GIS 4930: Special Topics; Project 3: Stats Analyze Week

Welcome to the continuation of our look at statistical analysis with ArcMap.  Recall that the theme being explored with statistics is methamphetamine lab busts around Charleston, West Virginia.  These past few weeks of analysis have been the bulk of the work for this project.  The overall objectives of the analysis portion are to review and understand regression analysis basics, and a couple key techniques.  Define what the dependent and independent variables are for the study as they apply to the regression analysis.  Perform (multiple renditions) of an Ordinary Least Squares (OLS) regression model.  Finally, complete 6 statistical sanity checks based on the OLS model outcomes.  

In the previous post, we looked at a big overview of the area that is being analyzed.  There are 54 lab busts from the 2004-2008 time frame taken from the DEA's National Clandestine Laboratory data.  Decennial census data from 2000 and 2010 at the census tract level was spatially joined to these 54 lab busts.  The data was then normalized into  a percentage by the census tract into 31 categories for analysis in the OLD model.  These 31 categories of data were then fed into the model and systematically removed while analyzing their affect on the model.  Ultimately as good of a model as possible was arrived at with some results shown below. 

This is a generated output depicting the OLS results created in ArcMap. Key things to note from the table is that there are only nine variables being incorporated into the OLS model of the original 31.  How were variables removed you might ask?  There are six checks, or questions, to answer to determine the validity of a variable's use in the OLS model: does an independent variable help or hurt the model; is the relationship to the dependent variable as expected; are there redundant explanatory variables; is the model biased; are there variables missing or unexplained residuals; how well does the model predict the dependent variable?  The first three of these were generally grouped into one solid check for determining if a variable should stay or go.  The remaining checks were applied to the model results as a whole.  As long as a variable had a coefficient that wasn't near zero, a probability lower than 0.4, and a VIF less than 7.5, it could stay.  After looking at this data table, it's time to transition to the visual interpretation, shown below. 

This map depicts the standard residual for the OLS model depicted in the table.  It symbolizes areas using a standard deviation style outlook.  However, rather than wanting a more Gaussian curve style of data showing some of every color, you ideally want values to be in the +/- 0.5 range because that is said to be highly accurate.  Darker browns indicate areas that the model predicted less meth labs actually were.  Whereas, darker blues indicate high value areas where the model expected more meth labs than those that were actually present.  

This week's focus was not to describe the data results, but to accomplish the analysis leading up to it.. 

GIS 4035: Remote Sensing; Module 9: Unsupervised Classification

This week's assignment revolved around unsupervised classification with remotely sensed imagery.  This is a multifaceted lab looking at a number of different processes culminating in the unsupervised classification and manual reclassifying of the resulting raster dataset for a permeability analysis.  The main objective was to understand and perform an unsupervised classification in both ArcMap and ERDAS Imagine.  Imagery for two different areas was provided, ultimately the UWF area, as seen in the map below, was the final subject matter for exploration of these topics.  
Unsupervised classification is a classification method such that a software suite utilizes an algorithm to determine which pixels in the raster image are most like other pixels throughout the image and groups them accordingly.  After the software has grouped the various pixels together, it is up to the user to define what the group classes represent.  For this type of classification, the software is given certain user defined parameters such as number of iterations to run,  confidence or threshold percentage to reach, and sample sizes.  These essentially tell the software how long to run, what the minimum "correctly grouped" pixel percentage is, and how many pixels to look at adjusting at a time.  

A high definition true color image of the UWF campus was used for the analysis shown below.  This entailed performing a clustering algorithm on the true color image to group like pixels together and then export them as a slightly less defined image for storage space and processing speed concerns.  The clustering algorithm created 50 classes, or shades of pixels which approximated the true color image.  The software was told to produce 50 classes with 95% accuracy overall.  Then I manually reclassified each of those 50 original classes into one of 5 labeled classes.  I accomplished this by highlighting the pixel shade and reviewing it against the true color image and assigning it to the classes described.  Four of the five classes are straightforward and represent what they say, with some possible error.  The mixed class, however, represents certain pixel shades applied to different items that represent both permeable or impermeable surfaces.  For example, some dead grass showing could show a tan pixel while a tan rooftop could also be showing the same value. So recoding this pixel to be grass or buildings would be wrong for at least some of that cluster of pixels.  To account for this the mixed class was created, which is why you can see some rooftops as blue, grassy areas as blue or green and some blue sprinkled throughout. 
Overall, this is a fairly course analysis, but it does do a great job of exercising the process and creating likely results. 

GIS 4035: Remote Sensing; Week 8: Thermal and Multispectral Analysis

This week's assignment was designed to focus on being able to compose a series of different raster bands into a composite image utilizing both ERDAS Imagine and ArcMap.  A couple different images were provided by UWF to exercise these skills and to ultimately come up with a user derived analysis of some particular feature.  

The map shown below is a thermal overview of Florida's Emerald Coast. The image, provided by UWF, dates back to February of 2011.  The main map is a True Color image displaying an overview of where the two inset maps are located.  The central feature of the two inset images is a large oblong clearing.  A clearing is one of many available military firing ranges located along the panhandle.  The main objective of this assignment was to try and differentiate the area of interest from it's surroundings using thermal imagery.  The purple image, located in the top right, comes from a unique combination of infrared both short wave and thermal bands to provide brightness to the "hottest" areas.  These are areas that heat up and/or emit the best.  You can see that there is a very similar spectral pattern all along the island to the south.  Santa Rosa Island is made up of white sand beaches and dunes and appears as the only feature that might be spectrally similar to the artillery ranges. The color inset, located on the bottom right, is another infrared look at the area, but rather a grayscale color has been used to help give characteristic spectral pattern to the other images.  

GIS 4930: Special Topics; Project 3: Stats Prepare

Throughout the next few weeks, we will be delving deep into the clandestine, the dangerous and ultimately bad world of drugs.  Specifically we will be examining the role of GIS statistical analysis as it applies to aiding law enforcement with determining ideal locations to find methamphetamine labs.  Meth has been around since the early 1920's and have been illegal since the 60's, which drove the illicit trade underground.  Meth labs have been found in every state, but surprisingly only in about half of the country's counties.  Over the next few weeks, we will be analyzing two different counties of West Virginia, Putnam and Kanawha. These counties are credited with 187 meth lab busts from 2004-2008.  Chances are, we all know someone who has been impacted through drugs, or drug use, or at minimum you can see it all too prevalent on the news. The idea behind this lab is to examine the socioeconomic trend information that can aid in determining where meth labs are most likely and be able to give that information to local law enforcement agencies. The end deliverable for this year will be a scientific paper discussing the issue and analysis being done on the study area shown below. 

As stated earlier, the study area is of the Charleston vicinity in West Virginia and is home to 187 meth lab busts.  This information has already been summarily broken down into a meth lab density by census tract shown in the main map provided below.  This essentially means the total number of busts per census tract was divided by the area of the tract to provide us with density values seen in the legend.  This map also provides a basic overview of the subject counties and provides state context as well.  

GIS 4035: Remote Sensing; Week 7: Multispectral Analysis

This week's focus revolved around multispectral analysis through spectral enhancement. Essentially this means to take existing spectral data and present it in a manner that might bring out certain relationships or patterns not readily present in other presentations.  The main objective for this assignment was to study an image set and identify certain spectral relationships that aren't readily seen looking at a standard true color image.  This is accomplished by manipulating the pixel values to show other relationships through gray scale panchromatic views of single spectral bands or different combinations of multiple bands such as that seen from a standard false color infrared image.  Both ERDAS Imagine and ArcMap were used to explore the image provided.  Several tools within ERDAS were used, such as the Inquire Cursor to look at particular groups of pixels for their relevant brightness information.  Histograms and contrast information were utilized to identify patterns within multispectral and panchromatic views of one or more spectral bands. The image shown in all three maps below was provided by UWF.  This week's assignment required us to identify three different sets of unique spectral characteristics present within the image and to build maps displaying our results.  These results are shown below. 

The first criteria involved locating the feature in spectral band 4 that correlates to a histogram spike in value between 12 and 18.  Band 4 is generally associated with near infrared (NIR) energy and is good for looking at vegetation and soil and crop land and water contrasting.  With this task, I needed to look at the histogram and find the resulting spike, which is shown in the lower right of the image displayed below.  From there, I specifically made this the only “visible” feature in the map.  Both images shown on the left are using band 4, which proves that the water does stand out quite significantly. 

The second criteria involved locating a feature that represents both a spike in the visual and NIR bands with a value around 200, and a large spike in the infrared layers of bands 5 and 6 around pixel values 9 to 11.  The main features of this image are displayed in a false natural color employing bands 5, 4, 3.  This combination of colors does particularly well at letting the areas that are being inquired about be displayed.  I’ve also created insets to display the different extents of the same data in different spectral scenes.  The two separate breakdowns of pixels of value 200 in visual bands and values 9-11 in the infrared bands are compared on the lower right. 

The third, and final, criteria revolves around water features in which viewing bands 1-3 become brighter than usual, but remain relatively constant in bands 5 and 6.  A true, or natural color, image is shown on the upper right.  Looking at this image, you can see a river in the upper right portion that is much darker than the water ways featured in the other images.  The image shown adjacent to the natural color photo is a custom combination of bands 6, 3, 2 to focus the brightening of the inlet/bay feature while not pronouncing the IR energy in the same ways as the typical false color IR, which is shown in the lower right corner.  Extent indicators were used to show the different looks at the specified band and pixel value combinations identified.  A gray scale image was created to reflect band 3 which shows the largest brightening values. 

This assignment wasn't easy, since it involves many concepts I'm still trying to understand. However, I feel as though I've learned some new things throughout this week's assignment.  I'm looking forward to learning more throughout the remainder of the semester. 

GIS 4930: Special Topics - Module 2: MTR Report

Honestly, I absolutely hated this MTR assignment.  Throughout these past couple of weeks, we were supposed to be working as a group to complete the final project.  Even though I signed up for a group, I felt like I wasn't a part of one.  I would ask the group leader questions but would not receive a response in a timely manner.  Basically, I felt like I was "out of the loop" throughout the entire assignment.  If I had any extra time to work on this assignment, I would most definitely start from the beginning.  I'm really disappointed with the decisions I made in regards to Non-MTR and MTR during the Analyze week a few weeks back.  

GIS 4035: Module 6 - Spatial Enchancement

This week's lab assignment was centered on working within ERDAS Imagine and utilizing the Fourier Analysis tools, the Convolution tool, and the Focal Statistics tool (found in ArcMap).  The project required us to work with a Landsat image that has bands running horizontally across.  The tools named above were required in order to minimize the banding to the greatest extent possible without losing too much detail in the image. The assignment was built for us to experiment with the different tools in ERDAS, as well as learn how to use them in getting better results.  

In performing this exercise, I used all types of combinations using the tools previously listed in an attempt to remove the strips in the image.  The image shown below is the result of using the Fourier Transformation Editor tool and the Convolution of a 3x3 Low Pass kernel. I made sure to finish by adjusting the histogram of the image for better visualization.  My results aren't 100% perfect, since the bands are still visible from the left side of the imagery. However, the right side looks pretty good.  If I had more time, I would definitely experiment with the different tools in an attempt to get the final result to balance out.  Overall, I enjoyed this assignment.  It was definitely challenging, but I feel as if I've learned some new things along the way.