Present and Future:

The intent of this research product was to derive a quality-controlled (QCed) bleaching observations dataset that was matched to various satellite spatial scales to validate satellite algorithms that aim to predict aspects of coral bleaching.

NOAA Coral Reef Watch (CRW) released the first version of the QCed Coral Bleaching Observations product (dataset) in October 2012. Subsequent discussions with multiple CRW users and stakeholders resulted in our decision to begin evolving this into a more generic product with broader aims. As a result, Version 2 of this research product (presented here) was released in June 2013. CRW may continue developing this product to suit a range of uses other than just satellites. Ideas for improvements that are already under consideration include:

1. Development of a quality control for reports of no bleaching.
2. Continued inclusion of datasets held by other NOAA divisions and external agencies/partners.
3. Including quality flags for each step in the quality control methodology so that users can choose which tests they wish to include. In other words, all data will be retained and users will be able to choose which quality control (QC) tests to use on the complete dataset. If there is sufficient demand, we will consider developing a web-based GUI so that users can choose which QC tests to implement. The result will be a downloadable bleaching dataset with the user's personalized QC test.

For the current version of this research product, however, QC of the coral bleaching observations data was conducted with satellites in mind; therefore, output was tailored only toward validating satellite heat stress products such as CRW's Degree Heating Week (DHW) and Light Stress Damage (LSD) products (see: http://coralreefwatch.noaa.gov/satellite/index.php).

Purpose:

One of the main aims of NOAA Coral Reef Watch (CRW)'s decision support system for coral bleaching management is to provide managers, scientists, and other reef stakeholders with information about the onset, severity, and duration of heat stress that can lead to mass coral bleaching and mortality. To quantify the ability of a satellite product to provide useful information about coral bleaching events, it is important to have a QCed validation/calibration dataset of in situ bleaching observations that are tuned to the same resolution as the satellite data. The purpose of the Quality-Controlled (QCed) Coral Bleaching Observations product (dataset) presented here is to provide data for calibration and validation of CRW's satellite-based heat stress products.

Many organizations and individuals collect coral bleaching information and data from their local reefs. One of the more well-known, publicly-accessible global datasets is ReefBase, administered by WorldFish. Because of its geographic coverage and popularity, CRW decided to populate the first version of the QCed Coral Bleaching Observations dataset (released October 2012) using ReefBase data only. Bleaching observations data in ReefBase are provided voluntarily by interested persons, regardless of their qualifications of experience; the result is a database with varying quality. For this reason, CRW developed a QC methodology to filter the original ReefBase data and produce a higher quality bleaching observations dataset tuned specifically for validating CRW's satellite products.

For the second version of the QCed Coral Bleaching Observations dataset (presented here), CRW chose to incorporate bleaching observations data provided by coral reef scientists and management partners in the Caribbean during and subsequent to the 2010 mass coral bleaching event in that region. These data were incorporated with the same parameters as the ReefBase data so that they merged easily.

Methodology:

The below methodology was developed and run separately at 0.044 degree (approx. 4 km), 0.1 degree (approx. 11 km), and 0.5 degree (approx. 50 km) resolution to form three separate data files tuned to three different satellite spatial resolutions. Note: the data files at 4, 11, and 50 km resolution that comprise the QCed Coral Bleaching Observations product should be viewed independently of one another, not comparatively. Data were passed through a number of QC checks. Only those reports that passed each step were retained in the final dataset presented here. Although many of these quality control steps can be performed in any order, the QC steps we followed (in order of use for the current version of the QC methodology) were as follows:

1. Since not all bleaching records have a month and year in their associated date (some only have a year), in this first step, all records without a month in their date stamp were discarded. Note that this step was only necessary for the ReefBase data.
2. Since this particular QCed dataset is aimed at obtaining bleaching data only, all reports that were marked as having "no bleaching" (i.e., zero bleaching observed) were discarded. Although "no bleaching" reports are very important for a complete understanding of the relationship between stress and bleaching, a different methodology is needed for analyzing and QC-ing "no bleaching" reports. This will be one of the next steps in the future development of this product.
3. All bleaching reports include a latitude and longitude associated with the report reef. For each quality control resolution (4, 11 and 50 km) all pixels containing reefs were determined using a UNESCO-derived 1 km map of global reef locations (also available from ReefBase). At each resolution, the report latitude and longitude were compared to the location of reefs. If the report did not lie within a reef pixel, it was discarded. Future versions of this product will incorporate the use of the Millennium Coral Reef Map.
4. The temporal scale of a thermal-driven bleaching event is on the order of a number of weeks to months; the temporal scale of most of the ReefBase datasets, in particular, is monthly. As a result, for this step (which applied almost entirely to the ReefBase reports), if multiple reports occurred within the same pixel of a given spatial resolution (4, 11, or 50 km) in the same month, they were treated as a single report.
5. Since we do not have access to satellite remote sensing data prior to 1981, and the current aim of the QCed Coral Bleaching Observations dataset is to validate satellite algorithms, all bleaching reports older than 1981 were omitted. Note that this step was only necessary for the ReefBase data.

Data File Description:

The output from the QCed Coral Bleaching Observations product consists of three ASCII files - one for each of the 4 km, 11 km and 50 km satellite resolutions. Each row within a file is one bleaching observation report from ReefBase or the 2010 Caribbean bleaching event that passed all of the QC tests (described in Methodology above). The total number of rows within each file equates to the total number of QCed bleaching observations at each resolution. The top row consists of the column headings. Column descriptions are as follows:

1. Latitude of record (lat)
2. Longitude of record (lon)
3. Year of record (year)
4. Month of record (mth)

Contact Us:

If users and other interested parties wish to contribute to future development of this research product, please email CRW at coralreefwatch@noaa.gov. Given that we may develop the QCed Coral Bleaching Observations product for other users, including non-satellite users, we are interested to form partnerships to obtain bleaching observation data (including reports of 'no bleaching') and to hear users' suggestions for future directions.