Supervised Classification

The links posted below are for a map created using ERDAS IMAGINE 2010 software and the ground control points as part of the University of West Florida On-line Geographic Information Systems (GIS) Certification program class, Photo Interpretation and Remote Sensing (GIS4035/L). The links require Internet Explorer to open and display properly

Though the lab focused on a supervised classification, this module covers the different methods of classifying multispectral images into different patterns that represent clusters of statistically different sets of multiband data, which can be correlated into separable classes. The two main methods of classifying images are Unsupervised and Supervised Classification.

I found this lab more time consuming that expected.  I lost count of how many iterations were done to get to this point.  One thing that did work well for me was to create a flow chart for each of my steps and to take careful notes.  This ensured that precisely the same steps were taken for each class which intern created a MUCH more consistent set of results. Select .xps link below or .pdf map for a larger view. germantown_18d_classified.xps

GeoRectification

The links posted below are for a map created using ERDAS IMAGINE 2010 software and the ground control points as part of the University of West Florida On-line Geographic Information Systems (GIS) Certification program class, Photo Interpretation and Remote Sensing (GIS4035/L). The links require Internet Explorer to open and display properly.

This week's lab objective is learning to use control points from one image or map to rectify another image.  In this case we were provided a USGS map of Pensacola as our ground control point source.

Pensacola_Rectified image (bands 3,2,1)

In the "real" world, one always wants to check and recheck work before it is submitted.  This is a little bit of a challenge since the image is a much coarser resolution than what seems to be available on the map.  In this case the ground "truth" comes from the map. So once I was satisfied with the selection of the WHICH points were to be used I let the imagery point position float until I was satisfied with the visual position and RMS error. Click here for a view the ground control point results. 

Fast forward to the completed image you see below (left side).  The RMS was excellent, (<1), the coordinates seemed right so one more step was taken.  The rectified image and the map image were placed in separate windows, linked and inquire cursors activated.  Voila a perfect match.  I could also have dropped them in the same window and looked to see how they overlaid each other but that will have to be for another day.

Look at the cross hairs on both images.  They are centered on the runway intersections

Module 2 - LandSat Band Combinations

The links posted below are for maps created using ERDAS IMAGINE 2010 software as part of the University of West Florida On-line Geographic Information Systems (GIS) Certification program class, Photo Interpretation and Remote Sensing (GIS4035/L). The links require Internet Explorer to open and display properly.

This module covers satellite based sensors that operate in the visible in infrared wavelengths. 

"Features can be identified visually, by looking at the image, but also by examining the histograms under the layer information.  Certain features have definite and relatively constant reflective properties and therefore will always fall at a certain spot on the histogram.

In this exercise we examined the reflectance values found in the layer histograms to identify real features on the map." from lecture material

 Map of showing shadows on slopes of the Olympic Mountains using false Color Infrared (CIR).  I used layers 4, 3 and 2.  There was a spike between pixel values 12 and 18. 

 Map of Lake Tapps with dissolved calcium suspensions which appear as bright blue CIR and Black in layers 5 and 6. There was a spike around the pixel values of 9-11.

 Map of Grays Harbor shallow estuary showing basin bottom using the bathymetric combination with layers Red=3, Green = 2 & and Blue = 1.

Module 1 Challenge - I love Challenges

I am not sure why these are called "Challenges" rather than Labs but I like the idea. 

So this is the drill.
Import an image into Erdas Imagine 2010.
Then convert it to a map using some of the built in Erdas Imagine capabilitites.
Finally save it as a MS .xps file and publish it to a blog.
So click the link below and see what we can do!

Module_1_Challenge_AVM.xps

Final Project - Bobwhite Manatee Transmission line

Wow this was such a challenging project.  This project encompassed nearly everything we learned this semester.   This required research.  The primary tool was ArcGIS.  Support tools included Erdas Imagine MS Excel, Powerpoint and Word.

A PowerPoint presentation accompanied by a written slide‐by‐slide summary is required as the final deliverables

Carto Final Project. Mean SAT Scores and Participation Rates

Bottom line, I chose a B&W format since this was for a newspaper and it was not stipulated whether the paper would be color or B&W.

I will expand on this on Saturday.

Week 11 - Proposed Wind Turbine Locations

Ohio currently has some wind power producing capabilities. We are to study one existing wind farm and then use established criteria to determine if the Great Lakes should be the site of the next wind farm for the Midwest.

It turns out there are many sources of legitimate information about shipping lanes, average annual wind speeds, bird migration, parks, and more. For me, the issue was sifting through the volumes of data to get what I needed for the lab.
Some of the sources I used are: http://www.greenenergyohio.org/ , http://www.odh.ohio.gov/, www.bgohio.org/utility-director/trubines.html, and http://www.berr.gov.uk/energy/sources/renewables/planning/onshore-wind/ornithology/page18703.html.


I found the reading rather interesting.  There was a considerable amount of literature the could be summed up as "not in my back yard".  However, if one read all the available literature one would find most of the complaints or issues could be easily remedied.  In some cases, technology and creative thinking fixed things.  An example was noise from the turbine blades was lowered by doing three things.  Turning the turbine 180 desgree, eliminated much of the wind noise, as did restyling the blades. Finally, insulating the inside further elminated the noise so that it was no longer an issue.

Bowling Green site
Using Google Earth one can view two wind machines located near a landfill 6 miles southwest of Bowling Green, OH. Bowling Green is home to Ohio’s first utility-sized wind farm. There are four turbines that are 257 feet tall. These turbines are as tall as a 30-story building and generate up to 7.2 megawatts of power — enough to supply electricity for some 3,000 residents. Located about six miles from the city, the turbines can be seen for miles and have become a local attraction.


Lake Erie
I selected two sites on Lake Erie just north of Cleveland OH. I based the selection on the site assessment index from the BERR planning criteria. http://www.berr.gov.uk/energy/sources/renewables/planning/

- Average annual wind speed of between 16.8 and 17.9 mph
- There are no residences near the sites
- There are shipping ports but not in close proximaty but there are shipping routes near the site
- Positioned approximately 15 miles  from shore the coastal horizon is not impacted
- Natural habitats will not be affected

- Installation and constructions costs were a major reason for this selection
Some turbines have been shipped via the from Pensacola to Cincinatti. The proposed sites are near a shipping channel for that reason.  Additionally, notice the proposed sites are not in the highest wind areas since there was no financial gain for the higher speeds but there are higher construction costs.  Finally, the closer selection also means there will less energy loss via normal power transmissions.