) used by the EPA. This model calculates the change in the pollutant concentrations resulting from the change in land cover during a typical storm event (2 year/24 hour). Air quality is determined using the UFORE Model for air pollution developed by the USDA Forest Service. The urban for- est effects (UFORE) model is based on data collected in 55 cities.
- 6. Remote Sensing & Classified Land Cover: Essential Land Use Decision Support Tools 6 GLOBAL ECOSYSTEM CENTER Figure 4 – An overview of an Urban Tree Cover Analysis Steps in Green Infrastructure Development 1. High-resolution imagery is collected 2. The imagery is processed into land cover classes by an image analyst 3. Existing natural resource data is added (rainfall, soils, etc.) 4. Additional ancillary data is identified from city management files 5. Data is combined in a Geographic Information System 6. Ecosystem services calculations are processed and the following data produced: • Green infrastructure statistics • Storm water calculation • Carbon storage and sequestration • Air and water quality • Scenario modeling
- 7. 7MARCH 2012 Land Cover Land cover metrics are measurements of Earth’s land surface, including vegetation, geology, hydrology, or anthro- pogenic features. Land cover directly impacts biological diver- sity while contributing to local, regional, and global change. Classifying high-resolution imagery into discrete land cover classes gives managers the data needed to model development scenarios and test management strategies. Remotely sensed imagery is an accurate, effective and least costly means of obtaining data describing the landscape. High- resolution imagery is used to develop data for managing urban landscapes. The imagery becomes a powerful data source for decision-making when it is classified into discrete land cover types. Once the land cover data is in a digital form, it can be used by the community’s GIS. The GIS allows the image data to interact with other ancillary data to produce critical decision- making information. This imagery provides snapshots of land cover. The data is collected as a spectral image and is translated into discrete land cover categories by image analysts. The land cover classifica- tions derived from high-resolution imagery provide accurate assessments of both green and gray infrastructure. Converting the high-resolution image from its original spectral form to a digital file requires the skills of an image analyst who is trained in the classification of remote sensed imagery. The technical field of study is called digital image analysis. Highly accurate descriptions of the land can be devel- oped using this approach. Different land covers like trees, water and roads reflect light in different wavelengths or bands. Our eyes can see the visible bands while a digital image can record beyond the visible spectrum to the infrared and other bands. For example: Trees look green to our eyes because they reflect green light that our eyes see. They also reflect infrared light that our eyes cannot see. The more wave lengths that an image can sense, the easier it is to separate the different land cover types. In the case above, the infrared portion of the spec- trum increases our ability to tell trees and crops apart. Common land cover classes for high-resolution green infra- structure are: Trees Grass Impervious surfaces Bare soil Water NAIP Imagery The National Agricultural Image Program (NAIP) was established in 2003 to provide high- resolution land cover imagery needed to support agricultural programs at the federal, state and local level. This program has been extremely effective and evolved into the largest single civilian mapping program in US history. Imagery has been collected over every state many times since 2003 and in most states, imagery is available for five different years. All the imagery collected is in the public domain. Although the original purpose of the program was to serve the agricultural community in rural areas, it has proven to be an ideal source of data for urban areas. NAIP imagery is obtained at the height of the growing season and therefore documents green infrastructure in urban areas as well as crop and pasture land in rural areas. Urban areas have a great need for green infrastructure data, but have been slow to develop green data layers in their Geographic Information System because of the perceived cost. With the availability of this data at no cost, this is no longer a concern. The quality of the NAIP imagery has kept pace with the rapidly evolving technology of Remote Sensing (RS) imagery collection. In recent years, almost all the data has been collected with digital RS gear that produces four digital bands of spectral data (Red, Green, Blue, and Near Infrared). Since 2009, all the data has been collected at one meter resolution or better and the collection interval is generally three years. This four-band multi-spectral data provides image analysts with robust data for analysis. For more information about NAIP imagery go to http://gis.apfo.usda.gov/gisviewer/
- 8. Remote Sensing & Classified Land Cover: Essential Land Use Decision Support Tools 8 GLOBAL ECOSYSTEM CENTER Figure 5 – High-resolution land cover classifications are able to be temporally updated through re-classifying only areas that have changed. The 2005 classification is the“base”(below-left) and only the areas that have changed are re-classified, creating a updated land cover classification (below-right). Case Study: Decatur, Georgia - Low-Cost Urban Tree Canopy Assessment o Low cost analysis o Ideal for small towns and cities o Uses NAIP imagery (no cost) o High-resolution change analysis o Urban Tree Canopy assessment o Ecosystem services Decatur, Georgia 2005 2010 45% 27% 28% Landcover Categories
- 9. 9MARCH 2012 Case Study: Berkeley Heights, New Jersey - Mid Level Analysis Percent land cover: o Tree canopy 51% Impervious understory 6% Impervious 21% Open Space 22% Bare .4% Water .3% o Highest tree cover in county parks o Scenario modelling +1% impervious change = $23.8 million stormwater runoff values Landcover Categories 51% 21% 22% 6%
- 10. Remote Sensing & Classified Land Cover: Essential Land Use Decision Support Tools 10 GLOBAL ECOSYSTEM CENTER MARCH 2012 Case Study: Bellevue, Washington - High-End Analysis o Detailed analysis designed to address specific management needs of the community o Requires detailed interaction with managers and GIS department Figure 6 – Bellevue Washington - When the land cover results are analyzed through zoning pat- terns, different perspectives emerge (need more work here). Bellevue, Washington Central Business District Zoning Suburban Residential Zoning Tree Canopy by City Zoning: Citywide 36% Commercial 21% Urban Residential 30% Industrial 19% Suburban Residential 35% Parks 67% Central Business District 7% Right 0f-Way 20% Landcover Categories 21% 16% 20% 15%10% 14% 51% 8% 32% 27% 17% 15% 16% 9% 14%
- 11. MARCH 2012 11 Conclusion Every municipality in the United States should have a green data layer in their Geographic Information Sys- tem (GIS) so they can manage their green infrastructure. Studies of the land cover in hundreds of municipalities show that maintaining a robust green infrastructure is an extremely good financial investment. The value of invest- ing in green infrastructure becomes obvious to decision- makers as soon as they have the opportunity to evaluate growth, development and management options with both gray and green infrastructure data on-the-table. The most fundamental piece of data needed is a digital map of the community’s foundation i.e. infrastruc- ture. This map is processed from high-resolution imagery. The imagery must be collected at two different times of the year, once when the leaves are off the trees and once when they are on. Finally, both imagery-derived data sets are merged using GIS software and analyzed. Unfortunately almost all municipal management decisions being made today are based on gray infrastruc- ture data alone. It is rare for a municipality to have data on its green infrastructure and this condition is clearly a mistake. A healthy green infrastructure is more than a nicety. While aesthetic benefits of greening a community are well known, the financial value of this green system is not. The work performed by the green infrastructure or natural system can be calculated and used in the decision- making process. The value of the work performed can be calculated by measuring the structure of the landscape using high-resolution imagery in a Geographic Informa- tion System, calculating the function of the landscape by programming mathematical algorithms into the computer software, and finally converting the functional metrics to financial currency. Developing a green infrastructure for use in a com- munity’s GIS has never been easier or less expensive. There are several reasons for this, the first is that the imagery is now available at no cost from the National Agriculture Imagery Program (NAIP), and second, because the Global Ecosystem Center has develop a low-cost classification technique that use high-resolution imagery, a unique analysis technique, and third, the expertise of local urban foresters and landscape ecologists to help communities interpret the data. SS Calculating Stormwater Value 1. Data from the gray and green infrastructure maps are combined in a GIS project to determine the structure of the landscape. 2. Ancillary data documenting rainfall and soil condi- tions are added to the project. 3. The findings from 1&2 above are imported into a hydraulic engineering model developed by the Natural Resource Conservation Service and the stormwater run-off volume is determined. 4. A scenario model is developed to test a proposed change in the community infrastructure. 5. The difference between the existing landscape and the future scenario is determined. The difference in the stormwater volume is the gain or loss in ecosystem service. 6. Once the volume is known, a dollar value can be placed on that volume commensurate with the incremen- tal cost of increasing or reducing the size of a stormwater facility.
- 12. Remote Sensing & Classified Land Cover: Essential Land Use Decision Support Tools GLOBAL ECOSYSTEM CENTER 1607 22nd St. NW, Washington, DC 20008 Phone: 202.290.3530 Fax: 202.683.6729 http://www.systemecology.org Gary Moll, President Kenneth Kay, Geospatial Specialist Binesh Maharjan, Geospatial Specialist FEBRUARY 2012 Remote Sensing & Classified Land Cover Essential Land Use Decision Support Tools Using Moderate-Resolution Imagery GLOBAL ECOSYSTEM CENTER www.systemecology.org Also available: Essential Land Use Decision Support Tools Using Moderate-Resolution Imagery 12