Some remarks on the estimation of geothermal topocorrections