In　this　article,　we　establish　the　nonlinear　signal　model　of　the　surface-based　frequency-modulated　continuous　wave　ice-sounding　radar,　and　propose　a　novel　range　processing　strategy　that　aims　at　removing　the　frequency　ramp　nonlinearity　effectively.　The　proposed　algorithm　takes　full　consideration　of　the　dependence　of　nonlinearity　on　the　round-trip　delay　time　while　providing　certain　robustness　of　noise.　The　theory　analysis　and　implementation　steps　of　the　proposed　algorithm　are　demonstrated.　The　full-scale　simulations　with　different　kinds　of　nonlinearities　and　various　signal-to-noise　ratios　verify　the　effectiveness　and　robustness　of　the　proposed　algorithm.　We　also　apply　the　proposed　method　to　real　dataset　collected　during　the　31st　Chinese　Antarctic　Research　Expedition　(CHINARE　31)　and　CHINARE　33.　The　result　shows　the　effectiveness　of　our　algorithm　on　illustrating　clarified　internal　reflecting　horizons　of　ice　sheets.　Compared　with　the　echograms　processed　by　the　typical　range　processing　scheme,　our　algorithm　performs　better　in　nonlinearity　elimination.