GIS 4035 - Final Project - Estuary Comparison - Sediment Deposition

For the final project of Photo Interpretation and Remote Sensing we were tasked with performing manual or automated image processing techniques on remotely sensed data. The landsat imagery for this project was obtained from USGS.com utilizing their GIS database known as Earth Explorer. The imagery I chose was captured by the Sentinel II satellite. Preprocessing techniques for this project involved clipping, scaling and visualizing the imagery to assess for estuary location, bay shape, water color gradient, and tidal influx/outflux.Programs utilized to perform this analysis included Erdas Imagine and ArcMap. Utilizing Erdas Imagine I was able to perform a supervised classification for the study areas of Pond 6 and Saultsman Cove.
 

Utilizing ArcMap I was able to create the following map deliverable, a combination of two maps showing study area, supervised classification, state extent, and spectral Euclidean distance for the two selected estuaries. An overall dark Euclidean distance output being indicative of pixels that have a higher likelihood of being classified correctly. My map, displayed below, contains not only all deliverable map elements including scale bar, north arrow, study area extent, legends, credit, and author, but also class area and total area derived from tables created and imported from Excel after analyzing attribute table data added in Erdas for each respective study area. In conclusion, I found this project to be an accurate assessment for testing an array of GIS skills related to land use land classification. In performing this sediment deposition analysis I have learned that by analyzing pixel coloration one can determine sediment deposition accurately. Sediment deposition rate was assessed for by analyzing the area values for high, medium, and low sediment classifications for each respective study area, and Pond 6 was found to have a higher chance at silling up the quickest.

GIS 4035 - Module 10 - Germantown, MD - Supervised Classification

This week we were tasked with another image classification project. For this project a 57.42 square mile study area located in Germantown, Maryland was subjected to a supervised classification. The following classification consisted of a recording to 8 categories. The 8 categories are as follows: Urban, Grass, Deciduous Forest, Mixed Forest, Fallow Field, Agriculture, Water, and Road. The following features were selected for visually and based on provided coordinates using a drawn polygon method and the region growing properties > at inquire > spectral euclidean value method. The spectral euclidean value and the neighborhood value were adjusted accordingly to obtain features in which there was minimal spectral overlap.
The attached map deliverable highlights the 8 feature categories and their corresponding areas in square miles. The Spectral Euclidean distance map shows a representation of spectral Euclidean distance with brighter areas being represented as values with a larger pixel difference thus having a higher likelihood for wrong classification. The inset spectral Euclidean Distance map is overall dark indicating correct classification.

GIS 4035 - Module 9 - Unsupervised Classification

This week we were tasked with performing an unsupervised classification of our very own UWF campus here in Pensacola, Florida. The unsupervised classification was ran using ERDAS.
UWFclass50.img was created with the following settings : classes 50, RGB 321, maximum iterations 25,
convergence threshold 0.950, skip factors x: 2 y: 2.  Attribute table data was opened and the following category names were applied to the following pixel features in the UWFclass50.img:
Trees – dark green, Grass – green or chartreuse, Buildings/Road – grey,
Shadows – black, Mixed – light green
All 50 classes were recolored and renamed T, G, B, S, or M
Area was manually calculated and determined to be .90 square miles. Impervious percentage
calculated as 25% and Pervious percentage calculated as 75%.

GIS 4035 - Module 8 - Thermal and Multispectral Imagery

For this week's Module 8 we were asked to analyze two composite satellite images at different RGB and landsat band setups in order to highlight specific features. The composite image that I chose was the EMTcomposite.img created in module 8 and the feature I chose to identify was the two fires located in the NE corner of the original EMTcomposite.img. I chose 4 data frames to show 4 different image configurations. These 4 data frames represent, in my opinion, the most effective ways to spot a forest fire via satellite imagery. The configuration of the 4 data frames is as following: RGB 123 - True color image for comparison, RGB 137 – fire appears brighter and more vivid, Landsat Band 1 – Grey Scale – shows brighter areas, Landsat Band 1 – Color Gradient – shows brighter areas as red. Landsat band 1 was chosen as it is the only band that effects the fire. This may be in part due to the fires intense brightness and its physical appearance being higher in elevation when compared to surrounding ground elevation.

GIS 4035 - Module 7 - Multispectral Analysis

This week, for module 7, we were tasked with analyzing satellite imagery for a mountain range in Washington state, tm_00.img. By analyzing histogram data one can extrapolate important information about the .img file.
The following map deliverable focuses on three features located in the tm_00.img.
Feature 1 criteria: In Layer_4 there is a spike between pixel values of 12 and 18. Name the type of feature responsible for this and locate an example of it on the map.
Feature 2 criteria: Identify the feature that represents both A) a small spike in layers 1-4 around pixel value 200, and B) a large spike between pixel values 9 and 11 in Layer_5 and Layer_6.
Feature 3 criteria:  In certain areas of water, layers 1-3 to become much brighter than normal, 
layer 4 becomes somewhat brighter, and layers 5-6 to remain unchanged. Locate an example area that clearly shows these variations in water, using a band combination that makes them stand out.

The following map deliverable was created in arcMap highlighting all 3 features and showing their location in the original tm_00.img with colored extent indicators. The following RGB band combinations are listed as legends for each corresponding feature and an explanation for how I located the feature is found to the right of each corresponding feature.

GIS 4035 - Module 6 - Spatial Enhancement

"The Landsat 7 sensor suffered a malfunction in 2003 called a Scan Line Corrector failure. Since then, Landsat 7 images have frequent lines of no data, giving the images a striped appearance." -Module 6 Background

For this weeks module we were tasked with enhancing a .img file of satellite imagery using Erdas Imagine. Utilizing Erdas one can edit an image in an assortment of ways and this module goes into depth into a variety of filters and tools that can be utilized in order to enhance an area to highlight certain features or in the case of l7_striping.img fix large gaps running diagonally through the image caused by satellite processing malfunction. The main tools used in order to correct for this malfunction were the Fourier Transform, Fourier Transform Editor, Spatial Convolution, and Focal Statistics. A preview of the Fourier Editor settings is included in the attached map and shows the technique I used in order to hide the gaps by applying wedge voids to the "star line" gaps. A Spatial Convolution of 3x3 kernel was applied in order to sharpen features in the imagery. The output of these tools was my deliverable final.img. I imported this .img into Arcmap, added essential map elements, and created the layout you see in the attached map.

GIS 4035 - Module 5b - Intro to Erdas and Digital Data 2

This week we continue module 5 with part b, and continue working with Erdas. For this weeks project we looked at a total of ten .img files and one .shp file in Erdas to analyze properties and characteristics that distinguish areas of the same imagery. The first step in any properties analyzation is to always first review the metadata, and that is precisely what I did. In reviewing metadata, that is unique in Erdas, one can view valuable data about an image that can let one determine the following but not limited to: spatial resolution, radiometric resolution, bit type, pixel size min/max, image dimensions, and projection info. For the final exercise of module 5b we were supplied an .img file soils_95 and a shapefile hydro_00. We were asked to create an area column and a percent column in the soils_95 attribute table in order to select for characteristics. The attached screenshot is a preview of a map in Erdas showing soil layers that contain humus and fine grain particles. This weeks lab concluded the introduction to Erdas and I'm sure that I will find it to be a valuable tool in future GIS projects.

GIS 4035 - Module 5a - Intro to Erdas Imagine and Digital Data

Module 5a Intro to Erdas Imagine and Digital Data had us start off by giving a tutorial into the program Erdas Imagine. This program is designed to process imagery and in this weeks lab I used Erdas to perform another ground truthing / accuracy assessment for a new area.

The region is located in the pacific northwest of the United States and is ~24 square miles in area.
Utilizing Erdas I was able to manipulate the color bands into a RGB 5,4,3 combination to present the following color scheme. I then imported the selected subset .img into ArcGIS pro to assemble the attached map. This ground truthing / accuracy assessment map highlights various features represented in the legend across the diverse landscape of the satellite imagery.

GIS 4035 - Module 4 - Ground Truthing

This week in photo interp. & remote sensing we were tasked with performing an accuracy assessment of the classification area, also known as ground truthing, for module 4.
This assessment builds off of last weeks LU/LC module 2. 30 assessment points were chosen using a random selection method taking care to evenly distribute the points throughout the classification area. Out of the 30 points selected 6 were categorized as error (n) and 24 were categorized as correct (y).
This assessment resulted in an accuracy calculation of 80%.

GIS 4035 - Module 3 - LULC classification

For this weeks project, module 3, in GIS 4035 we were tasked with created a land use / land cover classification map for a provided aerial image file. A LULC level II code format was followed and the following codes were applied to the study area and the reasoning for why:

11- Residential - areas that contained buildings that were smaller than commercial size
12 - Commercial and Services - areas that contained larger than home sized buildings and large parking lots
14 - Transportation - main highway
43 - Mixed forest land - all woodland area was categorized under this classification of mixed
51 -  Streams and Canals - smaller streams leading from the bay were categorized as this
53 - Reservoirs - a few small retention ponds in the residential area
54 - Bays and Estuaries - the bay
62 - non-forested wetland - wetland area surrounding the bay that had no trees

GIS 4035 - Module 2 - Visual Interpretation

For my first project in GIS 4035 Photo Interpretation and Remote Sensing, we were tasked with analyzing and interpreting 3 different aerial images.
For the first .tif aerial image file we were tasked with interpreting tone, by classifying 5 polygon feature class files from very light to very dark, and texture, by classifying 5 polygon feature class files from very fine to very coarse.
The following map deliverable was created.

For the next .tif aerial image file we were tasked with interpreting features based on shape, shadow, pattern, and association.  The following map deliverable highlights each of the mentioned features and indicates the color code for that feature in the legend.

For the final .tif aerial image file we were tasked with interpreting color features when compared to a false color IR image of the same study area. The following differences were noted:

Feature	True Color	False Color
Stream	White	Blue
Forest	Green	Red
River	Green-blue	Blue
Building	Light-grey	White
Lot	Dark-grey	Blue