What is Infrared Imagery? (Part 2) Relevance of Infrared Aerial Photography

What is Infrared Imagery? (Part 2) Relevance of Infrared Aerial Photography
November 16, 2009 Mike Tully

In Part 1 of this article, we reviewed the elementary physics of infrared light. Because it is generally captured with new digital aerial cameras coincident with normal color photography, it is quickly becoming an inexpensive and ubiquitous imagery resource for geospatial professionals. For this reason it is incumbent upon the profession to understand how this new information resource can best be exploited. Traditionally, color infrared aerial photography has been used for land use classification (plant species identification) and impervious surface delineation. Because a growing number of researchers and professionals are beginning to study and use this more common and higher resolution imagery resource, many new and important uses for this will be defined in the next few years. Many of these uses will align well with geospatial professionals and engineers that need to extract more information from resources spent on aerial acquisition projects. In times of shrinking budgets, these developments will make the return on investment more favorable.

IR.jpgThe USDA National Agriculture Imagery Program (NAIP) has been collecting 4-band (Red + Blue + Green + IR) photography in some areas for several years. Because this imagery is acquired after trees and crops have grown leaves, it has considerable value to land use professionals for species identification, forest inventory, forest pest detection, and land use mapping. This illustrates an important technical aspect of CIR imagery: infrared energy is strongly reflected by plant chlorophyll.

The more “green-ness” the brighter “red” plants will appear when using CIR imagery (Figure 1: Reflectance curve showing chlorophyll absorption of visible and near infrared light). Different plant species can be identified by their unique spectral characteristics. But, more than that, plants under stress from drought or pest pressure also show different reflectance characteristics in the infrared spectrum. Some of the most important uses of CIR imagery are based on these factors. Because of the historically high costs of CIR imagery much of the research and application had been directed out of necessity on high value crops or uses. But now that CIR imagery costs using digital aerial cameras are significantly less expensive, more uses in this area of plant species identification and pest pressure may be described. As this occurs the potential value of this imagery to the professional will increase.

2_IRvRGB.jpgFundamentally, with each additional band of imagery more information is available. Then, more sophisticated analysis using software like Erdas Imagine or Definiens ECognition is possible. So, for example, using only panchromatic (black & white) imagery (a single band) many features are discernible, but when additional bands of color are added much more information can be seen: green versus brown grass, cloudy versus clear water, different species of deciduous trees and plants, etc. When a fourth infrared band (or additional hyperspectral bands) are added even more information is available. Although it is not possible to identify every plant species using only the infrared band, many individual species are possible to identify (Figure 2: Leaf-off photography showing CIR on the left and full color on the right [North Dakota, August 2009, 1 meter GSD]). Sometimes the value of infrared imagery when used for species identification and land use classification is not so much that a species can be identified but that it is easier to automatically or semi-automatically delineate than if using only color imagery.

3_IRforest.jpgThere are a large number of additional uses for CIR imagery. For example, large epidemics of the mountain pine beetle, Dendroctonus ponderosae, in lodgepole pine forests of Colorado, Wyoming, and British Columbia are more extensive now than in the past century of forest management. Mapping the extent of these infestations over time and identifying how and to where the infestations are spreading is an important component of wildland fire management (Figure 3: A dead stand of lodgepole ping showing the “red attack stage” of the mountain pine beetle in CIR photography). Once a tree is overwhelmed by the beetle and dies, the foliage remains green for a few months (called “the green attack stage”). Approximately 12 months after death, the needles turn red (called “the red attack stage”). This is followed after maybe 3 years by “the gray attack stage” after most of the needles have dropped from the crowns. Color infrared photography is helpful to more easily identify the red attack stage. Because there is less chlorophyll in the dead needles the crowns will appear gray or much less “red” than adjacent healthy trees. It seems to be easier to identify and delineate these stands from the infrared band than using full color photography only.

4_IRvRGB.jpgThe Iowa Department of Natural Resources (DNR) has acquired, from Aerial Services, 4-band imagery over much of the state (Figure 4: Rural Iowa showing full color imagery [bottom] and CIR [top]; green vegetation appears bright red when viewed in infrared; water appears very dark indicating low reflectance in infrared wavelengths). They are using this for high resolution land cover classification and prefer CIR imagery over full color because it is a richer source of spectral content and makes classification work more efficient and complete.

Because water is known to absorb most infrared energy, it appears quite dark in CIR imagery. Exploiting this physical property of the light energy, NOAA uses IR photography to delineate the water-land interface. Other organizations in the gulf coast states use infrared imagery to map invasive plant species in tidal zones or river channels (Figure 5: Water hyacinth along the upper river bank of the Rio Grande river is easy to distinguish from other water species; it appears as bright red along the upper river bank in this photo; hydrilla along the lower bank appears as reddish brown).

Impervious surfaces also turn out to be good absorbers (poor reflectors) of infrared energy. Image classification software like ECognition and Erdas Imagine is commonly used to semi-automatically extract all these features from CIR imagery. Considerable cost savings can be realized using these methods over more traditional manual means.

5_IRcoast.jpgThese are a few examples of how color infrared aerial photography has been used to date in the geospatial profession. Color infrared imagery coupled with sophisticated classification software can help increase the return on investment on the aerial photography acquisition project when this extra band of information is acquired simultaneously with the other three color bands. This is now possible using new digital aerial cameras like the Leica ADS80 camera system. This information from high resolution aerial photography is increasingly helpful in answering questions and augmenting more sophisticated planning than has been commonly possible prior to the advent of these new digital aerial cameras. If you are planning an acquisition soon, it may be frugal to include the small marginal cost of infrared imagery. Even if you have not identified a use for it today, there is a high probability that users will have a need for it in the next 12-24 months as professionals learn to exploit this valuable source of rich information.