Gizmodo's NASA Scientists Find More Subsurface Lakes on Mars links to Characteristics of the Basal Interface of the Martian South Polar Layered Deposits (paywalled, but there is this 52nd Lunar and Planetary Science Conference 2021 conference proceeding by the same authors at the same time)
Water is suspected when there's a region of "shiny" subsurface material reflecting modulated radar pulses (subsurface radar). There are some nice subsurface radar images of ice layers in Mars in this answer to What's the scientific evidence of water for return trip methalox on Mars?.
But what is it that allows one to distinguish with substantial certainty that either reflections from ice, or from these newer "subsurface lakes" are not coming from dispersed metallic particulates or geological structures?
The reasons that ice and liquid water can return reflections are a little complicated, but include that that the individual molecules are polar so the ice or water-containing material is polarizable (dielectric or Fresnel reflection) and also that brackish water is nicely conductive so it can act more like metal in that the wave induces a current and that re-radiates back with specular-like directivity as long as you are below the plasma frequency.
Some minerals can have high dielectric constants and those containing metal particle (e.g. pyrite) can have conductivities comparable to salt water.
Question: How do they know the sub-surface radar reflections (potential underground ice or "lakes") on Mars are not natural dielectric or metallic in origin?
From the University of British Columbia's Geophysical inversion facility page Geophysics foundations: Physical properties: Electrical resistivity of geologic materials
Of all the geophysical properties of rocks, electrical resistivity is by far the most variable. Values ranging as much as 10 orders of magnitude may be encountered, and even individual rock types can vary by several orders of magnitude. The next figure is a representative chart (adapted from [Palacky, 19871 illustrates very generally how the resistivities of important rock groups compare to each other. This type of figure is given in most texts on applied geophysics.
1Palacky, G.V. (1987), Resistivity characteristics of geologic targets, in Electromagnetic Methods in Applied Geophysics, Vol 1, Theory, 1351
presumably "massive sulphides" includes pyrite formations (FeS2 crystals)