Data Availability (DA) and Quality Monitoring (QM) Methodology

Data Availability (DA) Buttons

: Data Availability is greater than 90% the button is green and if 98% it is set to good
: Data Availability is less than 90% the button is set to orange with the corresponding Availability %
: Data Availability is 0%. This means product has not shown up in the desired window allowed.
: Data Stream is not Tracked, but could be produced or it could have been produced in the past, like decommissioned Satellites
: Data Stream or specific product is not Produced anymore or never was.
: To Be Decided and could be tracked in the future, just a place holder generally.

QMetric is defined as 3-day median Availability.

Availability is calculated by taking the ratio of the number of products available in the 24 hour time window allowed for each of these categories (swath files, granules, daily composites etc) to the expected number of products. This is done for each Sensor and Product and Parameter type in some cases, when multiple parameters exist for a single Sensor stream.


Quality Monitoring (QM) Buttons

: Data Quality is good with no significant outliers (greater than 3) in the last 7 days.
: Data Quality has changed, with at least three outliers seen in the last 7 days.
: Data Quality is not available yet, though produced or it could have been produced in the past. It is set to blue unlike in the case of Data Availability as it is less of concern and needs no immediate action.
: Data Quality is not routinely done or the product itself is not produced or distributed at CoastWatch. It is set to blue unlike in the case of Data Availability as it is less of a concern and needs no immediate action.
: To Be Decided for the future, just a place holder generally. The only strong statements for Quality are good or fair for its status.

QMetric in the case of Quality Monitoring is defined to be good when all products are good or else the status changes to fair.

The back-end Analysis subtracts the daily time series statistics from the corresponding 30-day running average to create a "Noise" time series. This product is used to identify suspect and real outliers to make the final decision on the Quality Status of the product. No other reference is used. It only requires the existence of the original time series in retrieval space for any parameter.