Howdy, I'd be tempted to go chemo-thermal: a slow exothermic reaction inside an insulation box And outside another insulation box. I'd gather a representative temp curve from some weather people then talk to some chemical engineers about reaction rates & civil engineers about insulation characteristics of temperature gradients & heat transfer.
Between the two, you should be able to get a good guess on what'll work. Next step is to stage the temps to be encountered (serious refrigeration) & try some combinations with an RF temp gauge and see what Really is workable. Good Hunting... <<<)))