St. George suffered from water shortages from 2011 to 2015. The desert community also endured record-breaking freezing temperatures during the 2013 winter. These extremes will affect the health of vegetation in the community. Two NAIP scenes were collected and then classified using the same number and types of classes, with the goal to have a degree of uniformity for the comparison. Because the St. George scenes are very heterogeneous, ten classes were chosen: Shadow, Grass, Trees, Water, Soil, Red Soil, Asphalt, Pavement, Exposed Rock, and Rooftop. At least 400 pixels were chosen as training data in each class. A Maximum likelihood classification algorithm was ran for both scenes and the results were validated. At first, the two scenes didn't line up very well, pixel for pixel. A 2nd degree polynomial warp was done on the 2011 scene using the Cubic Convolution algorithm. The warp allowed for the two scenes to line up better. The classification difference between the two classified scenes were then calculated. After the classification difference was calculated, the results were evaluated. Some portions of the scene showed a large decrease in vegetation (Trees and Grass), such as the agriculture field captures below.
Summary
Vegetation Change Detection in St. George, UT
using NAIP 2011 & 2014
A capture of the same field in 2014, NAIP 2014 dataset
​
A capture of a field in St. George from the 2011 NAIP dataset
The difference of classifications from 2011 to 2014. The yellow pixels indicate cells that changed from the 'grass' class to 'soil'. The change from grass to soil may indicate water-stressed vegetation as the two scenes were both collected in the Summer of their respective years. This is just one example of the study area where a vegetation class was changed to a non-vegetation class.
MSGIS Skills: GIS Analysis, Spatial Analysis