I think you need to DEM previously to analysis the mask in order to apply the volume difference. I'm afraid that you couldn't identify the pound with this technique.
Well there can be many ways you can calculate the depth and volume of the lake....one way would be by volume area scaling...however the application of scaling methods can lead to underestimation or overestimation depending on the region it is developed for...so i would suggest using an ensemble estimate of the different VA scaling available....then you can use the DEM to draw depth countour lines starting from the lake surface elevation..moving inward towards the center of the lake such that it contains the calculated volume which is the ensemble estimate. This would give a spatially distributed lake depth of a lake...well this is just an alternative...the real Bathymetry is only possible using sonar surveys. An interpolation technique using GIS tools can generate the spatially distributed depth of the lake base on the input of the depth contour lines.
Generally the surface of the lake, although there are fused datasets that show both bathysphere and elevation. And that 'surface' in the DEM needs be interrogated carefully by reading the documentation of the dataset to see how they handled seasonal effects and recharge / discharge of reservoirs. Most global datasets are merges of all sorts of different datasets ( see the 'Index indicating the data source and the number of source scenes' value in the table at https://developers.google.com/earth-engine/datasets/catalog/NASA_NASADEM_HGT_001 ).