What is your forecast method projecting for September 2016 ?

Very unreliable before July. (more at the end)

2) How did you determine land snow extent on a daily basis?

Linear interpolation

What is the formula you use to estimate future ice loss at any given day ? And does that formula change over the melting season ?

Only daily kWh change. The latitude changes follow roughly the ice edge and main melting areas in the 2007-2015 period. The further back you go in time, the more inaccurate it gets. The first attachment shows that the model still correlates with the actual area, but consistenly predicts a higher minimum.

what does that mean ?

Example: The daily outgoing longwave radiation per m2 is 5.686kWh at 10C (283.15K)

In my Land Albedo calculations (incoming radition) this 5.686kWh is first reached on 14. May. On 20. June it is 6.54 kWh. On 27. July the incoming radiation goes below that value.

As long as incoming radiation is bigger than outgoing radiation then the difference is considered to be energy melting sea ice.

As formula: If(incoming>outgoing , incoming-outgoing , 0)

As I said before this isn't reality and was just used to (curve)fit the June volume loss. If you have much better idea I will try it out.

5) "With four parameters I can fit an elephant, and with five I can make him wiggle his trunk."

I would add even more variables to control his legs, tail and ears. Only then he can be considered an elephant.

OK, seriously what I have actually done is fit my energy calculations to the PIOMAS volume change, since volume change is the heart of my model. My volume calculations also include average temperatures above the ice, for the 2007-2015 period. If I want to model other years I need to adjust these as well.

I never tried to fit my calculations to area change. It just turned out to be pretty accurate in late season if I divide the volume change by average thickness. As you might have seen in spring my model has most of the time a few weeks below actual area and a few weeks above actual area. Only if these average out by July the final area is quite accurate.

If I use my current volume model and the 2015 extent/area ratios then 2016 will reach a 2.66 million km

^{2} minimum. However the June cliff is the most uncertain time and I rate the current error at at least 500k km

^{2}.

-----------------

Edit:

If you have 9 data points and 4 variables

To be more precise: I modeled volume change for every day during the melting season with 5 variables.Thats 185 data points per year or 1480 for the entire 2007-2015 period.