March 14, 2014 / by John
Our team has been working hard to nail down simple workflows around processing Landsat 8 imagery and we thought it would be fun to highlight two maps that leverage remote sensing techniques. These raster surfaces can be easily generated in house by small watershed organizations and are used regularly in environmental monitoring efforts. Remote sensing is a field that is exploding with the advent of new technologies and environmental applications. Defined simply, it is the science of obtaining information about objects or areas from a distance. Mounted from an aircraft or satellite, remote sensors collect data by detecting the energy that is reflected from Earth and can be analyzed in a variety of open or proprietary software.
Passive sensors record radiation that is reflected from the earth’s surface and store these data into multispectral bands. These sensors need an external energy source like the sun, in order to record data. Active sensors like LIDAR (Light Detection and Ranging) and Radar (RAdio Detection And Ranging) rely on their own, self-powered stimuli to collect data about the earth. We could literally spend a decade explaining remote sensing fundamentals but what’s key is you need to know the multilayered data produced from active and passive collection methods often unlocks a tremendous amount of information about the landscape and how it is changing. Did I mention it’s free?…Well not all of it but a good bit, check out EarthExplorer.
Estimating Surface Temperature
The steps required to build a surface temperature layer are fairly simple. First one must convert the raw thermal Landsat 8 bands into a unit that represents Top of Atmosphere Radiance (TOAr). While the actual physics behind how these values are generated are quite complex, the concept is fairly logical. A large radiator (in this case the sun) fires energy at the earth.
The wavelengths of emitted energy fall somewhere along the electromagnetic spectrum. In this case we are interested in wavelengths on the thermal infrared portion of the spectrum, which bounce off the earth’s surface and get picked up by one of Landsat 8’s two thermal infrared sensors (band 10 and 11). Different types of surfaces have different emissivity values which in turn affect the radiance values picked up by the satellite. Note how areas on the map like roof tops and bare dirt tend to have a higher surface temperature than a forested area. The following is a simple overview of the major steps in the process:
- Apply DOS1 atmospheric correction to correct for Rayleigh Scattering caused by atmospheric haze
- Conversion of raster bands from DN to reflectance and At Surface Temperature
- Land cover classification of study area
- Reclassification of the land cover to emissivity values
- Conversion from At Surface Temperature to Land Surface Temperature
There is excellent documentation on generating these types of surfaces. Here are two leads for ArcGIS users and open source QGIS users.
- Deriving temperature from Landsat 8 thermal bands (TIRS)
- Land Surface Temp. w/Semi Automatic for QGIS
Measuring Cover Crop Productivity
Normalized Difference Vegetation Index or NDVI is a measure that has many applications in agriculture and forestry managment. The map we have created is a December 3, 2013 Landsat 8 scene processed to show a true color image (to the right of the swiper) and the image’s NDVI (to the left of the swiper). The rationale behind NDVI is that live green plants absorb solar radiation for use during photosynthesis. Leaf cells scatter solar radiation in the near-infrared region of the electromagnetic spectrum to prevent tissue damage and overheating. By normalizing the ratio of reflectance and absorption values along the visible light and near-infrared sections of the electromagnetic spectrum, NDVI can be used to identify areas of high and low biomass production as well as pinpoint healthy growing vegetation.
In the case of our map, the greener portions of fields indicate healthy winter cover crops that are actively growing. When fully grown they will mitigate runoff from fields and absorbs nutrients that can flow into the Bay. Notice that we can use this map to identify portions of the field that are not growing as fast as others. Further, the map helps users differentiate between what the naked eye would perceive as bare soil but is actually newly germinating crops. NDVI is simply a ratio of imagery bands and is calculated as follows:
NDVI = (Near Infrared-Red)/(Near Infrared+Red)
If you are an ArcGIS for Desktop User you can use the raster calculator in spatial analyst or the NDVI function in the imagery analysis window. I strongly recommend learning about the raster calculator as it is a powerful tool for generating new surfaces based on simple map algebra. The following two articles should help get you started:
To wrap up, these are just starting points and examples we hope will get our followers excited about the various applications of remote sensing. We have a few projects on deck that will help smaller nonprofits implement similar workflows that support restoration decision making and we will be sure to release more tips and techniques in the future. If you have general questions or comments on remote sensing, ping me on twitter, @JohnDawes4.