Please be a little more specific, as you have only spoke in generalities.
What is the linearity and repeatability spec of your device?
(I'll answer my own question :
Accuracy +/- 0.2 % of max range.
Resolution: 0.079" (2mm)
Dead Band: 8" (20cm)
What is the resolution of your analog input when using a 4-20mA range?
Give real world numbers for the amount of fluctuation you are seeing. "Not a lot" also tells us little.
What is the frequency of the fluctuation? Can you put a scope on it or does the IDEC software allow you to trend it a high enough frequency to see what is going on?
We need to determine, based on your resolution, what level of fluctuation you should expect.
We need to determine if this is high frequency noise, or somthing low frequency such as waves in the tank.
Ultrasonic sensors by nature are not super precise. I would expect fluctuation.
Regarding your question about calibrating to a table, that depends both on how accurate you need to be as well as if the tank is irregular. If the tank gives you a very linear relationship between displacement and volume and repeatability is more important than accuracy, by all means stick with the equation. If your tank is irregulary shaped or you need precision beyond the linearity specification, then you should calibrate to a known standard and extrapolate between data points.
In either case, the calibration and scaling will not affect noise or fluctuation, so if that is why you are thinking about doing it there is no point.
I personally am not a big fan of filtering a signal just to make it look more stable that it actually is. There are times when you may need to, but all too often it is used to cover up a problem.
If every inch is 72 gallons, then every tick of the sensor ouput is worth 0.079* 72 = 5.6 gallons. Thats just based on the field devices resolution.