Lab 3A - Spectral Indices
3.1 - Overview
In this lab, we will work with the spectral characteristics in our data to visualize and extract insights that go beyond basic visual interpretation. We will work with the different spectral bands offered by Landsat 8 to find unique patterns that can help us solve problems and conduct analysis. By the end of this lab, you should be able to understand how to build and visualize existing indices, as well as construct your own, identify how different indices can help your use case, and understand the mechanism behind how they work.
3.2 - Spectral Indices
Spectroscopy is the study of how radiation is absorbed, reflected and emitted by different materials. While this discipline has its origins in chemistry and physics, we can utilize the same techniques to identify different land cover types from satellite data. In the chart below, land cover types have unique spectral characteristics. Snow has a major peak at lower wavelengths and is near zero above 1.5 micrometer, whereas soil has very low reflectance at lower levels of wavelength but relatively strong and steady reflectance after ~0.75 micrometers. Spectral indices are built to leverage these unique characteristics and isolate specific types of land cover.
Land covers are separable at different wavelengths. Vegetation curves (green) have high reflectance in the NIR range, where radiant energy is scattered by cell walls (Bowker, 1985) and low reflectance in the red range, where radiant energy is absorbed by chlorophyll. We can leverage this information to build indices that help us differentiate vegetation from urban areas. In the next few sections, we will cover several of the most important indices in use.
3.2.1 - Normalized Difference Vegetation Index (NDVI)
The Normalized Difference Vegetation Index (NDVI) has a long history in remote sensing, and is one of the most widely used measures. The typical formulation is: