Ground reference geospatial data collected on Fire Island National Seashore, NY, USA, September 16-19, 2024

PRE-FIELDWORK PLANNING: Prior to fieldwork, four areas of interest across FIIS were identified: (1) the “Lighthouse” AOI that encompassed an area near the Fire Island Lighthouse on the eastern end of the island, (2) the “Sunken Forest” AOI that encompassed the Sunken Forest, (3) the “Otis” AOI that encompassed an area of the Otis Pike Fire Island High Dune Wilderness (Note: the “Otis” AOI was previously called the “Bonus” AOI during data collection as we were unsure if weather would allow us to conduct fieldwork all four scheduled days and it was lowest priority), and (4) the “Breach” AOI that encompassed the western side of a previously breached area that has opened and closed on the eastern end of FIIS over the past few years – also located within the Otis Pike Fire Island High Dune Wilderness near Goose Point. These AOIs were selected based on the variety of land cover they contained as observed in aerial imagery, their geographic spread across the island, and the travel logistics required to access them on the ground. We generated random points within each AOI in our geographic information system (GIS) to create sampling locations, using the “Create Random Points” tool in ArcGIS Pro and specifying a 10 m minimum distance between points to prevent clustering (ArcGIS Pro version 2.8.2). We generated 50 random points for the “Breach” AOI, 50 random points for the “Sunken Forest” AOI, 30 random points for the “Lighthouse” AOI, and 60 random points split evenly across two subareas (30 per subarea) within the “Otis” AOI. Sometimes additional points were manually added to the random point set in and around the AOIs to note areas with, based on aerial imagery, unique looking cover to try to visit them during our time in the field (e.g. we added 11 points to the Lighthouse point set to bring the total number of points up to 41). See file “2024-022-FA_FIIS_Sept2024_AOIandPointMap.jpg” for a map showing the AOIs and the predetermined point locations. The final point count per AOI was 60 in the “Otis” AOI (30 per subarea), 41 in the “Lighthouse” AOI, 50 in the “Breach” AOI, and 50 in the “Sunken Forest” AOI. Shapefiles containing each area’s points were uploaded into projects on the “Emlid Flow” cellphone app (version 11.2), used on two different phones) to use the app’s map feature to navigate to the predetermined locations in the field. All field staff checked that geolocation tagging was enabled in their cell phone camera applications prior to fieldwork.

IN-FIELD POINT SAMPLING: We collected data at one AOI per day. The “Breach” AOI was completed on 20240916. The “Sunken Forest” AOI was completed on 20240917. The “Otis” AOI (both subareas) was completed on 20240918. The “Lighthouse” AOI was completed on 20240919. XYZ survey measurements were collected with an Emlid RS3 Global Navigation Satellite System receiver in real-time kinetic mode using corrections from the National Park Service (NPS) Real Time Differential Service (NPS NTRIP base station FIIS_RTCM3). The GPS measurements were collected using a 5 second average per measurement. The Emlid RS3’s tilt compensation feature was enabled. The point sampling protocol was as follows: 1. Navigate to a predetermined point via the Emlid Flow (Survey Plus) app “stake out points” feature. The “stake out points” mode helps orient you in the field to where a point is and tells you how far the pre-determined point location is relative to your current location. 2. Position a 1 m x 1 m quadrat on the ground with the predetermined location in the center. 3. Confirm that the recorded sample point number in the survey app and in the field notebook match at each location. 4. Place the survey rod in the center of the quadrat and take an XYZ measurement. 5. Take field photos with quadrat in place. These include: a. A photo of the phone application screen with point number visible (this is to help keep photos organized in phone gallery and provide a time stamp close to the survey measurement time stamp for later reference and file organization), b. A context photo, which includes quadrat in larger landscape setting (take a few steps away from the quadrat when taking the photograph) c. A photo showing the top-down view of quadrat, illustrating the contents of the quadrat. d. If applicable: a vegetation height photograph, which shows the vegetation closest to the survey rod with a measuring stick for scale. e. If applicable: a sand photograph, which shows the grain size of the sand present in the quadrat relative to a quarter in the photograph for scale. f. Any other photographs that may help capture unique conditions at the point as needed (e.g. a photograph looking at the sky that shows canopy cover over a point). 6. Take notes and additional measurements: a. Ecogeomorphic State: The ecogeomorphic state present in the quadrat. b. Land Cover 1: The majority (>=50%) land cover type within the quadrat. If the cover in the quadrat was split evenly (50%/50%) between two types, “split” was noted in this field along with the two types of land cover present (e.g. “split herbaceous and shrub”). A visual density guide (Terry and Chilingarian, 1955) in combination with group consensus was used to evaluate this attribute. While the percent coverage they occupied in a quadrat was factored into land cover estimates, bare woody stems and sticks were not considered a land cover class and ignored (e.g. If a dead branch covered 10% of a quadrat and the rest was woody cover, “Land Cover 1” was recorded as “woody”, “Land Cover 2” was recorded as “NA” despite “Land Cover 1” only occupying 90% of the quadrat, and the vegetation density was recorded as 90%). c. Land Cover 2: The second majority (<50% and >=5%) land cover type within the quadrat. If the secondary cover was less than 5%, “NA” was written in this field as the quadrat was deemed close enough to a pure sample. If “Land Cover 1” was 100% cover, this field was “NA”. If the cover was determined to be split in “Land Cover 1”, “split” was noted in this field. If there was a majority land cover type recorded in “Land Cover 1” and the remainder of the quadrat was split about evenly between two types, the two types were noted in this field (e.g. “herbaceous and woody”). As described for the “Land Cover 1” attribute, sticks and bare branches and stems were not considered a land cover class and ignored. d. Plant Functional Group: The plant functional group present within the quadrat. If two functional groups were prominent in the quadrat, they were noted together (e.g. “woody and herbaceous”). e. Vegetation Height (cm): The height of the vegetation closest to the survey rod as measured with a measuring stick. If no vegetation was present within about a 6-inch (15.24 cm) radius of the rod, this field was noted as “NA” as the vegetation height would not be closely associated with the surveyed location. Occasionally, for taller (>2 m) vegetation, we estimated the vegetation height relative to our survey rod height (e.g. “2 ft (60.96 cm) above rod”) and reconciled the note to a number value post-fieldwork. Vegetation height estimates reflect what was on the ground and do not factor in overhead tree canopy. Vegetation height was sometimes not collected due to the presence of poison ivy. f. Vegetation Density: The density of the vegetation within the quadrat. A visual density guide (Terry and Chilingarian, 1955) in combination with group consensus was used to evaluate this attribute. Each evaluator in the group shared their initial approximation of cover without the input of others. The group then discussed any notable differences in their perspectives and came to a consensus for each measurement. As described for the “Land Cover 1” attribute, sticks and bare branches and stems were not included as vegetation in the vegetation density measurements. Vegetation density estimates reflect what was on the ground and do not factor in overhead tree canopy. Vegetation density measurements were only made when a quadrat was used at the sampling location (i.e. no poison ivy was present). g. Substrate: The substrate type present within the quadrat. This was a visual estimate and meant to be a qualitative descriptor, falling into broad categories like “sand”, “mud”, etc. h. Any additional miscellaneous notes. E.g. Is there anything unusual about the point that is notable and is not captured in the other columns? Was this point not at the predetermined location due to the planned location being precarious/inaccessible? Is there mixed land cover to note? Is the point under a canopy and, if so, what was on the ground under it? Are there a notable number of sticks or bare stems present? Etc. When no quadrat was used at a sampling location, it was normally due to poison ivy presence and noted in this field. 7. Confirm the point has an accompanying survey measurement in RTK app to signal that the measurement was successfully recorded, and that the point was visited. 8. Pick everything up and navigate to the next “staked” point. We also took opportunistic point samples, in addition to the random points, as we noticed unique and/or important for our purposes land cover and features in the field. We did not successfully collect measurements at every pre-determined point due to site navigability, time restrictions, and/or lacking GPS signal. This was most notable at the “Sunken Forest” and “Breach” AOIs. GPS signal was poor throughout much of the Sunken Forest area due to tree cover. We collected one beach transect spontaneously at the “Sunken Forest” AOI (point names 27 through 39; taken at 10 step intervals from the wrack line on the beach to the dune crest) after we discovered that we could not reliably get sufficient GPS signal within the forest to take many “fixed” GPS measurements. Five point samples were collected in the “Sunken Forest” AOI entirely relying on our phone GPS to geotag the photographs we took of the quadrat to serve as the primary location data as we could not get sufficient GPS signal for the survey equipment to get a “fixed” signal -- the cover was important enough to document that we decided to try this backup approach in the moment. The “Breach” area predetermined points were, in hindsight, generated too far apart to feasibly navigate to all of them in the allotted time, so we opted to visit a subset of the points across basic terrain types (e.g. marsh, dune, beach, upland, etc.). Eleven of the 50 pre-determined point locations in the “Breach” AOI and 1 of the 50 pre-determined point locations in the “Sunken Forest” AOI were successfully visited in the field, requiring us to heavily supplement the pre-determined points with opportunistic samples to bolster the viable dataset. Some pre-determined point locations throughout all the AOIs were inaccessible due to terrain like dense shrub thickets and poison ivy. In such cases we often took a measurement at the edge of the thicket (as close as we could get to the pre-determined location) and noted this deviation in our field notes. Occasionally we were unable to collect vegetation height and/or vegetation density data due to poison ivy being present at the location. We did not want to risk contaminating the quadrat and/or measurement stick and transfer poison ivy oil onto our hands. Cellphone photographs were taken using either a Samsung Galaxy S22 Ultra (camera model SM-S908U) or an Apple iPhone 15 Pro.

IN-FIELD LANDSCAPE PHOTOGRAPHY: While navigating to, from, and across the AOIs, we opportunistically took cell phone photographs of the landscape and any characteristics or features that would be helpful to document and to quickly increase our coverage of data. These photographs were taken while walking, as a passenger from a vehicle, and from the top of the Fire Island Lighthouse. The photographs from the Lighthouse provide a unique perspective and are landscape photos 863 through 919. Cellphone photographs were taken using either a Samsung Galaxy S22 Ultra (model SM-S908U) or an Apple iPhone 15 Pro.

POST-FIELDWORK DATA PREPARATION: The measurements and notes in the field notebook were transcribed into a tabular spreadsheet alongside the corresponding survey measurements. Minor changes to the fieldnotes were made upon reviewing the data, such as swapping the dominant landcover class from “herbaceous” to “sand” if the primary landcover class had been recorded as “herbaceous” in the field but the vegetation density of a quadrat containing herbaceous vegetation and sand was recorded at less than 50%. These sorts of errors sometimes occurred in the field as we evaluated land cover labels before assessing vegetation density, but they were rectifiable based on the in-field photographs and vegetation density measurements. We also combined the plant functional group classes “woody shrub” and “tree” into a “woody” class, as, in hindsight, we did not consistently distinguish between the two during our in-field evaluations. Any points that were sampled under canopy cover had their relevant land cover class and plant functional group fields changed to “woody”, as we did not consistently label these points with either (a) what was on the ground under the canopy or (b) “tree” or “woody” as the point was under canopy. We opted to recategorize these points as “woody” as that cover type is what would be seen from an aerial perspective, which was relevant for our project goals, and we have the ground photos as reference if the labels need to be reprocessed for other projects and uses. We finished calculating vegetation height values for taller vegetation that was noted relative to our survey rod height. Notes were added to the “Notes” field upon reviewing the quadrat photographs if they were missed in the field, such as noting the presence of the invasive species Phragmites australis or if the presence of bare woody stems and/or sticks influenced the land cover and vegetation density estimates. The AOI names were added to the "Point_Name" values to make each point name a unique value, with no repeating point names across all of the rows of the CSV. The photograph file names were systematically renamed to indicate whether they were associated with a measured sample (“quadrat” image files) or taken as an opportunistic landscape photograph (“landscape” image files). The quadrat photos were renamed to include the field activity number, the location descriptor “FIIS”, and “Photograph Reference ID” (“PhotoRefID”) numbers that correspond to the PhotoRefID values in the survey data spreadsheet, tying the measurements and notes in the spreadsheet to the relevant photographs. The quadrat photos were also appended with a short descriptive phrase at the end to indicate what perspective of the sampling area was shown: “context” for context photographs, “topdown” for topdown photographs, “vegheight” for vegetation height photographs, “sand” photographs for grain size scale photographs, and “log” for photographs of the field notebook, the timestamps of which were used to assure that the proper photographs were assigned to the right survey measurements -- which were also time stamped. Some one-off descriptive phases were used for photographs taken under unique circumstances, such as “nearby” to capture land cover nearby the measurement but outside the quadrat or “canopy” that captured the canopy cover above the quadrat sampling area if it was under trees. If multiple photographs of the same descriptive phrase were taken, they were numbered chronologically from first to last taken (e.g. “topdown1”, “topdown2”). The PhotoRefID numbers were assigned per sampled point chronologically, from first collected to last collected. The “landscape” image files were renamed to contain the field activity number, the location descriptor “FIIS”, the date they were taken, and the word “landscape”. The landscape image filenames were also appended with a unique number assigned to the photographs in the chronological order they were taken, separate from the “PhotoRefID” used in the quadrat image filenames. Cell phone photographs from the Apple iPhone had to be converted from HEIC files to JPEG files while maintaining the geolocation metadata for display in GIS. This was done using the following website: https://heic.online/ (accessed in December 2024). The locations of the photographs (both landscape and quadrat) as originally recorded via integrated cellphone GPS in WGS84 were downloaded from IDS in tabular form via the IDS tool for downloading image details. This table was imported into ArcGIS Pro (version 3.4.3) and displayed as points using the “XY Table to Point” tool. The “Project” tool was used to project the points from WGS84 to NAD 1983 (2011) UTM Zone 18N using the “WGS_1984_(ITRF08)_To_NAD_1983_2011” geographic transformation option. The resulting UTM coordinates were then added to the “2024-022-FA_FIIS_September2024_ImageLocations.csv” file as two additional columns. There is a small (less than 1 cm) difference between the original WGS84 and converted NAD83(2011) coordinates. This difference is within expected accuracy estimates for the positions recorded via integrated cellphone GPS. The addition of the converted coordinate columns to the image locations CSV file was done during data release review, which is why this processing step has a completion date of April 10, 2025. The rest of this processing step was completed January 24, 2025. Note that the EXIF headers as stored in the image files on IDS only contain the original WGS84 coordinates.

IMAGE EXIF TAGGING: All image headers of the renamed image files were edited with exiftool (version 13.00) to add standard USGS tags. The image location EXIF data (latitude and longitude) CRS is WGS84 (EPSG:4326), and is a different CRS from the RTK-GPS measurements as the image EXIF locations were recorded via integrated cellphone GPS.