Supervised classification of a Landsat image
This is a supervised classification of a patch of western Victoria, southwest of Ballarat. The original satellite image was recorded in November 2004.
The idea is to generate statistical models of the characteristic spectra of different vegetation classes, by demarcating 'training regions' on the original Landsat image, and then getting the software to calculate the spectral band means and covariances for those patches. Then, using the statistical descriptions of each vegetation type, the software looks at every pixel in the full scene and decides which of the classes the pixel is most likely to belong to (the maximum likelihood algorithm). This is a small portion of the resulting classified scene.
Pine plantations are colour coded dark blue, dry grazing is pinkish buff or brick orange, eucalypt forest is dark green, wheat is bright blue-green, pasture is bright green, ripe crops and mown hay are yellow, other crops are sage green. The classification was performed in ENVI. Assayed accuracy of the classification was 86%.
By doing this for images taken spaced several years apart, you can count the pixels indicating each land use class in each year, to quantify how much and in what direction agricultural land use has been changing over the years.
Supervised classification of a Landsat image
This is a supervised classification of a patch of western Victoria, southwest of Ballarat. The original satellite image was recorded in November 2004.
The idea is to generate statistical models of the characteristic spectra of different vegetation classes, by demarcating 'training regions' on the original Landsat image, and then getting the software to calculate the spectral band means and covariances for those patches. Then, using the statistical descriptions of each vegetation type, the software looks at every pixel in the full scene and decides which of the classes the pixel is most likely to belong to (the maximum likelihood algorithm). This is a small portion of the resulting classified scene.
Pine plantations are colour coded dark blue, dry grazing is pinkish buff or brick orange, eucalypt forest is dark green, wheat is bright blue-green, pasture is bright green, ripe crops and mown hay are yellow, other crops are sage green. The classification was performed in ENVI. Assayed accuracy of the classification was 86%.
By doing this for images taken spaced several years apart, you can count the pixels indicating each land use class in each year, to quantify how much and in what direction agricultural land use has been changing over the years.