Thursday, April 20, 2017

Lab 12: Earth Imagery with ArcGis Pro


The objective of lab 11 was to learn how to use Landsat imagery, multispectral data, and vegetation indexes to assess and manage lands. The top image reveals the impacts fire and the nuclear power plant accident has had on Ukraine. Multispectral data was used to determine how vegetation's response to fire. Higher vegetation indexes indicate healthy vegetation while lower ones indicate little/ or no vegetation.

Lab 11: Ground Truthing

The objective of lab 11 was to evaluate the accuracy of the previous lab's land use classification of Pascagoula, MS. I began by creating a shapefile and adding random points to the map. I then created new fields to record whether the points were accurate and had the correct code. As I created points, I recorded whether they matched the previous classification in the attribute table and the actual classification of the point. I used Google Maps to help me classify 33 randomly selected points on the map. I found that 21 matched previous classifications, while 12 were not accurately classified. My maps overall accuracy was 63.64%.

Thursday, April 6, 2017

Lab 10: Land Use/ Land Classification

Land Use Assessment of Pascagoula, MS
The objective of lab 10 was to assess and classify an aerial image of Pascagoula , MS. The image was classified according to features using polygons. Polygons were displayed using colors that best fit the features they symbolized. I had to go back and clip polygons from previous ones that I missed before finalizing the map. Overall, the LULC.shp file that I created made it easier to distinguish features within the image. 

Monday, April 3, 2017

GIS 4035 Lab 9: Supervised Classification

The objective of lab 9 was to create a supervised classification from lab use assessment of Germantown, Maryland. Classifications were made using UTM coordinates and polygon digitization from visual interpretation. Once features were identified,  features could be classified by similar pixel values. I struggled getting the recode process to run, so feature of the same feature are still separated in the legend. However, the areas calculated are for that of land use cover for the entire feature. Although this lab was tedious, it was interesting to evaluate and compare land use of Germantown, Maryland.

Thursday, March 23, 2017

GIS 4035 Lab 8: Unsupervised Classification

The objective of Lab 8 was to perform an unsupervised classification in ArcMap and ERDAS, create values for pixels of select features, and combine pixels of the same feature into one class. Although the lab was tedious, it was interesting to note how tremendously shadows affect the way features are projected in an image. When categorizing the image into categories of trees, grass, building/roads, and shadows, I found that mixed values made it hard to assign values to a particular class. The image below is a classification at my discrepancy.

Tuesday, March 21, 2017

GIS 4035 LAB 7: Spatial Enhancement


The objective of lab 7 was to learn how to use spatial enhancements in ArcMap and ERDAS to alter the way features are displayed within an image and correcting errors. Filters used included low pass filters, high pass filters, and Fourier transformation.  Filters work by changing pixel values. A low pass filter removes high frequency data while a high pass filter suppresses low frequency data. Fourier transform changes spatial frequency to surface underlying values. The image below is that of an edited Landsat 7 sensor image following a Scan Line Correction failure. The lines were removed in ERDAS using Fourier transform in attempt to remove lines in the image. I adjusted the wedge tool shape multiple times to get the output I wanted. After using Fourier Transform and Convolution Filtering, I changed to focal statistics to rectangle height 3 and width 3 which worked best to display the image features without too much distortion. 

Thursday, February 23, 2017

GIS 4035 LAB 6: Thermal and Multi-spectral Analysis

In lab 6, students learned how to create multi-spectral images in ERDAS and ArcMap. A single image was created from multiple individual layers using Composite Bands in ArcMap and Layer Stack in ERDAS. I had to repeat the Import Data process in ERDAS Layer Stack because I did not set the Output File correctly. Multiple band combinations were used to distinguish features in composite images. I selected farmland as my map feature for the deliverable map. I chose to display the composite multi spectral image in NIR because it made it easy to differentiate between farmland and varies other vegetated lands. I also used a color scheme for a second composite image of the selected area. 
As seen in the image above, the color ramp composite image makes it hard to differentiate between features compares to that of NIR.