Wednesday, April 26, 2017

Earth Imagery with ArcGis Pro

Earth Imagery with ArcGis Pro

Exercise 1: Change Detection Over Time at Chernobyl with ArcGIS Online
 I used historical archive of Landsat imagery to explore how Ukraine change 40+ years after a nuclear meltdown. I looked at the accident before and after to compare. 

Exercise 2: Forest Fires in Australia with ArcGIS Pro
 This exercise used Landsat data.  In this exercise, I identified burn scars in Australia after a series of wildfires that happened in 2013. I used a vegetation index to see how the vegetation has regenerated since the fire. 

Ground Truthing

Ground Truthing 

In lab 3 I made a LULC map. This week my goal was to verify the classification schemes. There are three different types of ground truthing accuracy:  Error matrix, producer accuracy, and overall accuracy. My LULC map was based in Liousiana since I am not in the state and can't physically go to the sites I can use other sources like Google Maps. Google- Maps has a street view feature to easily ground truth sites. 

Steps I had to take: 

I created the point shapefile and named it truthing. In catalog, I set the coordinate system to match the LULC which was NAD_1983_HARN_StatePlane_Mississippi_East_FIPS_2301_Feet. Then I went to the attribute table and added two text field named true_YN and New_code. I went to editor and started an editing session. A pop up will occur asking which to layer to edit and I selected truthing. I clicked truthing in the create features box, then I picked 30 random points on the map. Once done, clicked stop editing.  I went to google maps. For the first point, in editor, I clicked on the point and selected identify. The pop opens with some information and there is a location but it is in a format that google maps will not recognize. To the right there is a drop-down area and selected degrees, minutes, seconds. And the location pops up in that format. I copy and paste it into google maps and press enter. From there I can see if the classification was correct or not. I can zoom in and out and use the pegman. I used the satellite view. If the location contradicted the classification, I referenced the LULC classification summary to determine what the point is really classified as. I did this for all 30 points. If it was true I put a YES in the true_Y field. And if it was no I put a N in true_YN field and also the correct classification in the New_code field. I calculated the overall accuracy. 4/30= 86.7% accurate. 

Monday, April 24, 2017

Unsupervised Classification

Unsupervised Classification

Exercise 1: In ArcMap I completed an unsupervised classification of satellite data using iso cluster and maximum likelihood classification tools.

Exercise 2: I used  ERDAS Imagine to complete an unsupervised classification of an aerial photograph. I accurately classified images of different spatial and spectral resolutions.

  • Reclassification- I determined what feature the new image classes represented.
  • Recoding- I simplified 50 classes into 5 features.

Lastly in Arcmap, I created the map including the essential map elements as well as the percentages of the impermeable and permeable surface from the image.

Wednesday, April 12, 2017


LULC- Land Use and Land Cover

During the exercise, I Identified various features using aerial photography in ERDAS imagine. I started an editing section under create features. In the construction tools, I clicked polygon.  As I created each new polygon I named their code and name. I classified the features up to Level 2 classification. In Arcmap I made a map including all essential elements. 

Thursday, March 30, 2017

Supervised Classification

Supervised Classification 

 I Learned how to perform a supervised classification by collecting sets of pixels on an imported image to define spectral signatures and add these new signatures to an existing signature file. The picture below is the mean plot of all of the classes. I did this by going to the signature editor and display mean plot window. I switched it from single to multiple signature mode and scale chart to fit signatures. I used ERDAS imagine for this work. 

Lastly, I created my own image. I created spectral signatures and AOI features, Recognize and eliminate spectral confusion between spectral signatures, added a distance file image and with all essential map elements. I made the map in Arcmap. 

Thursday, March 9, 2017

Spatial Enhancement

Image enhancements help users to interpret data beyond what is initially apparent in an image. Spatial frequency are gray-scale values that change relative to their neighbors. If there is a slowly varying change in gray scale it has low spatial frequency and vise versa. 

Exercise 1: Acquire Satellite Data

USGS Earth explorer and USGS global visualization viewer.  I learned how to find and download data, look at the station id to know what station the information is coming from and understand what quality bands are. 9 is a band without any errors. 

Exercise 2: Formatting Data

In ERDAS I learned how to batch data to make layers into an image.

Exercise 3: Perform spatial enhancement in Arcmap and ERDAS.

I learned how to use a low-pass filter, high-pass filter and fourier transformation. I learned to the convolution tool. I learned various filter tools in Arcmap and learned about the focal statistics tool.

Exercise 4: Enhancement and making a map. 

Lastly I learned how to correct scan-lines which can cause a striping effect on the image.I used the fourier transform editor to make the files into a file I can edit on. In the fourier transform, I used Wedge and low-pass buttons. I also used the convolution and histogram to make the image pop-out more. 

Thursday, March 2, 2017

Thermal Analysis - ERDAS/ArcGis

Thermal Analysis - ERDAS/ ArcGis

Thermal Energy

On the electromagnetic spectrum, thermal remote sensing is 8-15 ┬Ám. Emissivity is the portion of internal energy of the surface that is transformed into radiation energy. Blackbody can transform all the internal energy and they have a emissivity equal to 1. Water is 0.98. 
Wein's displacement law - lambda (wavelength)= 2897.8/T(temperature).
Higher temperatures emit more radiation. Hotter objects emit peak radiance at shorter wavelengths.

Creating Multi-spectral Images  

In Arcmap under composite bands, I opened 8 raster layers and merged them into a new image  called ETMcomposite. I did the same thing in ERDAS by using layerstack, althought this took a longer amount of time.

Creating a map

I used the multispectral image from the exercise to create a map on a certain feature. I was looking at  the Pensacola area and decided on Pensacola International airport. I went to the inquire cursor and looked under lat/long to get the coordinates.  I soon found the best combination I like which was (4,5,2) RGB. I looked at the histogram and adjusted it as needed to make the airport runway pop more. Once I was content with both the band combination and breakpoints, I created the subset image in erdas, saved it and opened it in Arcmap. From there I opened up the image, set the band combination, and made this map.