| Irrigated Lands Mapping |
Initially test the ability to reproduce the RASA methodology using 1992 Landsat TM satellite imagery dataset. Develop guidelines for the methodology to process data for the entire study area. Outline methods for using ground-truth data to improve accuracy of irrigated lands estimates.
Batch process imagery using the guidelines and methods developed in the previous step. Imagery will be classified using a band-ratio (band 4/band 3) method to create a brightness theme and enhance the vegetation signature. A brightness threshold will then be selected to separate irrigated and non-irrigated land.
Iteratively improve classification accuracy by using ground-truth data to calibrate the selected threshold.
Randomly select 1 to 10 one-square-mile sections from 139 counties and acquire ground-truth information on crop type and irrigation status. Compare with classified imagery and compute accuracy statistics.
Aggregate the classified image into a 2km by 2km grid of percent irrigated land to compare with the 1980 RASA data set and create a change map that indicated the location of increases and/or decreases in the amount of irrigated land in the High Plains.