National Marine Debris Monitoring Program: citizen science observations from 1997–2007

Metadata:

Identification_Information
Data_Quality_Information
Entity_and_Attribute_Information
Distribution_Information
Metadata_Reference_Information

Identification_Information:

Data_Quality_Information:

Attribute_Accuracy:

Attribute_Accuracy_Report:: The GPS coordinates for each monitored beach are considered to be accurate and reliable.

We do not have actual accuracy measures for the debris item counts. What is known is that the Ocean Conservancy defined an error rate. The maximum allowed error rate for this study was twenty per cent (20%). This was monitored by quality assurance checks run on a random sample of site surveys performed by each field survey director. When the computed error rate exceeded 20%, the survey director performed a recount (recall that all debris items were removed from the beach and so were available for the survey director to check). The final report to EPA stated that "in most cases the QA data card showed less than 20% error". This error rate was constructed as follows:
Define Error 1 = the difference between the recorded total count of indicator debris items for survey X and the actual value.
Define Error 2 = the count of indicator items misidentified by volunteers during survey X.
Define QA Points = the actual number of indicator items present on the beach segment for survey X.
Define Error Rate = (Error 1 + Error 2)/(QA Points).

Logical_Consistency_Report:

The core of the dataset - the indicator item counts - is logically consistent.

When wind speed was left blank, wind direction was also left blank. For the 680 data cards on which wind speed was coded "1" (i.e., no wind), 172 cards nonetheless recorded a wind direction. While this represents a logical inconsistency, we could not tell whether the recorded wind direction was the error or if the observer paid insufficient attention to the wind speed scale printed on the data card and thought a code of "0" would represent no wind, and a code of "1" would represent some wind. The other 508 cards recorded "blank" when wind speed was coded "1"; this is logically consistent.

Data entries when a wind direction was identified were easy to understand 99.6% of the time. The other entries listed what appear to be compass degrees or used descriptive terms rather than a direction.

Survey end times, when provided, are always later than survey start times.

Completeness_Report:

The dataset is complete as far as the collected data allow. The indicator item counts are complete. Weather data were occasionally not recorded by the volunteers. Of the 7,555 surveys in the dataset: 354 surveys did not include air temperature.

175 surveys recorded “blank” for both wind speed and wind direction; wind speed should have always had a value, so one cannot distinguish between simply not recording values and the observer assuming that two blank entries would be interpreted as "no wind", therefore no wind speed and no wind direction.

508 surveys recorded “blank” for wind direction when wind speed was recorded as greater than zero; this represents missing wind direction data.

258 surveys did not include weather at the time of the survey (note: all surveys did have coded weather at the time of the survey).
757 surveys did not include weather in the week prior to the survey (note: all surveys did have coded weather in the week prior to the survey).
194 surveys failed to record the survey start time; whenever survey start time was not recorded, survey end time was also not recorded.
190 surveys failed to record the survey end time when the survey start time had been recorded.

Information about the areas used for survey sites is complete for all survey sites used in published journal articles. Of the 160 survey sites whose data are reported in this publication, 78 sites were not used in the journal articles. A subset of those sites lack some site information. Whether the site was categorized as beach or reserve is undetermined for 73 of the 78 sites; spatial location of the survey site (latitudes and longitudes) is incomplete for 16 of the 78 unused sites.

Lineage:

Methodology:

Methodology_Type: Lab
Methodology_Description:: PLANNING THE SURVEY
The National Marine Debris Monitoring Program was developed to create a standardized and statistically valid method of monitoring marine debris. As a result, a Marine Debris Monitoring Workgroup – comprised of representatives from NOAA, the National Park Service, Ocean Conservancy, U.S. Coast Guard, the Marine Mammal Commission and selected scientists – developed a protocol and monitoring methodology that was reviewed by all federal agencies that monitor marine debris.

Site selection: Within each defined continental region, candidate sites were identified based on the presence of 1,550 meter shoreline units that had sand or small gravel substrates, moderate slope, facing open ocean, free of regular community cleanups, and not affected by high-usage recreational activities. (A complete list of criteria is provided on page 25 of the Escardó-Boomsma et al. (1995) report, which is provided in the data publication package.) From these regional candidate site pools, sites were randomly selected for each region.
Methodology_Citation:

Methodology:

Methodology_Type: Field
Methodology_Description:: SITE PREPARATION
Each survey site was measured and marked to assure length accuracy and repeatability of successive surveys. A surveyor’s measuring wheel was used for accurate site measurement. Semi-permanent markers were placed at the beginning and ending points of the 500-meter study site. In the event that it was not permissible or possible to mark a study site, a natural or man-made landmark such as a large rock or building was accurately noted. In addition to establishing visual boundary markers for each site, global positioning system (GPS) coordinates (latitude and longitude) were also recorded for use in possible GIS mapping of the study sites and database manipulation.

Photographs of each selected survey site were taken, noting unique features and landmarks that would identify the site’s location, along with securing a detailed map of each location. Volunteers conducted initial beach cleanups at each of the survey sites to clean the beach of all debris that had accumulated over an unknown time period. This cleanup provided a “clean slate” in preparation of the subsequent debris survey and beach cleanup to be conducted 28 days later. Volunteer groups also used this initial cleanup as continued training for proper debris identification and as a practice run for future surveys.
Methodology_Citation:

Methodology:

Methodology_Type: Field
Methodology_Description:: CONDUCTING THE SURVEY
Surveys were conducted on 28-day intervals. Volunteers had up to three days before to three days after the scheduled survey date to complete the survey and beach cleanup. The time to conduct each survey varied from site to site, but on average each survey took one to two hours to complete. One volunteer on each survey was designated as the data recorder, and filled out the standardized data card (example from 2001 is provided in this publication) and turned the cards in to the site’s survey director.

All indicator and non-indicator debris items were physically removed from the site at each survey.

SURVEY WALKING PATTERNS
Method 1
Volunteers sweep up and back from the water’s edge to the back of the beach along the 500-meter study site. Volunteers or sweeps should be spaced about every two meters (approximately six feet). This method was determined to work the best for small groups of volunteers (two to five people).

Method 2
Volunteers line up, spaced about every two meters (approximately six feet), from the back of the beach to the water’s edge and sweep along the length of the 500-meter study site. This method was determined to work best with larger numbers of volunteers (>5 people).
Methodology_Citation:
Methodology_Citation:
Methodology_Citation:

Methodology:

Methodology_Type: Lab
Methodology_Description:: QA REVIEW POST-PROJECT
The Excel file that held the base data set was subjected to a quality assurance review by Rugg and Ribic. When questions were raised about a particular record, Sheavly checked the original data card to determine whether a fix was needed.

Fewer than thirty-five (35) data cards warranted a review. The review resulted in elimination of a few sites due to interference from other volunteer groups that cleaned the survey beach, and thereby disrupted the monitoring protocol. For the same reason, some individual surveys were deleted for some sites.

The published dataset does not include any of the disrupted sites or surveys.

Process_Step:

Process_Description:: EFFECTS OF SAMPLING INTERVAL
The sampling protocol specified completing a site survey every 28 ± 3 days. However, the volunteers were often not able to maintain this schedule across the entire sampling period. Thus, there was a question of whether longer inter-sample intervals affected the amount of debris sampled on the beach. If true, then either data would need to be removed from the analysis set or additional steps would be needed to compensate for these effects in order to use the data properly.

To assess whether there was an inter-sample effect, inter-sample lengths (days) were calculated for each of the surveys at sites with long observation records. The first observation in each series was assigned an inter-sample length of 28, reflecting the protocol of cleaning the beach then doing the first sample 28 days later. The inter-sample lengths were binned for analysis. The starting bin was the protocol range of 25 – 31 days; this was extended backward and forward until 26 bins were created (roughly covering 6 months). Inter-sample lengths greater than 185 days were dropped into a single bin.

79% of observations were in the target window; 5% missed the window altogether and executed the next observation at the next regular sample interval (56 days). Five percent (5%) of observations were taken with a 21-day interval; 6% were taken with a 35-day interval.

Analysis of the ocean-based indicator items found no significant effects due to inter-sample length in Regions 1,2, 3, or 5; significant effects were found in Regions 4, 6, 7, and 9. The problem intervals were either very short (≤14 days) or very long (>180 days).

Therefore, surveys separated by fewer than 20 days were removed from the data set. For long gaps in the series, the first survey was removed, which restarted the series (i.e., the first survey was treated as the clean-up survey in the protocol for initiating monitoring at a site).
Process_Date: 2008

Entity_and_Attribute_Information:

Overview_Description:

Entity_and_Attribute_Overview:: \Data\NMDMP_Site_Coordinates.csv: Comma-separated values (CSV) file containing geographic coordinates and other basic information about the 160 survey sites.

Variables include:

Region = U.S. coastal region code; integer values from 1 through 9

RegionDescription = high level description of the geographic extent of the region

SiteNumber = survey site code that is unique across all regions; 160 sites that are integer valued

SiteName = name of survey site structured as "beach/reserve name, state abbreviation"

SiteDescription = text description of the larger beach/reserve on which the survey site was located

SiteType = Beach (63 sites), Reserve (24 sites) or empty (73 sites); empty cells denote unknown site type

Article = code denoting whether the site was used for the Atlantic Coast article (A; 41 sites), the Gulf Coast article (G; 18 sites), the Pacific Coast article (P; 23 sites), or no article (N; 78 sites)

LatitudeA = latitude for the north or west end of the survey site; decimal degrees

LongitudeA = longitude for the north or west end of the survey site; decimal degrees

LatitudeB = latitude for the south or east end of the survey site; decimal degrees

LongitudeB = longitude for the south or east end of the survey site; decimal degrees

Source = source of latitude and longitude data: from the volunteers as entered on the data card (GPS), from the web (Web), or no source ("Not locatable on Google Maps")

Notes:

Twelve sites hold LatitudeA and LongitudeA data that were obtained from the web by the authors to locate the beach/reserve; for these sites, there are no LatitudeB or LongitudeB entries.

Three sites hold no latitude or longitude data because no data were entered on the data card and the beach/reserve location could not be found on Google Maps.

One site provided LatitudeA and LongitudeA data on the data card, but no LatitudeB and LongitudeB data.
Entity_and_Attribute_Detail_Citation:: none provided

Overview_Description:

Entity_and_Attribute_Overview:: \Data\NMDMP_Analysis_Surveys.csv: CSV file containing the data for the 7555 site surveys.

Variables include:

Region = U.S. coastal region code; integer values from 1 through 9

SiteNumber = survey site code that is unique across all regions; 160 sites that are integer valued

Date = date of the debris survey being reported; MM/DD/YYYY

Season = text value denoting the terrestrial season: spring (March 21 - June 20), summer (June 21 - September 20), autumn (September 21 - December 20), or winter (December 21 - March 20)

Temperature = air temperature at the start of the survey; degrees Fahrenheit

Wind_dir = wind direction, free text

Wind_speed = coded value for wind speed, integer-valued from 1 through 5

Weather_now = free text description by the volunteer data recorder of the weather conditions at the time of the survey

coded.Weather_now = standardized coding of the free text description; 12 possible text values (“approaching storm”, “cold front”, “drizzle”, “foggy”, “heavy seas”, “none”, “rain”, “snow”, “storm”, “strong winds”, “tropical storm”, “tropical storm/hurricane offshore”)

Weather_prev_week = free text description by the volunteer data recorder of the weather conditions over the week prior to the survey

coded.Weather_prev = standardized coding of the free text description; 11 possible text values (“cold front”, “drizzle”, “foggy”, “heavy seas”, “hurricane”, “none”, “rain”, “snow”, “storm”, “strong winds”, “tropical storm”)

Time_Start = start time for the survey; local time formatted for 24 hour clock HHMM

Time_End = end time for the survey; local time formatted for 24 hour clock HHMM

The next 31 columns are the integer counts of indicator items found during the survey (see Table A4 in the Atlantic Coast article, Ribic et al. (2010), for details on the indicator items; the NMDMP Volunteer Handbook also contains descriptions for some of the items)

TotalItems = Count of all indicator items found during the survey; integer
Entity_and_Attribute_Detail_Citation:: National Marine Debris Monitoring Program. Volunteer Handbook. 12 pp. (available in Supplements folder: NMDMP_volunteer_handbook2.pdf)
Entity_and_Attribute_Detail_Citation:: Ribic, Christine A.; Sheavly, Seba B.; Rugg, David J.; Erdmann, Eric S. 2010. Trends and drivers of marine debris on the Atlantic coast of the United States 1997-2007. Marine Pollution Bulletin 60: 1231-1242. https://doi.org/10.1016/j.marpolbul.2010.03.021

Distribution_Information:

Metadata_Reference_Information:

Metadata_Date: 20220415

Metadata_Contact:

Contact_Information:

Contact_Person_Primary:

Contact_Person: Dave Rugg
Contact_Organization: USDA Forest Service, Research and Development

Contact_Address:

Address_Type: mailing and physical
Address: USDA Forest Service
Address: 1 Gifford Pinchot Drive
City: Madison
State_or_Province: WI
Postal_Code: 53726-2366
Country: USA

Contact_Voice_Telephone: 608-231-9234

Contact_Electronic_Mail_Address: david.rugg@usda.gov

Metadata_Standard_Name: FGDC Biological Data Profile of the Content Standard for Digital Geospatial Metadata

Metadata_Standard_Version: FGDC-STD-001.1-1999