@User: Probably you're not aware of it, but as part of my work for both MultiGiG and for Analog Devices, I designed several temperature-compensation circuits for clocks coming from a crystal oscillator. Your solution looks very well suited to a temperature compensation scheme. The basic premise is that you have a circuit that encodes a piece-wise-linear or polynomial curve that takes a temperature value, and applies a nonlinear correction, and then applies that correction to a clock generator (such as my all-digital PLL for the striVe processors) to produce a temperature-compensated clock. It's something worth looking into.