The way to pick the right point of measurement in a Parshall flume is…not to pick at all!
As a short-throated flume, the Parshall flume has only ONE primary point of measurement (Ha). Short-throated flumes control the flow in flow in a very specific region to produce curvilinear results. At any location other than Ha, the level-to-flow equation is NOT valid. To say otherwise is simply not to understand how the flume operates.
If the level reading is taken at some point other than Ha, the flume will under / over read. As the point of measurement gets closer to the throat DOWNSTREAM of Ha, the level will be lower (and the flume under read). As the point of measurement gets further away UPSTREAM from Ha, the level will be higher (and the flume will over read).
For small and medium sized Parshall flumes, Ha is located 2/3 of the distance upstream of the beginning of the throat, as measured along the sidewall of the flume. For larger Parshall flumes, Dr. Parshall had to increase the length of the converging section (usually W/2+4) to obtain satisfactory flow conditioning, but kept the point of measurement relatively close to the throat of the flume.
Long-throated flumes control the flow with a throat that is long enough to produce parallel flow lines in the flow control section. As a result, long-throated flumes such as the Palmer-Bowlus or RBC can have a range of locations where the level-to-flow equation is valid. This location begins D/2 upstream of the flume throat ramp and extends upstream. These are the only flumes where there is variability in the primary point of measurement.
Image: Wikipedia