Imagery and GIS. Kass Green

      ___________________________

^1. Most remote sensing systems record electromagnetic energy, but some, such as sonar systems, record sound waves.

^2. Hans Karl Heidemann, “Lidar Base Specification,” ver. 1.2, November 2014, US Geological Survey Techniques and Methods, book 11, chap. B4, https://pubs.usgs.gov/tm/11b4/pdf/tm11-B4.pdf.

      Chapter 4

       Choosing and Accessing the Right Imagery

      Chapter 3 introduced the characteristics that differentiate the types of imagery available to civilians. This chapter examines those characteristics through the eyes of the image user, first by presenting a framework for matching imagery characteristics to user requirements so as to ensure the best imagery is chosen for each project, and then by describing and cataloging the wide variety of imagery datasets available at the time of the printing of this book. However, please be aware that the supply of both public and commercial imagery is very dynamic and constantly changing, especially with the advent of unmanned aerial systems (UASs) and small-payload earth observing satellites (often referred to as cubesats, microsats, or smallsats). There is a high probability that new imagery sources have arisen and others have failed since this book has been published.

       Selection Framework—What’s Required versus What’s Available

      The usefulness of a particular imagery product will depend on its technical and organizational characteristics as well as how those characteristics meet the needs of users. Sometimes an analyst has no choice regarding the type of imagery to be used and has to make do with what has been provided by their organization. However, in today’s ever-expanding imagery marketplace, multiple datasets are often available and accessible. Because so much imagery is available, it is also now common for a mapping project to use multiple datasets. For example in the Sonoma Vegetation Mapping project, multiple imagery datasets were used including two years of National Agriculture Imagery Program (NAIP) imagery (4 bands, 1-m spatial resolution), multiple years of Landsat 8 imagery (11 bands, 30-m spatial resolution), Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) hyperspectral imagery, 4-band multispectral, 1-foot imagery collected in 2011 by the county, and quality level 1 lidar and 4-band optical imagery (6-inch spatial resolution) flown specifically for the project.

      No matter how large or small, each image acquisition will require trade-offs between imagery characteristics. It’s not uncommon for imagery analysts to want to use the best spatial, temporal, and spectral resolutions available. However, not every project requires or can afford the highest resolutions possible. Higher resolutions usually translate into higher costs and limited accessibility or both, due to licensing, increased storage, and processing times. Rather, imagery should be chosen to fulfill the project requirements. Products should be substituted for one another and trade-offs made, especially between price, spatial resolution, spectral resolution, temporal resolution, and licensing restrictions.

      To determine what is required by a project instead of what is desired or available, the analyst can ask several simple questions that will narrow the choice of imagery. They are:

      1 1.Will the imagery be used for visualization or to make a map?

      2 2.What is the smallest item to be identified on the ground?

      3 3.What is the time frame of the project and its results?

      4 4.What types of features need to be mapped?

      5 5.What is the size, shape, and accessibility of the project area?

      6 6.What are the requirements for spatial and spectral accuracy?

      7 7.Will the imagery be shared with other organizations?

      8 8.Is the imagery accessible?

      9 9.What is the project budget?

      It is the combined answers to all these questions that will determine which type(s) of imagery will best meet the needs of the project. Often some project requirements must be relaxed, and the questions asked and reasked multiple times before the answers converge on a particular set of imagery products.

       Will the imagery be used for visualization, or to make a map?

      Imagery is most often used as an aid in the visualization of map information. Imagery used in map visualization allows the user to understand the context of a location—to be able to visualize its surroundings. The most readily available imagery is served for visualization to mapping websites such as ArcGIS Online, Bing, Apple, and Google at no charge to the user. Usually, imagery for visualization is optical and served in true color. For example, a common base image, Esri’s World Imagery, is served at no charge in true color and is created from a variety of sources at multiple scales including Landsat, NAIP, WorldView, and Pleiades imagery. Imagery used for visualization is always dated, because it is derived from archive imagery. However, the currency of visualization imagery is rapidly increasing with the increasing supply of worldwide high- and very-high (HVH) spatial-resolution imagery.

      However, the imagery in visualization applications is available only for viewing and not for analysis. It is made available as cached, tiled services, which offer imagery in highly compressed JPEG or PNG format (please see chapter 5 for more detail on image compression). As a result, the band combination that is displayed (usually natural color or color infrared composite) cannot be changed by the user, and the pixels values are not available for analysis. These constraints limit how much information can be derived from the imagery.

      While suitable for heads-up digitizing of objects easily discernible (e.g., streets, buildings), the cached imagery does not provide enough data for deciphering subtle changes such as vegetation types or camouflaged items. The creation of many maps requires that the values of the imagery pixels be available for digital analysis. Most cached imagery is also not current enough to support disaster response activities.

       What is the smallest item to be identified on the ground?

      The imagery spatial resolution requirements of a project are determined by the smallest item to be identified on the ground—the project’s minimum mapping unit (MMU). Smaller MMUs require higher spatial resolutions. For example, mapping forest types with an MMU of 10 acres can easily be accomplished using 30-m Landsat imagery. However, mapping individual trees will require 1-meter or higher spatial resolution imagery. Similarly, mapping four-lane freeways can just barely be accomplished with 30-m imagery, but mapping two-lane residential streets or small unpaved secondary roads will require a higher spatial resolution.

       What is the time frame of the project and its results?

      The time frame of the project and its results will affect the temporal resolution required and can also affect the choice of using an active or a passive sensor. Land-use and land-cover information required immediately for decision-making will often be based upon readily available archived imagery, so that the user does not have to wait for new imagery to be collected and processed. Using archived imagery is viable as long as the date of imagery capture is not so distant as to make the imagery obsolete. Conversely, disaster response requires immediate postevent imagery that shows the extent and impact of the disaster. Similarly, imagery used for weather prediction must be up to date so that weather models can be run from current weather conditions. Obviously, mapping troop and military equipment movements also requires high-temporal-resolution imagery. Mapping perpetually clouded areas