Your project seems a perfect application of my old patent EP0564012
http://www.freepatentsonline.com/EP0564012.htmlhttp://www.freepatentsonline.com/EP0564012.pdfwhich has expired hence is available for free. The text must exist in other languages.
I considered the deposition of a metal by electrolysis because metal 100nm thin suffices and needs very little current, far less than changing the degree of oxidation of an ion in a thick solution. The display has a transparent upper electrode (consider a metal mesh also, not just ITO), a metal-containnig electrolyte few µm thin, and a back electrode of some colour. The current deposits the electrolyte's metal under the top transparent electrode. Au
2+ reduced to Au for 100nm on 5cm
2, or 1mg, needs 1C or 10mA*100s for instance.
-20°C shouldn't be a worry as the dissolved salts prevent water from freezing. You might use some powder or fabric to hold the water in place, or gel it. Even solid electrolytes exist.
Changing the colour once must be easy, and if choosing the proper metal and ions, 10 years isn't a worry (except to test it quickly). Gold cyanide has the drawbacks of cyanides, but there are more options.
I wanted it as a depletion indicator for telephone prepaid cards when they still existed. Later, people wanted a reversible indicator, which would have been more difficult.
Beware I didn't test it! It looks reasonable to my eyes, the experts here may well have a different opinion, and engineering is always full of surprises.