is this a one-time measurement or a continuous service? if it is the latter, i would look into optical systems.
also the accuracy needed would be a factor. if it is sufficient to know a ball park number, you could go for counting the number of gel particles and calculating mass by multiplying with the mean particle size (weight). but this would be very ball-park-ish, so it is probably not really suited for your application.
another possibility coming to my mind (also not very precise) would be to weigh the tank content alltogether, measure the volume then get the difference between the actual weight of the content as opposed to the pure water content. the difference would enable you to iterate the actual total weight of the gel if you know its density as well as that of the water. complicated calculations but very easy measurement.
both of these ideas are just spontaneous ideas. don't take it too seriously, but i think either will work, is accuracy is not the highest concern.
just my two cents,
chris