A few people follow my NRT AWP site and use it as a predictor for future melt. The model though only calculates how much energy went into a grid cell and thus is only good to predict re-freeze. It does not give a good indication about the amount of ice that has melted. Ice-free regions can appear because the ice has melted, or the ice was transported away by winds and currents. If it gets compacted on the other side of the Arctic, it is harder to melt. A smaller area taken up by the ice means less sunlight is available to melt it. This phenomenon is already known in the compaction ratio (area/extent). During the summer, a low ratio will result in more melt.
The new Ice-melt AWP model better visualises this phenomenon by only accumulating AWP if the sea ice concentration (SIC) is above 25%. In case of an ice-free region driven by wind, the new model will remain at zero while the regular AWP model will accumulate huge amounts due to the low ocean albedo. In the Eastern Greenland Sea, the sea ice is continuously replenished and melting throughout the summer. On average this region melts the most ice.
Initially I intended to run this model just as described, but while the maps looked good the values for graphs were not helpful for determining strong melting seasons. Due to its size the central Arctic had by far the most accumulated AWP even though there is still little melt. Instead of an upwards trend in melting over the decades there was a significant downwards trend. This was not a model error. In the past there was simply more ice present to absorb more sunlight. Additionally, less greenhouse gases meant more of the absorbed energy was lost to space instead of melting ice.
To address these two issues, I decided to test the model with a “heat loss to space” component from my Sea Ice Forecast Model for the Sea Ice Prediction Network. The heat loss is 4-6 MJ/m2/day depending on a mean temperature and CO2 level. This level of heat loss is just a third of the measured outgoing longwave radiation for the Arctic. Its purpose was to improve the forecast model, which is only a local energy model without any heat transfer from lower latitudes. Since the Ice Melt AWP model does not include any heat transfer either it seemed appropriate to use the same level of intensity.
The heat loss adjustment per CO2 level is necessary to avoid over prediction of ice loss in the 1980s and under prediction of ice loss in the 2000s and 2010s. In the model the late 2010s at around 410ppm CO2 have an 11% reduced heat loss compared to the early 1980s at 340ppm.
With these model additions the central Arctic now has the lowest ice melt AWP of all regions and the melting energy has a clear upwards trend over the decades. All years with the lowest sea ice extent/area (2007,2012,2016,2019) are also the ones with the highest Ice Melt values. Apart from comparison to other years I would not put any other use case into the presented values. Especially relating negative values with freezing conditions. It is still not even close to a volume model. For sea ice volume we already have PIOMAS, Cryosat2 and my AMSR2 Snow & Ice Volume.
Maps & Graphs:
https://cryospherecomputing.tk/melt-awpDatasheets:
https://github.com/NicoSun/ScienceData/tree/master/Ice_Melt_AWPNRT charts will follow soon