Mini-Term Project

Background

Flow Model

1. MosaicPro  1990 and 2015

2. Subsetting

3. Binary Detection

4. Signature anaylsis (population, vegetation, etc.)

Methods

December 1990 Charolette and Lee Counties Landsat 4-5 TM

February 2015 Charolette and Lee Counties Landsat 8 OLI

Import Data and Layer Satcking

-Landsat 4-5 data import bands 1, 2, 3, 4, 5, and 7 (band 6 is a thermal band and not needed)

-Landsat 8 data import bands 2-7

band info: http://landsat.usgs.gov/band_designations_landsat_satellites.php

MosaicPro

-Imported the layer stacked images into Erdas Imagine

• first, highlight the image and select Raster Options: make sure "Fit to Frame" and "Background Transperant" are selected. 
• also, in the multiple tab select the option for Multiple Images in Virtual Mosaic
• the images should be imported into the same viewer

-Import images into MosaicPro

• highlight the image and for the Image Area Options select "Compute Active Area" and then "Set" at the bottom. Accept the default values and click Ok to import the image

Color Corrections

• Use Histogram Matching 
• SET
• Matching Method should be set to overlap areas (in the histogram matching window)  

Set Overlap Function

• Select Overlay

Process
-Run Mosaic
-Save mosaic



Subsetting

Creating the Shapefile
-used mgisdata to import counties from the usa geodatabase
-make a layer from the selected  charolette and lee counties in florida
-exported the data to a shapefile in a personal geodatabase used for this project

-import the image
-import the shapefile into Erdas Imagine
-Hold shift- select both counties
-Paste from Selected

-File--> Save As--> AOI Save As --> name file


Raster--> Subset and Chip --> Create Subset Image
-Input image
-name output image
-Click AOI button and select AOI
-navigate to the AOI previously created
-click OK to run the subset


Binary Detection

Mosaic 1990 images with only band 4 (the near infared band) and then mosaiced the 2015 images with band 5 (bear infrared band)

subset the images to the charolette and lee counties shapefile...

sync views
Raster--> functions--> two image functions --> subtaction 15 -1990

Metadata
mean: 7440.311
std. dev. 6574.905

1.5 std. dev = 9862.3575


upper: 17302.6685
lower: -2422.0465

Since the values for 3x std. dev. were way outside of the range for the image, I looked at the histogram and decided on 17680.4 to be the threshold for change in brightness of the pixels. As shown in the figure...

Signature analysis

Discussion

Sources