Top graph, page 2 of PDF linked above. Alkalines are worse than NiMH in cold. I've heard that Eneloops and other LSD NiMH fair a little better than standard NiMH in cold, but I really don't know. The electrolyte is a number of different chemistries freezes and crystallizes at low temperatures. Li-Ion fairs better, but will also eventually freeze.
Sanyo advertised better low temp performance with eneloops, but I've never seen any document ion to quantify "better". I've done discharges at 55 deg, 39 deg, and 27 deg and -5 deg F with nearly no appreciable reduction in capacity. Keep in mind these were "low tech" discharges in a first rev Maha. I did use a thermometer to verify temp at the beginning and end of the cycles, and I allowed the hardware to reach ambient temp before beginning the test. I ran the tests with Eneloop and Enduro cells. I did notice that the -5 F test ended relatively warmer at the end than the others, mostly due to the plastic bag I used in my freezer, limited airflow and the plastic provided some insulation so ending ambient temp was about 15 F.
Take this for what it's worth... I wish Sanyo would publish temperature specs on the eneloops.
Maybe this winter will be cold enough for me to do it again in the below zero range, this time with a fan blowing over the unit to keep things closer to ambient as it's something that just doesn't work quite right in the freezer.