NSIDC melt area is almost unchanged at 25%.
As Friv has posted above, DMI shows melt increasing significantly. Daily SMB loss increased to 9.5Gt, the daily peak mass loss to date this season.
The only realistic way to track GIS is to wait for grace data 3-4 months after real time.
But combing modis, jaxa, temps, promice, and so on we get a decent idea of how it is going.
But even on DMI one method shows 2014 with 2012 or even above a bit another shows 2014 well behind 2012.
I tend to work backward from the DMI SMB model (right or wrong) and them look for the reasons for any change. I think the model responds fairly well to any change, and has most of the inputs you are looking at. Of course this only works for hindcasts, not forecasts.
I don't tend to look at the Total Loss model as often, as you say it is running above 2012 at the moment. Looking at the model guidance below the graphs (copied below) it says that the only inputs are Albedo and GRACE data. Albedo is very high this year so could be affecting this in spite of the colder conditions across the earlier melt season. We'll have to wait for the GRACE data to confirm what is happening
Method 2 is developed by scientists at GEUS and combines past gravity measurements from GRACE with satellite measurements of ice-reflectivity (albedo) and this makes it possible to get near-real-time estimates of the total mass change. The albedo data is retrieved from the MODIS sensor on the NASA Terra satellite.
With forecasts inertia due to ice/snow conditions is a real issue. After a large storm it can take several days in the melt season for the melt to return to pre storm values due to lower albedo etc. I think this is the reason why some of your melt estimates based on mainly on the weather are slightly too high, the weather simply does not have a chance to overcome the inertia of the current ice/snow conditions before returning to lower temperatures.