Tuesday, October 17, 2017

Module 6:

Module 6 highlighted the spatial enhancement features of both ERDAS Imagine and ArcMap.  We began by minimizing the striping produced in Landsat 7 images with Fourier Transformation.  Then with spatial effects and focal statistics, refined the image.

Sunday, October 1, 2017

Module 5a: Intro to ERDAS Imagine

Module 5a was a crash course in waves, energy, and electromagnetic radiation.  We also were exposed to ERDAS Imagine.  With ERDAS we manipulated Advanced Very High Resolution Radiometer (AVHRR) imagery and Landsat Thematic Mapper (TM) imagery.  The following map is a subset image from a larger TM file.  The subset shows 6 distinct area classifications and their acreage found in that subset image.

Tuesday, September 26, 2017

Module 4:Ground Truthing

The goal of Module 4 was to assess the accuracy of the LULC areas I created in last week’s lab.  First, I created a shapefile with 30 randomly selected point to assess the “ground” accuracy of the previously created LULC.shp.  Then I located the image’s location in Google maps.  By using streetview and zooming in, I assessed every sample point to Google maps.  If the google map assessment correctly correlated to my LULC shapefile, I gave it a “yes”.  If it did not, I assigned it a “no” and recorded the correct LULC in the truthing attribute table.  Below is the final product with a list of the sites with their percent accuracy.

Tuesday, September 19, 2017

Remote Sensing Module3: LULC

Module 3’s goal was to make us familiar with land use land cover and distinguishing features on aerial photography.  First, I decided to create the land use land cover polygons at a 1:5,000 scale.  At this scale, individual features were recognizable but it still allowed for a “larger”, bird’s eye view of the image.  I began feature identification with the water bodies and wetlands, since these were straightforward, smooth, black/blue areas to find.  I continued drawing a large polygon around the rest of the image (ie the land).  Since most of the image was residential, I thought it would be easiest to make smaller polygons out of the bigger area designating the commercial, transportation, agricultural, and industrial areas.  In the future, instead of drawing two separate and bigger polygons (in this case one for land and one for water), I would start with one big polygon over the entire image and then start dividing up the smaller polygons.

Below is my land use/land cover map drawn over aerial image 10700320_m5.tif.

Monday, September 11, 2017

Remote Sensing: Module 2 Visual Interpretation

My first map with Remote Sensing highlights differences in both tone and texture in aerial imagery.  Tone is a measure of brightness and five areas on the map have been located showing very light, light, medium, dark, and very dark.  Texture is a measure of smoothness and five areas on the map have also been located showing very fine, fine, mottled, coarse, and very coarse.

The second portion of this assignment was to create a map locating features based on the following criteria:  size/shape, shadows, and patterns.  Three buildings were located based on their larger (by comparison to other features), blocky nature.  I located a palm tree, a water tower, and a building based on their shadows which offer very mirrored reflection of the actual object.  Patterns helped me locate a lightpost (skinny shadows located routinely along a road), waves (oblique, diffuse shadows located along the beach), and a fence (a uniformly, thin, closed, and curvey pattern).  Features can also be identified based on their association with other objects, criteria, location, etc.  I was able to identify a pier based on a long, slender, and straight shadow located over water.  I also identified a subdivsion based on a clustering of smaller structures with small roads connecting them to a longer, main road.

Thursday, August 10, 2017

Applications: Final

Recently, I have been wanting to go on an overnight, camping/backpacking trip.  As I live in lower Alabama near the Alabama/Florida state line, I have spent time enjoying the woods of the Florida panhandle, specifically the Blackwater River State Forest in Santa Rosa County, Florida.  There I have seen signs for the Florida National Scenic Trail (FNST).  As this trail is local, I have decided to do my camping excursion on the FNST that is contained in Santa Rosa County.  I only wish to spend a night or two on the trail, so I want to make these campsites as ideal as possible.  Qualities for my ideal campsite include:

1.  Location to the trail:  I would prefer to stay within half a mile of the trail but obviously, the closer to the trail the better.
2.  Located in state forest/land:  The trail does cross near private property, and I would obviously not wish to trespass.
3.  Not located in a city:  Backpacking is about enjoying nature, so I do not want to camp within one mile of a city.
4.  Location near a stream or pond (ie a swimming hole):  I would like to rinse off after a long day of hiking.  The campsite does not have to be directly on the water (I am willing to walk a little ways from the campsite to the swimming hole).
5.  Located on dry ground:  Although I want to be near water, I do not want to be sleeping in a wetland.  A 50 foot buffer around a wetland should be sufficient to stay dry.
6.  Located in a tree dense area:  I use a hammock to sleep in so I will need trees to hang up my hammock.

After acquiring the appropriate data, I had to analysis each of my campsite prerequisites.  My ultimate goal was to compare all 6 campsite criteria at the same time so I knew that I was going to create a weighted overlay model.  For this model, I would need each criteria in raster form and for their attributes would to be assigned an indexed value.  Before I converted any of my feature layers whose attributes could not be classified into rasters, I added a “Value” column to their attribute table and assigned a value between 1 and 9.  1 being the worst value and 9 being the best value.  I could then create the model and have values to compare and/or adjust to select the most ideal campsites.  The following slides outline the analysis methods used on each criteria.
First I needed to identify where the FNST actually was in Santa Rosa County.  From the Florida trail layer, I selected all portions of the FNST that was contained in Santa Rosa County.  Then I performed an Euclidean distance analysis for half a mile around the FSNT and reclassified the distances to simplify the values (giving a higher value to the distances closest to the trail).
To ascertain state land status, I used a county parcel layer.  I selected the county parcels that are earmarked by the state for parks/recreational and timber usage. I gave park parcels a value of 9, state timberland parcels a value of 5, and all other parcels a value of 1.
Since I know that I will not be seeking any campsites further than two miles away from the trail (even that is a bit generous) and to help narrow down any future analysis, I contained my searches to the county parcels that were within a 2 mile radius of the FNST.  I selected these parcels, exported them as a feature layer, and then converted them into a raster.  I used the 2-mile parcel feature layer as the boundaries for all my future analysis.
After identifying all the cities in Santa Rosa County, I placed a 1-mile buffer around the cities and gave all buffers a value of 1.  By merging the city buffer layer with the county layer, I had something to compare the buffer value too.  Any of the county area outside of the buffer zones received a value of 9.  I then converted that feature to a raster and extracted by mask to the 2-mile parcel feature.  Next, I wanted to look for swimming hole proximity.  After clipping county water bodies to the 2-mile parcel layer, I performed an Euclidean distance analysis for 2 miles around the water bodies. I then reclassified the distances to simplify the distance values (making sure that the higher values were assigned to the distances closer to the actual water bodies).  To establish a campsite on dry ground, I looked for wetlands in the county.  From the wetland layer, I placed a 50-foot buffer around the wetland areas, merged with the county layer, gave the wetland buffer a value of 1 and county a value of 9, converted to a raster, and extracted to the 2-mile parcel feature.  Landcover analysis would help me locate wooded areas for placing my hammock.  I extracted by mask the Florida landcover raster to Santa Rosa County.  I did not reclassify but just added another value field to the attribute table and manually assigned values to the different land classes.  Anything with water, development, or agriculture, I assigned a 1.  Forested areas received a 9 and anything in between received a 5.
I was then ready to create a weighted overlay model that used all the raster layers as inputs.  A model let me perform the weighted analysis numerous times in order to fine tune the percentages for each input. I ultimately gave proximity to the trail 10% weight (I knew that I could walk further from the trail if necessary); 15% weight to proximity to a swimming hole (this too was not absolutely essential); 15% weight to being on dry ground (I will not actually be sleeping on the ground, so I knew I could sacrifice this requirement); 30% weight to being on State property (this was a vital and I actually restricted the value associated with being in a parcel not owned by the state); 15% weight to being outside of a city buffer; and 15% to hammock placement.
From the results of the weighted analysis, the ideal campsite areas are all found within the Blackwater River State Forest.  I pinpointed 4 distinct locations for a campsite all in that northeastern section of Santa Rosa County.  They are spread around the State Forest so that I have options for hiking whatever distance I would like per day.  I have also included their coordinates so that I can easily locate them using a GPS.  If those exact locations turn out to be inaccessible, they are surrounded by other desirable areas that I could search in as well.

The weighted overlay dramatically decreases the area where I would need to look for a campsite.  I would wish to incorporate a base map/satellite imagery map with the weighted overlay results so that I could have better reference points for finding the ideal camping areas.  If I redid the assignment, I would probably incorporate some of the side trails off of the FNST just to know what options for getting off the trail were.  All in all, I am satisfied that when the time comes, I will be able to find the perfect camping sites for my backpacking adventure on the Florida National Scenic Trail.
To look at my presentation, please follow this link

Wednesday, August 9, 2017

Python Module 11: Sharing Tools

Module 11 dealt with sharing Python tools and scripts.  Fundamentally, its about staying organized and referenced.  Below are screen shots of the parameter dialogue box and subsequent randomly-generated point layer.

Here is a Flow Chart for updating the script tool for ArcMap users:

Coming into this course, Python was very much like a venomous snake to me.  Now I consider it more like a garden variety snake, and I am willing to get a little closer to it… but still consider it a snake needed to be treated with respect and vigilance.  Here are my main take-aways from GIS4102 GIS Programming:
1.      Python was very confusing until I started viewing it like a math problem.  Once I began really thinking in the terms of variables and expressions just as with calculus or algebra, then Python became a little less frustrating.
2.      Punctuation and spelling are critical.  Most of my errors or failure-to-runs were from a misspelled word.

3.      Tools within ArcMap now make a little more sense because I can see “behind the scenes” and am a little more willing to explore that environment.