This is based on there being a bit of wind, a reasonably clear sky and no snow/frost on the ice.

An ice sheet will, in theory, grow at a rate of roughly one inch per fifteen FDDs starting from ice between 1/2' and 3' thick (as the ice gets thicker the growth rate decreases as a result of the thermal resistance of the thicker ice). For example is the average temperature over a day is 17 degrees that day had fifteen FDDs. Freezing degree days (FDD) are the average number of degrees below freezing over 24 hours. Once the first layer of ice catches on a lake it grows thicker at rate that is dependent on air temperature, windiness, radiational cooling, the thickness of the ice sheet and any snow or frost buildup on the ice sheet.

It is described near the end of this page. It is handy for guessing how much ice will form over night or in the next day or two.

It is most accurate for thin ice (up to a couple inches). It uses a combination of air temperature, wind speed and radiational cooling. The second method is the Swedish method has been developed by Martin Ajne (and others?). It is a good, general purpose method when you do not have a good feel for the conditions other than temperature. One was developed by George Ashton in 1989 and is based on freezing degree days.

There are a couple of pretty good methods for predicting ice growth.

Openoffice calc graph forecast