Field Activity 12: Working with UAS Imagery

Introduction:

The class went out to soccer fields near campus and the Eau Claire Indoor Sports Center to gather Unmanned Aerial Systems (UAS) imagery (the same location as in Field Activity 7.  The imagery was gathered by attaching two cameras to a large balloon (Figure 1).  These cameras were set to take images at set intervals of every five seconds.  One of the cameras collected the images and the geospatial information of the images as well, while the other just collected images.  The balloon was walked around by the class to gather images of the soccer fields.  From here the class was required to mosaic the images gathered by the cameras together in order to create a seamless complete image of the soccer field study area.  Professor Joe Hupy instructed the class to explore some of the various software options to mosaic the images together.  He also explained that some sort of freeware would need to be found in order to attach the geographic information to the geotagged images.

The balloon had two cameras attached to it and was walked around the study area by the class as the cameras took pictures at five second intervals.  This is an inexpensive way to collect aerial imagery with little prior training.  However, if there is any sort of wind, a kite may be a better option than the balloon.  (Figure 1)

Methods:

The class gathered at the soccer fields that were to be the study area.  There was little wind and the cold had subsided for the day, it was near perfect weather to launch the balloon, though the wind did pick up later.  The balloon was filled with helium (Figure 2) and raised up while two cameras were attached to the string.  The cameras began taking images as the balloon was raised up.  When at full height, the class was instructed to walk the area of the soccer fields with the balloon and allow the cameras to do the rest.  Joe Hupy iterated that the course should be walked in a pattern that would ensure overlap of the images to aid in creating a 3D view of the area (Figure 3).

The balloon was filled with helium and allowed to float while the two cameras were attached to the string.  One of the cameras was set to collect geotagged images.  The class then walked around the fields bringing the balloon with them.  (Figure 2)

The track log of the balloon shows how the little bit of wind whipped it around as it was far away from the study area at some points and squiggled back and forth.  Thankfully, enough images were gathered of the soccer fields in order to create a mosaicked image of the area.  (Figure 3)

Once an ideal amount of imagery was gathered it was required to mosaic it together to create the final seamless image.  Two types of software that were recommended were Pix4D and PhotoScan.  The class was told to explore these options.  Drew Briski, a class member had some skills in PhotoScan and was able to help instruct the class as to how to mosaic the images together.  Hundreds of images were taken, though if they were all chosen to perform the mosaic, it would have taken hours.  Instead, for the purposes of learning, thirty-two high quality images taken in succession were chosen to best capture the soccer fields.  The images were also chosen from the camera that had the geotagged data with it in order to avoid having to later georeference the output image.  The program that was chosen to attach the spatial information was GeoSetter, a freeware created to work with geodata such as the data that needed to be attached to the camera (Figure 4).

The images were uploaded into GeoSetter and then the geospatial information was added.  As can be seen on the top right, the images were successfully placed in the correct location over the soccer fields.  This means that the information would hopefully not need to be georeferenced.  (Figure 4)

Now that the images that were geotagged were ready to be uploaded into into PhotoScan and mosaicked together according to the process outlined by Drew Briski (Figures 5-8).

PhotoScan at first appears to be a rather difficult tool that isn't very user-friendly.  However, if the right steps are taken, PhotoScan is extremely automated and mosaicking images is a matter of simply pressing buttons as it is extremely automated.  (Figure 5) 

The images have been added by using the workflow tab and selecting align photos.  This is a point cloud created from the images.  (Figure 6)

After a point cloud is created a mesh and texture can be built to help generate the 3D mosaicked image.  (Figure 7)

This is the final product as viewed in PhotoScan.  It's hard to picture the image in real life though as it is oriented extremely strange.  From here the orthophoto can be exported into a TIFF file to be viewed in ArcMap as a raster.  (Figure 8)

The TIFF of the image was then brought into ArcMap to view over a satellite image and compare to see how well the mosaic performed.  However, it was clear that the image wasn't oriented correctly (Figure 9).  The image would need to be georeferenced with the satellite image in order to come out correctly oriented (Figure 10).

When first brought into ArcMap and compared with the satellite imagery, it was clear that the image generated by mosaicking all of the captured images had some problems.  There was some distortion and the image wasn't oriented correctly.  (Figure 9)

After georeferencing the image to the satellite imagery it appears that the image has come out correctly oriented and with minimal distortion.  Six ground control points were used to georeference the image.  (Figure 10)

Discussion:

This process seemed rather daunting when first laid out to the class.  Images were to be gathered using a previously unlearnt technique, then they were to be mosaicked together using a previously unknown software.  This seemed quite difficult when first explained but it turned out to be a rather painless process.  The gathering of the images using the balloon is not a difficult or expensive task and gathered a good deal of decent quality imagery.  PhotoScan turned out to be rather user-friendly in part to Drew Briski providing a good deal of instruction as to how to use it.  Finally, georeferencing the image was an easy process that had been covered before.  It was however surprising that the image needed to be georeferenced considering it had geo-information attached to it already.  Also, the output image is of an impressive quality considering only thirty two of the hundreds of images were used to create it.


Conclusion:

This activity was a good way to teach the class that using UAS to gather images isn't as daunting as it would originally seem.  Some minor training and attention to details can be put together to put together high quality aerial imagery.  Even using the balloon in this case showed that good aerial imagery in a third dimension can be gathered using the simplest of tools, two normal cameras and a balloon.  The use of GeoSetter also showed that the industry is moving more towards freeware, making this whole process much more accessible.

Field Activity 11: Field Navigation using a GPS

Introduction:

This week's activity was a sort of "capstone" to many previously performed activities in the class.  It combined geodatabase creation, ArcPad deployment and use to gather data, field map creation, and field navigation.  The activity involved gathering points in using ArcPad at various stations set up in the woods surrounding the UWEC Priory (see Field Activity 10 for the study area).  The same groups of 3-4 people were assigned and groups were told they'd be navigating the entire course that they'd only navigated a third of in Field Activity 10.  However, this time the groups were allowed to deploy whatever data they felt necessary to a GPS through ArcPad and use whatever maps they wanted.  A large amount of data was provided by Professor Joe Hupy, including the locations of the stations.  Not only would this task be required, but the groups were to be given paintball guns and equipment provided by the geography program and Joe Hupy.

The activity was turned into a competition to see who could complete the course the fastest with rules being set regarding the paintball guns.  The groups would each have a different starting point and would attempt to complete the course from there.  Also, if a group member was shot the group would be required to wait a minute; these minutes would be crucial in determining who would complete the activity fastest.


Methods:

The first step in preparing for the activity was creating a geodatabase to deploy to ArcPad and store all of the necessary data.  The geodatabase was promptly created and the group set to work in trying to decide which data would be used of the data which was provided.  It was decided that the group would create both a map to carry in hand, and deploy data to the GPS.  This was done in order to have as little data as necessary in the GPS to help prevent it from having trouble loading while moving throughout the course.  The map was to contain much more detail and would be referenced if needed.

A geodatabase was promptly created with a domain added with point numbers as short integer coded values between 1 and 15 as there were fifteen points throughout the course.  Creating this coded domain allowed the group to save a little bit of time by just selecting the number from a drop-down menu and avoid error.  With the geodatabase created, a feature class was created in it to gather the station points with the domain properly set (a task learned in Field Activity 6), the group decided that it was crucial to create a path of travel.  The paths were not provided, though the locations of the stations were provided (Figure 1).  The next logical step since paths were not provided was to create paths.  This was done by digitizing lines between all logical neighboring points (Figure 2).

The locations of each station point were given as a feature class by Joe Hupy.  The points were subsequently numbered off and labeled based on the station number.  The Start points were for the previous activity.  In this activity each group was required to start in a different location throughout the course.  Station 5 was the starting point for group 2.      (Figure 1)

Logical paths were subsequently digitized as lines between points that made sense to travel between.  This was done to determine which paths would be the best route by looking at other data.  The red zones are no-shooting zones for the paintball guns.  These zones were to be avoided as much as possible.  (Figure 2)

To help determine the best paths, a slope value was created using a DEM to find the areas with the greatest change in elevation that should be avoided as much as possible to save energy and time.  This, as well as looking at satellite imagery, taking into account distance, and considering the locations of the no-shooting zones helped group 2 decide which paths would be best used and eliminate the rest (Figure 3).  This path of travel, the point locations, no-shooting zones, and the created feature class to gather new points were uploaded into ArcPad and the GPS (Figure 4).  These features were chosen as none of them were too large as to slow the GPS down and they were all the most necessary features.  A map was then created with more detail to help the group in occasions of need (Figure 5).

Three different slope values were assigned based on the slope degree.  Green is low, yellow medium, and red high.  It was attempted to avoid the red areas as much as possible and stay in the green to allow for fast travel.  Also the checkered no-shooting zones were attempted to be avoided.  This allowed for a creation of the red line which is the path of travel.  The idea was that as long as the group stuck to this line, all of the points would be easily found.  (Figure 3)

This is the final product that was uploaded into ArcPad.  It includes the chosen path, the given point locations, and the no-shooting zones.  Also included is the feature class that was to be gathered along the way though it's not pictured here as this is a before image.  These features were all chosen as they were important and not so large as to slow down the GPS.  (Figure 4)

The map created for the activity contained a larger amount of detail than that uploaded into the GPS.  Included in this map is everything included in the GPS, a satellite image of the area, the slope values of the area, and known trail paths in purple.  Also a scale and grid were included, as well as the estimated pace count between each point.  This was referenced several times throughout the activity in order to better situate the group when the info on the GPS was not quite enough.  (Figure 5)

When the class arrived to the location, guns and masks were handed out.  The group booted up the GPS and opened the map file that contained all of the necessary data.  However, the GPS couldn't find the signal because it could not find a conversion of WGS to NAD83.  This caused a problem for several minutes but was solved by opening up a new map and bringing in the required data.  This was a minor problem that was fortunately solved but could have been larger.  Also, it could have been avoided if proper tests were run beforehand.  This is a good lesson to take remember in the future.  The groups set off to their starting points with group 2 heading to point 5.  The point was found easily, along with the first six other points (5, 11, 4, 14, 10, 12, and 13).

The group was already almost halfway through the course when they encountered another team.  The other team held up the group for several minutes and the group actually got separated.  However, they were able to reconvene and continue to the next point, with only one of the group members getting tagged (in the face).  The next point (9) was found with ease as well, though the GPS acted up and had to be saved and reloaded which took up some time.  The path from point 9 to point 8 involved crossing through a no-shooting zone.  However, it was decided to simply go around to avoid any trouble as this area was in the open and it was told to us that it was best to stick to cover and avoid open areas lest people become suspicious of the guns.  This took more time, though point 8 was eventually reached, and the group set off to point 9.  It is here where another team was encountered, holding up the group for another several minutes.  Eventually the two groups passed by each other with a member from group 2 being tagged.  At this point there'd been two skirmishes with two losses.  The group was determined to win the next skirmishes.

Points 3 and 7 were encountered and recorded.  At each station, the group would record the point in the GPS and use the drop-down menu created thanks to setting the domain in order to number the point correctly.  It was near point 15 that a third team was encountered.  This time, all of the opposing team was tagged with one of group 2 being tagged as well (finally success!).  The two groups hung around awhile and chatted about the activity so far.  Though this used time, it reminded everyone that this was really an activity designed to be fun and help everyone use the skills they'd gathered throughout the entire class.

Directly after point 15 was gathered another team was encountered and ambushed with a member of their team being tagged twice.  At this point group 2 had gone through four skirmishes with two wins and two losses (a respectable record).  The final three points were subsequently encountered and recorded (points 6, 2, and 1).  All of the points had then been found and recorded (Figure 6).  Group 2 then returned to the starting location to see that they'd been just beat out by another team.

The blue triangle are the points that were recorded in ArcPad and uploaded into the geodatabase on the computer after the activity was over.  The points are for the most part extremely near where they were supposed to be and all the course points were recorded.  (Figure 6)

Discussion:

This activity was a capstone experience for the entire course and took skills learned in Field Activities 5, 6, 8, and 10 and combined them all into one activity with paintball guns which, of course, raised the pressure.  The initial creation of the database and feature classes along with map design all went extremely smoothly.  However, after he data was deployed to ArcPad, all the group did was open it to see if it was there, they didn't go outside to see if they could get the GPS signal.  This was almost a major problem as right before the activity the GPS wasn't finding, however it was solved by bringing all of the feature classes into a new map.  This taught the group to be sure to always double check and make sure everything works beforehand (a lesson which should have been learned at this point).

The actual navigation went well, with the group having minor problems with the GPS being unresponsive at one point, though this was solved by simply reopening a new map.  Using the paintball guns made the navigation more difficult as carrying he guns and wearing the masks made travel more strenuous.


Conclusion:

This activity reaffirmed many skills learnt throughout the course and even taught some more new lessons.  It also aided in building a stronger camaraderie among the class which is always a good thing to do.  The using of the GPS as opposed to the orienteering method performed in Field Activity 10 was a good contrast and showed advantages and disadvantages of both.

Field Activity 10: Field Navigation using Orienteering Methods

Introduction:

This weeks assignment involved using previously created field navigation maps in Field Activity 5 in order to navigate a course of points at the University of Wisconsin-Eau Claire Priory (Figure 1).  This navigation from point to point involved using orienteering techniques learned in Field Activity 5 and plotted points on a field navigation map to get from one flag station to the next.  The course was set up by UWEC's Joe Hupy and Al Wiberg, with the help of Zach Hilgendorf, a fellow geography student at UWEC.

Groups were assigned to navigate through a set of five stations spread out throughout an are of approximately 64 hectares.  Groups were made up of three to four people to best allow for distance-bearing navigation to be performed.  As mentioned in Field Activity 5, this form of navigation involves having a bearing finder holding the compass, a pace counter, and a runner.  The five points UTM and lat/long locations were given and had to be plotted on the previously created maps.  Each group was assigned a starting position and then set off from there to navigate to each position, with a punch card, which they could fill out at each station along the way, to prove they had completed the course.

Study Area:

The Priory buildings are the center of the chosen study area.  The area of the course that will be navigated is heavily wooded and very hilly.  The Priory itself sits on a hill.  (Figure 1)

The Priory is a real estate subsidiary of the University of Wisconsin-Eau Claire.  It is currently being used as a children's center and partially as a residence hall.  It is located three miles south of the UWEC campus in a forested and hilly area.  Most of the area of the course is heavily wooded (Figure 2) which made navigation in a straight line rather difficult in some areas.  There are more open areas though they are few and far between and the majority of the course navigation involved climbing through branches (Figure 3), over logs and even streams, and going around large trees, which made maintaining a correct bearing more difficult.  On the positive side, the weather was almost perfect for performing a navigation activity such as this one as it was around 60 degrees Fahrenheit and sunny outside.  This is much better than what last years class had to deal with; according to professor Hupy they weren't performing a navigation activity, they were "snowshoeing".  Thankfully the weather held up this week and helped aid in the groups navigation.

The point course was heavily wooded throughout which increased the difficulty of properly navigating the course as it was very hard to maintain a correct pace count and ensure bearing were constantly set correctly. (Figure 2)

Keeping a pace count was difficult as the logs on the ground and high prevalence of branches throughout the group's assigned point course increased the difficulty of the activity.  (Figure 3)

Methods:

When the class arrived at the Priory, Joe Hupy handed out each of the point station coordinates to each of the groups.  From here the groups were required to plot points on both of the UTM and lat/long maps using meters and decimal degrees respectively.  While the group was plotting these points it was almost immediately noticed that the UTM points were off on the map.  All the maps the group had printed off with UTM grids had the origin incorrectly set.  Professor Hupy explained that this is an easy fix and one simple click can correct it, though for this activity the lat/long maps would need to be used.  This sadly eliminated the group's option of using UTM if desired.  As the course will also be navigated in the future, the corrections will be made to the maps in order to use UTM in the upcoming weeks.

Zach Hilgendorf, a fellow geography student at UWEC knowledgeable in orienteering, then went over a review of how to properly use distance-bearing navigation to get from one point to another on a map.  The first step is to assign roles.  The three roles that are crucial in distance-bearing navigation are bearing locator, pace counter, and runner.  The bearing is found by using a compass (Figure 4) and aligning the edge up with the point currently located and the desired location point.  The direction of travel arrow needs to be pointing towards the desired destination.  Then the north arrow on the compass should be aligned with true north on the map.  The needed bearing can then be found by observing the bearing line.  From here the compass can be lifted off the map.  It is then the bearing locator's job to align the red north arrow up with the red north arrow on the rotating bezel (red in the shed).  The direction of travel arrow should then be pointing towards the exact bearing of the desired location.  From here the runner is sent out to a set landmark/point in the exact bearing of the location.  Once the runner has gotten to this point, the pace counter sets out towards him/her and keeps track of the pace in order to have an estimate of distance traveled.  This can be related towards the distance between the points measured on the map.  This is repeated until arriving at the desired location.

This is a typical compass used for distance bearing navigation.  The labeled features that will be mentioned in this write up are as follows:  1: base plate with ruler for measuring scale, 2: rotating bezel, 3: rotating needle, 5: orienting arrow fixed on rotating bezel used to indicate north,6: bearing line fixed on the base plate, 8: direction of travel arrow
(Figure 4)

The group had all the points plotted and distances measured and was guided to a starting location for course two of the three courses scattered throughout the area surrounding the priory.  From here the first bearing was found and set (Figure 5).  As the group looked to find a good point in the direction of the bearing, the first problem came up.  The forested area was so thick and had no prevalent features that could be properly used to send the runner to.  Due to this the group was forced to keep the runner close, within hearing distance, in order to communicate where to properly stand.  There were only about twenty paces between each area the runner was set to.  It was also quickly realized that the pace count of 65 steps equals 100 meters would not be able to be used as the forest was too dense and smaller steps were required.  Eventually it was settled that 90 paces would be set to be about 100 meters.  The first station (station 6 as the second course of five points was being navigated) was located rather easily after about ten minutes of repeating the process of sending finding the bearing, sending out the runner, and counting the pace.  From here a new bearing was set to find the next station.

This is the exact start point of course #2.  From here the first bearing was found using the orienteering techniques previously taught by Al Wiberg and reinforced by Zach Hilgendorf.  The first bearing was set well as the first station was located easily within ten minutes. (Figure 5)

The next station (station 7) was also easy to encounter, though this was a very difficult part of the forest to navigate through (Figure 6).  It took approximately twenty minutes to encounter station 7 from station 6.

What would have been a short stroll between station 6 and 7 was turned into a difficult climb through the thick forest.  Due to this, a walk that would have taken about three minutes in open ground took almost twenty minutes of navigating through the forest to encounter.  However, the seventh station was exactly where it was expected to be, unlike station 8.  (Figure 6)

The next station to find was station 8.  A bearing was set and the group set off, the runner going first followed by the pace counter and bearing setter.  The pace was counted and after observing the navigation map (Figure 7) the group believed they were in the correct location to see station 8, however it wasn't anywhere in sight.  The bearing setter stayed in one place while the other two group members went out to explore and see if they could locate the station.  After extensively searching station 8 was found with the help of another group looking to encounter it, though it was in a location that didn't appear correct on the navigation map that was created according to the given coordinates.

Station 8 was difficult to find.  It is hard to determine if this is due to navigation error or placement error of the point as the actual station seemed to be much nearer to the highway (further north) than the map and given coordinates tell.  The possible incorrect location of this point contributed to difficulty in locating station 9 as well.  (Figure 7) 

At this point, encountering the next point (point 9) was difficult as it wasn't where it was believed to be, even with two groups navigating.  This once again points to point 8 being incorrect, which in turn, made it difficult to find point 9.  Though after about 20 minutes of searching beyond just the navigation, the point was encountered.  From there, the final station (point 10) was encountered easily and the navigation activity was completed (Figure 8).

A different "punch" was obtained at each station.  This card shows the five punches obtained from stations 6-10 proving completion of the activity.  (Figure 8)

Discussion:

The distance-bearing technique used in this activity had its advantages and downfalls.  It was more difficult to navigate using this technique in thick woods than expected.  This is because it was hard for the bearing setter and the runner to completely distinguish a key point to stand to be at the correct bearing due to the large presence of many smaller trees and branches.  It was also difficult to keep an accurate pace count as stepping over logs, around trees, and under branches made it extremely difficult to fully estimate what pace equaled 100 meters.  On flat ground around 65 steps equals 100 meters, the group decided to go with about 90 steps equaling 100 meters, though at times this was inaccurate due to rapid changes in the slope of the elevation or the aforementioned thicket.

Despite these downfalls, the first two points (which were the stations in the thickest portion of the course) were encountered easily due to excellent bearing setting and navigation by the group.  Where the group encountered a problem was with point 8.  The point was far from where it appeared on the map, which was plotted based on given coordinates.  These coordinates were taken with a GPS earlier when the course was set up and could be inaccurate.  Although another possibility is that the group just slipped up in the navigation.


Conclusion:

The activity went extremely well due to the favorable weather and large amount of preparation on the instructors' parts.  The groups all managed to properly navigate through the assigned course with only a few hiccups (darned point 8).  Next week the class will be navigating the same courses but using GPS devices instead of the low-tech distance-bearing navigation.  It will be interesting to see how these compare, especially with the large amount of canopy cover that could throw off the GPS signals.