Validation time series

Time series of validation results for forecast lead times

Here are links to time series of aggregated validation results for the various forecast lead times. "Day 1" is the bulletin date. See below for a detailed description. These products are available for the ARC MFC V2 product (onwards), which was first produced for the bulletin date of 2011-12-28.

One year - Day 1 Ocean surface Sea Ice T,S profiles
One year - Day 2 Ocean surface Sea Ice T,S profiles
One year - Day 3 Ocean surface Sea Ice T,S profiles
One year - Day 4 Ocean surface Sea Ice T,S profiles
One year - Day 5 Ocean surface Sea Ice T,S profiles
One year - Day 6 Ocean surface Sea Ice T,S profiles
One year - Day 7 Ocean surface Sea Ice T,S profiles
One year - Day 8 Ocean surface Sea Ice T,S profiles
One year - Day 9 Ocean surface Sea Ice T,S profiles
One year - Day 10 Ocean surface Sea Ice T,S profiles

The satellite based observations for SST was discontinued in April 2012, due to the loss of Envisat. In the summer of 2013, a new recommendation of a remote sensing product for SST was recommended for validation of model real-time forecasts in MyOcean. The recommended product, based on data from the AVHRR instrument on the METOP-A satellite, has been implemented, and validation of model results for SST have been processed from January 2013 onwards.

The loss of ENVISAT also led to a degraded product for ocean color, and the original product was replaced by a MODIS-only dataset in May 2013. Furthermore, the observational product for sea ice drift from the period 2012-04-08 - 2012-06-17 is missing, also due to the loss of Envisat.

Description

Below is a sample of validation results from weekly bulletins (sea surface temperature vs. drifting buoy data; "Extended domain"/"N. Atlantic + Nordic/Arctic"). The sample bulletin date (2012-07-26; corresponding to "Day 1" in the list above) is printed in green at the horizontal axis.

Sample of validation at bulletin

The bulletin date is further indicated by a green vertical dashed line. Time series results that are given here, are only computed for the forecast product (blue curve). There are tick lines for every three days on the horizontal axis.

Note that the week day at which the validation analyses have been performed, changed from Wednesdays to Thursdays in May/June 2012 (the first Thursday bulletin is 2012-06-07). More importantly, the time analysis of analysis was shifted from bulletin - 7 days prior to June 2012, to bulletin - 3 days.

Consider, as an example, the validation results for "Day 7". This day is indicated by the black vertical dashed line in the image to the left. The validation result for this particular day is marked by a blue circle. Hence, this value is inserted in the time series for "Day 7" for this particular bulletin date, i.e., 2012-08-01 (2012-07-26 + 6 days).

The corresponding selection is made for the same forecast lead times (bulletin + 6 days) from other bulletins, forming the time series displayed below. The image above is for the "Extended domain"/"N. Atlantic + Nordic/Arctic". The corresponding date below is again given by a black dashed line, and the value is marked by a red circle.

Sample of validation time series (Day 7)