Image Analysis using "Sample Point" Software

Image Analysis of Percent Coverage using "Sample Point" Software

Our group analyzed percent coverage of vegetation for all quadrats within the 17 transects using free software called "sample point". Sample point software can be downloaded through the following website: http://samplepoint.org/index.html. Once the program is downloaded, go to "Options" and select "Database Wizard - Create DB" so that you can create a database name with the images you are wanting to analyze. Create a database name (example: Transect_1). Select "Create/Populate Database" chose the images and then select "Done". You can select more than one image for each database created. For instance, I selected all of the images that were located in one transect so that each image for all 5 quadrats within a transect were analyzed and stored in one excel spreadsheet. Go to "Options" and click on "Select Database". Chose the database you just created. If this is your first time using sample point, you will need to create buttons of the species you are going to be analyzing. Go to "Options" and highlight "Create Buttons" then select "Create Custom Button Files". It is best to abbreviate each of the species' name you are creating a button for because long names tend to get cut off. Save the custom button files once you have entered all of the species' names you will be analyzing. After you have created a database and created a custom button, you need to upload the button file to your database. Go to "Options" and highlight "Create Buttons" then select "Load Custom Button File". Chose the button file that you need. This has to be complete each time you create a new database. You can always create or edit your custom button's file if needed. Now you need to chose the grid size (points to be analyzed for each image). Go to "Options" and highlight "Select Grid Size" then chose the number of sample points. In our research project were used 64 points. Now you are ready to begin analyzing your data using sample point. Click on "Begin" to get started. You will need to zoom into the image until you see a red cross-hair cursor. Enter a zoom value from 1 to 12 and click "Refresh". Click on the species' button that the middle of the Red Cross-hair is positioned on. You can always click "Back" if you chose the incorrect species. Once you have selected species names for all of the grid points and you have exhausted all of the images, go to "Options" and select "Create Statistics Files". Using 64 sample points to analyze marsh vegetation, I was able to complete all 5 images for one transect in approximately 25 - 30 minutes. I believe that the percent coverage accuracy should increase with an increase in the number of sample points being used to analyze the images.   

Data Analysis and Preparation

Brian and Jesse 

Data Analysis and Preparation - November 23, 2013

Brian, Jesse, and I went to the library to work on our project after we spoke with Dr. Smith about how we should analyze our data. Jesse created a bar graph to represent the marsh-edge movement of the two control sites (north and south of the reef) and the restoration site (older northern section and newer southern section). Brian searched for a technique to determine if the two vegetation analysis methods were comparable. He found the Bland-Altman plot which can be used to determine if the different methods are interchangeable/comparable at estimating percent coverage. I worked at compiling our sample point data into one excel spreadsheet.

Day 3 - Marsh Vegetation Data Collection

Jason and Sharilyn analyzing vegetation behind the oyster restoration site

Day 3 - Marsh Vegetation Data Collection: November 16, 2013

Jason and Bryan

Brian, Bryan, Jason, Jesse, Dr. Smith, and I headed out to the restoration site to finish our final day of data collection. On our way to the restoration site, the GTMNERR mule broke down. Brian, Jason, Jesse, and I loaded our supplies onto the golf cart and rode to the restoration site while Bryan and Dr. Smith waited for help. The weather appeared to be better than the previous day, but as soon as we started collecting data another storm started heading in our direction. Bryan, Jason, and I took photographs and collected canopy height for the remaining 13 transects. Brian and Jesse collected measurements of the marsh-edge movement from previously marked quadrats in front of the marsh-edge. They also collected the remaining GPS coordinates for the end of the transects (16 m). Fortunately, with everyone's participation we were able to collect all of our data just in the nick of time before the storm arrived.

Brian and Jesse measuring change in marsh-edge

Day 2 - Marsh Vegetation Data Collection

Boat Generated Waves at the GTMNERR Restoration Site

Day 2 - Marsh Vegetation Data Collection: November 15, 2013

The weather seemed to be working against us on the second day of data collection, but Brian, Jason, and I were determined to accomplish the task we set-out to complete. We arrived at the restoration site early so that we would be ready once it was low tide. Unfortunately, we were only able to complete one whole transect (transect #4) because of the rain. Instead of returning to the GTMNERR Education Center and calling it a day, we changed our game plan. We decided to prepare for our third day of data collection (November 16th) by marking all of the five quadrats for transect 5 - 17 with yellow flags. I used this opportunity to get some really great photographs of the reef at low tide being bombarded by high energy waves generated by boats. The pictures illustrate how the oyster reefs protect the marsh vegetation by creating a barrier against the wave activity.

Brian marking the quadrats with yellow flags

Jason taking images of the marked quadrat

Day 1 - Marsh Vegetation Data Collection

Brian and Jesse analyzing vegetation using the visual method

 

Day 1 - Marsh Vegetation Data Collection: October 26, 2013

 

Previously Marked Quadrat

Our group decided to analyze marsh vegetation at three different locations adjacent to the Tolomato River: north and south of the restoration site (controls) and directly behind the oyster reef. Originally, we were going to analyze 25 transects positioned approximately 25 meters apart, but we decided to analyze 17 transects that were set-up behind previously marked positions by the GTMNERR staff. All transects contained five quadrats positioned at four meter intervals (0, 4, 8, 12, and 16). Each transect began at the edge of the shoreline where the vegetation started (0 meters) and ended 16 meters from the starting point.  One quadrat (1 m²) made from white PVC pipe sectioned into a 16 celled grid was to be placed around each area of interest on the transect. All of the original 25 transects were going to be analyzed using the visual method for determining the percent coverage of each species within the quadrats. However, Jason Lynn (biological scientist at GTMNERR) demonstrated a new technique for analyzing vegetation using photographic images taken directly above the marked quadrats. Percent coverage using the image analysis method was determined with free software called sample point.

Our group collected data on three transects using both the visual and image analysis method to see if they were comparable at determining vegetation percent coverage. We also collected and recorded the average canopy height of each species located within each quadrat.

Visual Analysis Method

The camera being positioned over the quadrat appeared to reduce errors that are generally made with the visual analysis method and decreased the time spent in the field. 

 

Jason collecting canopy height

Lessons Learned

This past Thursday was the poster presentation and I thought it was really neat to get to see everyone's finished projects and the results after so much hard work. It was very rewarding to see how the hard work has paid off and how we can contribute to the scientific community with our research even though we are all still undergrads. To me that is a really unique and captivating experience. That is one reason that I am thankful for this class. Another reason that I enjoyed this class was because it taught me how to be a scientist outside of the classroom. Most of the classes I take may have labs where we do some work in the field but the emphasis is always on what we are learning in the class room. This class reversed this and used the classroom time to emphasize the work we were doing out in the field. I enjoyed this fresh perspective especially since this is what I want to do in my career. This class has taught us how to take what we have learned in the classroom and apply it to real life situations and tests our ability to turn knowledge not into a grade but scientific results. I believe that this skill is invaluable to scientists seeing as much research is done in the field. This class also taught me several of the common methods of sampling used by researchers, such as seining, eckman sampling, transect sampling, collection trays, B traps and many more. We also learned how to collect and analyze water quality data. All of the hands on experiences that I have gathered from this class will definitely be useful later in my career. And for these reasons I am thankful for this class. 

Vegetation results

Our experiment yielded interesting results on the vegetation dispersal in the GTM NERR restoration site. We found that Spartina alterniflora was the most abundant along the shoreline while the smaller plants such as Batis maritima and Sporobulus virginicus dominated the more elevated regions. We postulated that because Spartina is very stress resistant it is able to thrive in these high stress environments. The Spartina also reduces the wave energy which we believe allows the smaller salt resistant plants able to grow behind the Spartina regions. However the results that we were mainly focusing on were not what we had expected. We thought the reefs would act as a wave buffer and make the areas behind them more stable which would allow for sediment accumulation and vegetation expansion. What the data showed was that there was no significant difference between the north site behind the reef and the control sites on either side of the reefs. This was the older section of the reef which was installed about a month prior to the south reefs installation. That is why we were surprised to see that the north section had lost more shoreline then the south section which had the oyster reefs for a shorter amount of time. This was the opposite of what we were expecting. To account for this we hypothesized that because of the reefs small size they may not prevent erosion but merely change the way that it occurs. To better understand our results further experimentation is required.