Copernicus Marine Service Arctic MFC: Validation of sea ice concentration
Validation results: Regional ice charts
Production update
Sea ice concentration
- An extensive analysis of ice edge displacement metrics validation results from 2017 was performed during the 2018/2019 winter, and the results were subsequently published. The study concluded with a recommendation of a suitable set of metrics that was selected from a perspective on information of simplicity and necessity. This recommendations were implemented in the presentation of validation vs. regional ice charts above starting with the 2019-05-16 bulletin. For the validation vs. OSI SAF data, the recommendations were miplemted starting with the 2019-06-06 bulletin.
- The article where the analysis is presented was published in Ocean Science, full reference:
Melsom, A., Palerme, C., and Müller, M.:
Validation metrics for ice edge position forecasts,
Ocean Sci., 15, 615-630, doi:10.5194/os-15-615-2019, 2019.
-
Starting with the presentation of validation results for the 2013-01-02 bulletin, significant changes to the validation of sea ice concentration and sea ice edge position have been implemented:
- In order to reduce the contribution of results that are far away from the marginal ice zone (MIZ), two new ice classes have been included in the observational product: (1) Ice free near the MIZ, (2) Very Close Drift Ice near the MIZ (see the contingency tables); here "near" means within a distance of 200 km from the MIZ
- RMS of sea ice concentration no longer includes results more than 200 km away from the observed MIZ
- The RMS values are thus not consistent with those computed before this update, and the time series of validation results for sea ice concentration now start with the bulletin from 2012-12-06 (results were re-processed with the modified algorithm for the 2012-12 bulletins)
New products
- Results for the sea ice edge position is now extended by including the position bias. A positive bias means that the modeled sea ice edge is on the open ocean side of the observed edge. Note that this metric can be interpreted as a proxy for sea ice extent.
- We now also provide the bias of the sea ice edge. In the validation algorithm, the sea ice edge is composed of a line of grid cells. The new metric thus gives this bias in units of grid cells (presently, each grid cell represent a square of size 12.5 km x 12.5 km). A positive bias means that the sea ice edge is longer in the model results that in the observational data.
Documentation
These web pages contain an outline of the validation algorithm. Download this document for a detailed description of the algorithm, the observations that are used etc. Furthermore, as stated above, the selection of displacement metrics follows the recommendations provided in this article.
Validation data: Regional ice charts
Validation of the Copernicus Marine Arctic MFC results for sea ice concentrations is performed using the ice chart data from the Copernicus Marine Sea ice & Wind TAC. The region for which the ice charts are available, is shown in the sample snap-shot below. Ice charts are available daily, but not for Saturdays, Sundays and public holidays in Norway. The horizontal resolution of the ice chart data is 1 km.
These sea ice classes correspond to the following sea ice concentrations (discrete values with an interval of 0.1):
- Open Water (OW): 0.0
- Very Open Drift Ice (VODI): 0.1, 0.2, 0.3
- Open Drift Ice (ODI): 0.4, 0.5, 0.6
- Close Drift Ice (CDI): 0.7, 0.8
- Very Close Drift Ice (VCDI): 0.9, 1.0
- Fast Ice (FI): 1.0
In addition, regions with no ice is defined as a separate class Ice Free (IF). Note that our model does not distunguish between classes VCDI and "Fast Ice", so these two classes are also merged into one class in the validation, which will be referred to as Very Close Drift Ice. Note also that for the observational products, classes IF and VCDI have been split according to the distance from the MIZ.
Ice chart data are available from Copernicus Marine product SEAICE_ARC_SEAICE_L4_NRT_OBSERVATIONS_011_002.
Validation results: N. hemisphere SSMIS sea ice concentration
Validation data: N. hemisphere SSMIS sea ice concentration
Validation of the Copernicus Marine Arctic MFC results for sea ice concentrations described above is supplemented by a validation using the OSI SAF multi-sensor SSMIS passive microwave data provided to Copernicus Marine by OSI SAF as the observational reference. This product has a much coarser horizontal resolution than the ice charts, and the model results are not independent of these data. However, in contrast to ice charts, the OSI SAF products covers the entire region with ice infested waters in the northern hemisphere, so results from this validation provide information about the model performance in regions not covered by ice charts.
Validation results are provided for Arctic sectors, and (for 2017 onward) for the 13 GODAE regions.
OSI SAF sea ice concentrations based on SSMIS data are available from Copernicus Marine product SEAICE_GLO_SEAICE_L4_NRT_OBSERVATIONS_011_001, dataset METNO-GLO-SEAICE_CONC-NORTH-L4-NRT-OBS.
Validation results: N. hemisphere AMSR2 sea ice concentration
Validation data: N. hemisphere AMSR2 sea ice concentration
Validation of the Copernicus Marine Arctic MFC results for sea ice concentrations described above is here further supplemented by a validation using the AMSR2 data provided to Copernicus Marine Service as the observational reference. This product is available on the same domain and projection, the same spatial resolution and with the same update frequency as the multi-sensor SSMIS product above.
Validation results are provided for Arctic sectors.
AMSR2 sea ice concentrations are available from Copernicus Marine product SEAICE_GLO_SEAICE_L4_NRT_OBSERVATIONS_011_001, dataset osisaf_obs-si_glo_phy-sic-north_nrt_amsr2_l4_P1D-m..
ARC-MFC PHYS (TOPAZ)
The model results are produced with the TOPAZ ocean data assimilation model system (Copernicus Marine product ARCTIC_ANALYSIS_FORECAST_PHYS_002_001_a). Presently, TOPAZ is run weekly with data assimillation one week prior to the bulletin date, followed by a one-week 100 member ensemble simulation ending on the bulletin date, and finally a 10 day deterministic forecast. TOPAZ was developed and is maintained by the Nansen Center.
ARC-MFC PHYS (TOPAZ) results are available as Copernicus Marine product ARCTIC_ANALYSIS_FORECAST_PHYS_002_001_a. An archive of historical forecasts are available from a thredds server.