SkyTruth Final Report

SkyTruth is a non-profit which provides expert analysis of various aerial images to reveal the "truth" that pictures speak louder than any words one's actions could say, including the action or non-action of any governmental bodies.  Their expert evaluation of the gulf oil spill revealed more than was being said to the media and resulted in a quicker fuller response to the disaster. We could all strive for such a target.   

Our class assisted SkyTruth by evaluating several LandSat images for the presence of Mountain Top Removal in the Appalachian Mountains.  Each group, about 5 members, was responsible for analyzing 2-4 images with various criteria.   The final collaboration is a poster presented from a pro or anti mining position.

Our Group 5  was assigned "pro" mining.   The group itself covered 10 time zones which challenged our level of cooperation and amount of sleep to be had between us.  Here's the team.

Karen Mathews, Grad, Group Leader
Robert Wilkinson, Grad
Alex Fowlkes
Michael Supranowicz
Mohamed M. Hussain

Each of us accepted a role in the presentation:

Alex Fowlkes stepped out as the "poster master". His quick wit to begin the layout design put us on a good path. He collected the elements as they were delivered and expertly placed them into their location, creating interest.

Robert Wilkinson, early on sent out quick research notes which helped to open our hunting skills for the pro-mining stance we were asked to support. He presented the Economy & Energy section.

Michael Supranowicz took on the Health sector which proved to be a challenge.

Mohamed M. Hussain, provided us with a wonderful map of the Group 5 area showing our final MTR polygons.

Karen Mathews, served as the group leader.  This required constant monitoring of our communication boards (school discussion board and email). many hours were spent relaying information between the group members and to see that the written sections stayed on point with the pro stance outcome. An important part was to set interior deadlines so that the total goal could be reached without straining the members.   It was my task to write the Land reclamation section, the background and the conclusion.   As each member completed their task, I reviewed it and submitted it to the group for final agreement. On occasion it was necessary to send a piece back to the drawing board for revision.  Also, an important task was to review the checklist to insure all of the elements of the assignment were completed.

In the end, all the members pulled together to create a good product.  We could all ask for additional tweekings to deliver an "excellent" result rather than a sufficient one. However, considering the circumstances which arose, a good product is acceptable and a great one would have been better.

It was a pleasure to serve as the lead in the project.

Here's a link to the final poster.

Group 5 - Sky Truth Poster

Skytruth Prep

We are now working as a team to analyze several sets of Landsat images for a 2010 evaluation of Mountain Top Removal Mining.   Although this portion not been a mountain top experience, it once again has been an experience for the record books.   Analyzing data that will be used by others on the web for future use is a first.  Understanding more about aerial photographs and how the layers of a Landsat image is clearer than before.  These pictures truly have more than a thousand words to share.

My part as a member of  Group 5 was to evaluate a portion of Landsat # LT50190352010243EDC00. The image was classed using the unsupervised raster program in ERDAS, then hand classified in ArcMap.  The pink portions are potential MTR areas.  The image contains numerous mixed urban areas and the clutter, noise from the classification has not been completed.  A state underlay map would have been helpful in hindsite.

Map updated 12/2/12

Getting Data to Work- A major Petpev of GIS

In the world of our imagination, all things will work smoothly in our designated time allotment. In the real world the "Laws of Murphy" prevail:  where anything that can does and will likely always either go completely wrong or  neither way anything was intended or desired to go.  Otherwise said, expect the unexpected.

As I have been taking this online GIS Graduate Certificate course with the University of West Florida, I have learned that things go wrong.  But as I have begun working as an intern with a local Regional Commission, serving the good of the local government communities, most of our data received from our clients has a bug in it.   Now granted all bugs are not big nor all bugs bad, but when humans are involved there will be something wrong nearly all of the time with client originated data. (Most of our clients are local government offices).

Much of our data is received  in csv - excel data sheets.  Some is sent as a hand drawing or old paper maps which are converted to digital data (even worse).

Our office has recently opened its computer servers as cloud servers for our clients. We publish their data on Arcgis online.  One of our clients sent their layers for us to place on a working map for their office to improve their public works efficiency.   At the meeting when our representative was demonstrating the services of arcgis online, there were many excited comments about the vast abilities of seeing the "live" data layers.  Before the meeting was over one of the workers had altered 4 hydrants and changed the size of 2 major water lines without realizing it.  He was following instructions to become familiar with the new software.  Thankfully, the inadvertent (accidental) changes were found before the meeting ended.  His actions opened immediate procedures for editing procedures.      In this situation, not only was the original data bad being more than 7 years old; there was someone actively editing it without any thought to the consequences.     So not only was the data old, it got worse.

Lastly, I would say that a break in the communication process is the second petpev out there.  This happens more often when a gis task is completed and sent to the client for the client to come back with "oh, that's not what I wanted".  When the client really means, "oh the results are not what I expected."    Which when interpreted by GIS is more likely that the client's desired intentions came out different than what the client thought they would.    Numbers and data visualization have no emotions, but its the spatial results that influence the decisions.

It's amazing to watch truth shine in maps and numbers. 

Lidar-Report Sharing Links & Data -

We have been using the hydrology tool in ArcMap's spatial analysis toolbox to explore it's function.  Our data comes from the Appalachian Mountain, SRTM study zone covering 5 states: Tennessee, Kentucky, North Carolina, Virginia and West Virginia.   It's derived from the overall study found at htt;://eros.usgs.gov/#/Find_Data/Products_and_Data_Available/SRTM  .      Our Group 5 used 4 DEMS from LandSat to convert our data into usable form.  Each DEM was 1201 x 1201 pixels, 16 bit signed, already orthorectified to GCS WGS 1984 UTM 17N, based on one band.
They are:  n36_w083_3arc_v1.tif
                 n36_w084_3arc_v1.tif
                 n37_w083_3arc_v1.tif
                 n37_w084_3arc_v1.tif

The final products were 1- a stream feature shapefile showing the streams of the mosaiced DEM and 2- the basin zones as a raster.  Both of these files along with a .mxd of the layers can be accessed from a combined zipped layer package compatible with Arcmap 10.1 at this address which map be copied & pasted:  http://students.uwf.edu/kfm4/Group5StreamBasin.zip
One may need to create a geodatabase to import the layers for extraction. 


Taking the clipped mosaic through the hydrology study process resulted in these individual layers being created as shown in the map below:

Data Process:

1- The originating DEMS have the characteristics as described above.

2- A mosaic was created with the 4 DEMS, which was clipped to the study area.

3- Using the spatial analyst tool in arcmap 10.1 the fill tool was the first step.

4- The resulting fill layer was run through the flow direction tool without any errors.

5- A stream raster (flow accumulation) was created from the flow direction result.

6- The total raster count was taken from the attribute table- sum value. One (1%) percent was calculated as 46,140.20.

7- The con tool was run using the stream raster as input; with an expression of value>46140, with a constant value of 1.

8- Although the stream to feature layer continually gave an error, a check in arcatalog reveiled it ran correctly but needed the coordinate system manually established. WGS 1984 UTM 17N was set.

This steam feature is the final output in the layer file provided.

9- A final hydrology analysis was created as the basin DEM which is the second layer of the map.

The layer zipped file will appear similar to this map without the study area zone and street basemap from arcgis online.

Lab 10- Supervised Classification

Supervised only means that I'm telling the computer what to do.  Easily the garbage in garbage out is always the maxium used, however, in this process of supervised classification, we can "train" the computer by giving it hints on how to look at the image.   That's it in a nutshell, there are many parts to this process.   We used Germantown, Maryland as our target to perform a Land Use Cover analysis.  It takes more practice than what we are afforded to accomplish it at a higher confidence percentage.

Unsupervised Classification- Lab 9 Remote Sensing

Using the ERDAS program has become much easier since we just perform smaller tasks at a time. This particular approach has brought a greater understanding of the remote sensing part of the GIS learning process.     Unsupervised just means that the computer decides which shades match its sets of code.  This is a good first step. Our Image is the University of West Florida, Pensacola campus. But as my map shows, when the green grass shows up on top of the building there is either a "green" project or a problem lerking.  Here is was the problem.   A second attempt was made at the recoding, however, ERDAS was having a bad day and chose not to save either the codes or the changes as they were entered.  Rather than griping at the program, I chose to go with the first less accurate recoding effort.  The map has been recolored showing more orange for mixed zones.   The second pic is the first color coding which shows the school's feature's fairly well.

This one is the original computer colors, a pretty good contrast of pine forest vs. clearer campus area.
Had to remove this pic. It was causing posting errors. So sad, it looked pretty good.

LiDAR - The Beginning -

Hey did you know that dot to dots, Monet and LiDAR sensing all have something in common? I do now.  LiDAR sensing is a interesting concept. An active sensor - sends "pings" across a surface, the ping returns, creates a picture.  What makes it special are the number of returns surfaces will give, some one, some three.  Then there is the art of learning to interpret the dots.  Certain types of filters are put into place.  Then whoosh, a picture (of types) emerges. It could be a profile when the image is sliced; it could turn into a 3d movie when one flies around.  I wonder if the artist Monet had a grasp of this before it was invented. Some of our filtered results look like he could have painted them.

Here's a slice of two las files from Charleston, West Virginia. One from the hills, the other at the river.   The river picture is filtered by ping codes which gives it a flare.  The hill image profile shows that trees can be evaluated as well.






Our study of elevation and intensity in las files (LiDAR) images shows some differences between the filters.