Wow, 9 pages on this. There are two things going on that affect any driver at various environmental conditions. Compliance will vary with temperatures and humidity and resistance of the coil will change with temperature.
First lets look at Cms. This is the compliance of the driver. The materials chosen for the suspension will determine how much this value can vary. In general, foam surrounds have less variance with temperature than do rubber surrounds. It also depends greatly on the type of rubber. Natural rubbers can often get very stiff at colder temperatures and often have issues with cracking. I had a chart at one point to show the working temperature range of different rubber materials, but can't find it at the moment. This page talks about some gasket materials that apply as well.
http://www.sealsfast.com/pages/materials/materials.html
NBR has a brittle point of -22F meaning it will crack and can shatter at any point below that. Also as you are approaching those temperatures it will get increasingly stiff. EPDM is a thermoplastic elastometer which is good to -49F. Santoprene has a working range of -81F to 275F with no cracking or tackiness. This makes it the most reliable of rubbers and it is also going to be the least affected by cold temperatures. It also happens to be the most well damped as well. The combination of damping and temperature range is why we use it for our AV series.
That all said, you need to look at the effects of compliance changing with temperature. As the suspension gets stiffer, Fs goes up, Q goes up, and Vas goes down. The change this gives will depend on the type of enclosure. In a sealed enclosure this means that Qtc goes up and the Fc point goes up. Below Fc you will begin to lose some output due to the stiffer suspension. You can model the effects by cutting Cms in half and seeing what happens in your enclosure. Keep in mind that this won't be exactly a linear loss either. As flexibility is lost in the surround the effects will be higher at larger excursions as more of the surround is required to move. This means the overall loss of output can be more than the model would suggest at higher levels. In the following graph you can see the results on the impedance curve and response when effectively doubling the stiffness of the AV12H suspension in a 1cf sealed enclosure.
In a vented enclosure it is a little different. Due to multiple peaks in the impedance curve you end up with some points of higher output and some of lower output. The following graph shows an AV12X in 2cf tuned to 30hz. The blue curve is with the stiffness of the suspension doubled. You can see that at the points where the impedance curves cross is where you will see a transition from one driver having more output than the other.
Now the other change as mentioned is in the resistance of the wire. As the temperature goes down, so does the DC resistance (Re) of the driver. The motor strength of a driver is typically stated as Bl^2/Re. The B is the amount of flux in the gap measured in Tesla and L is the length of wire in the gap measured in meters. As you change the Re of the driver due to temperature, the B and L will not change. As temperature lowers so does Re and as a result your motor strength goes up. On the other hand as temperature goes up, so does Re and motor strength drops down. This is one of the variables that is difficult to account for and part of the reason many SPL competitors believe you cannot believe what a simulation will claims possible. Not only does the motor strength drop, but as resistance rises, your amplifier now cannot deliver as much power to the driver. This is known as power compression.
Now lets take a look at what would hypothetically happen if the same sealed and vented enclosure comparisons were to happen with something like the NBR or butyl rubber that can get extremely stiff as it approaches it's brittle point. This shows the combined effect of higher motor strength as well as the extreme rise on stiffness of the suspension. The green is the original driver in both applications.
You can see that the lowering of DCR and more motor strength give more efficiency and output up higher, but the extreme stiffening of suspension causes for some serious loss in output at the real useable frequencies. There is a lot to be said for proper selection of materials for a given temperature range.
John