Noob question but how accurate is the PIOMAS model? If you look at the picture the DMI (center) and HYCOM (right) models look totally inaccurate over the CAB when compared to the Cryosat measured data. It's not reassuring that if those models are inaccurate at high latitudes that PIOMAS is.
Hi Trebuchet. First of all, CryoSat also has its problems and isn't perfect. Personally, I don't put much trust in the DMI and HYOM/ACNFS ice thickness distribution maps, but over the years I have to say that PIOMAS has always given a good feel of the general situation with regards to Arctic sea ice volume. Wipneus posts awesome distribution maps too, but I'm not sure how accurate these are either. But a diffcerence with other models is that observational data (thickness measurements by planes, subs, scientists on the ice) gets assimilated for the PIOMAS product.
HYCOM/CICE models are different in a few respects from PIOMAS.
HYCOM/CICE is a forecast model, it predicts what the state of the Arctic will be tomorrow, based on what it thinks the state of the Arctic is today. PIOMAS is a hindcast model, it calculates what the state of the Arctic was last month based on measurements of what the state of the Arctic was last month. So where measurements are available, PIOMAS is likely to be more accurate. In particular, since PIOMAS uses reanalysed weather, while CICE uses weather predictions, PIOMAS should normally have a rather better idea of what the state of the weather was, and project ice conditions through the month better as a result. So if you want to know what the state of the Arctic was last month, PIOMAS should normally be the better guide. However, if you want to know what it will be tomorrow, CICE will make a guess and PIOMAS won't.
PIOMAS takes a lot more data into account than HYCOM/CICE does, and some of the assimilation in HYCOM/CICE models is problematic.
Both use measurements of ice concentration, but the only other data assimilated by HYCOM/CICE is sea surface temperature, and that's not a good thing to assimilate in areas with ice, because water in contact with ice is constrained to be at its melting point, so you don't actually gain anything by using data, and you can send the model haywire if you assimilate measurement inaccuracies too strongly (as was seen last year when a HYCOM/CICE model sent salinity all over the place to get the melting point of ice to match up with the data it was assimilating).
The ocean structure in HYCOM/CICE has to work for weather prediction as well as heat transfer to ice, while in PIOMAS it just has to work for heat transfer to ice. I think this is behind the tendency of HYCOM/CICE to have a shorter and more aggressive melt season that PIOMAS. CICE puts heat directly from insolation into melting the ice in July, while PIOMAS puts it into heating the ocean and continues bottom melting with that heat into September.
However neither are able to constrain thickness details at all closely from measurement, and both will have a lot of compensating errors (they'll get what proportion is a given thickness about right, but not where it is) and you shouldn't pay much attention to the thickness details as opposed to the broad picture.
HYCOM/CICE if you want to know what the ice will look like in 2 days time.
PIOMAS if you want to know what the total volume was last month.