Goals and Background
The goal of this lab is to gain the knowledge necessary to
measure and interpret spectral reflectance signatures of various surface, and
near surface features of the Earth using satellite imagery. Spectral signature
curves are used to help analysts to identify, classify, and map land surface
features as well as to determine which bands should be used to differentiate
features with similar spectral signatures. Another section of this lab provides
experience with some basic Earth resource monitoring functions available in
Erdas Imagine. Vegetation density can be monitored using normalized difference vegetation
index (NDVI), which ratios reflectance of the red and near infrared bands from
Landsat TM, ETM+, or OLI. Another operation can be run to analyze distribution
of ferrous mineral in the soil by ratioing the reflectance levels of the middle
and near infrared bands.
Methods
In part one of this lab; spectral signatures are made by
creating an area of interest within the feature to be analyzed and using the
supervised raster function, signature editor. This process is done for twelve
different surface and near surface features found in a Landsat ETM+ image of
Eau Claire and surrounding areas. Each signature is then separately analyzed
for the bands exhibiting the highest and lowest levels of reflectance and the characteristics
of vegetation and soils of different moisture levels analyzed. The signatures are then all plotted together
and analyzed to find bands with the best separability. In the second part of
this lab, two different band ratio functions are performed for the purpose of
resource monitoring in an image of Eau Claire and Chippewa counties. The
processes use unsupervised raster functions to produce an image depicting
vegetation density and another depicting the distribution of ferrous mineral
content in exposed soil. Both of these images are then brought into Arc map and
made into interval classification maps with five classifications in each to
better display the information provided in each.
Results
This lab emphasizes the importance of a basic understanding
of the interactions of the electromagnetic spectrum with various surface
features is an important component in analyzing spectral signatures. It also
provides a small look at the broad range of applications in the field of
spectral signature analysis.
Spectral signatures vary between surfaces, but tend to be similar between sub-classifications of surfaces, such as different soil types. Moisture level of the surface feature effects its spectral signature and can be used to differentiate between similar surfaces, as can be observed when the spectral signature of wet and dry soil are plotted together (figure 1). Plotting all spectral signatures together (figure 2) is helpful for the determination of the best bands for separability of signatures.
Resource monitoring functions in Erdas Imagine produce images helpful in analyzing the resource being monitored, however Arc map has functions that allow the variations to be more clear (figures 3 and 4).
 |
| Figure 1. Spectral signatures of wet and dry soils |
 |
| Figure 2. Spectral signatures of all surfaces analyzed |
 |
| Figure 3. Vegetation abundance map produced using NDVI image |
 |
| Figure 4. Ferrous mineral map |
Sources
Satellite image is from Earth Resources Observation and Science Center, United States Geological
Survey.
Background information is from Analyzing Spectral Signature lecture slides, Cyril Wilson, 2016
No comments:
Post a Comment