Because crossings with very tall embankments cannot be measured using a grade rod (even a fully extended one) and hand level, we have taken elevations using an "artificial" eye height through the structure to an observer with a grade rod on the other side. How do we record our data in fields 52 through 55?
This question is related to field 83, which may also rely on a "real" eye height.
This can be very tricky, and is one reason why we made sure to purchase 25' grade rods. Understanding that not every assessment team has access to 25' rods, we know these measurements can be very difficult to accomplish correctly. Therefore, though we would prefer to have that data, we understand if specific sites are too problematic (varied terrain, steep embankments) to capture the elevation data.
If you do try to proceed at such a site, the most important thing is that the elevations are tied into one reference point, so that everything can be "re-baselined" with respect to that point and relative elevations are therefore accurate between the points.
If the elevations can be measured from the upstream road side, that is ideal; however, when they cannot, if a continuous reference point can be maintained across them, and described in the comments, than we can re-baseline appropriately. In these alternative scenarios, thy key then is to describe how the elevations would need to be adjusted to maintain their relative accuracy.
As an example, if at a certain site, the elevations can't be read from the upstream side of the road because the grade road doesn't reach up far enough (to the observer's eye height), but they can be read from 2 feet above the road - you would enter 2.0 feet into the road elevation and place int he comments something like "Read from 2 feet above the road; grade rod height was limited.".
For your situation, which is pretty complicated, the diagram would look like this:
So, for the diagram above, let's imagine the 'sighters' eye height is 5.5 feet. The assessment team would put 5.5 in the inlet invert elevation. For the road elevation, the road height would need to measured from the same point (let's say it is 13.0). The outlet invert would need to be measured from the the same location (at the inlet invert); in our diagram let's assume it was read to be 6.3. And then the hydraulic control, if present, would be read form the outlet invert measurement point. In the comments, you could write 'Invert and Road elevations measured from the inlet invert; hydraulic control elevations measured normally'. Then, in SADES/ArcMap, I would re-baseline like so: inlet invert elevation = 13.0 (original road elevation), road elevation = 0, and outlet invert elevation = 6.3 - 5.5 + 13.0 = 13.8 (original outlet invert measurement - inlet invert measurement + road elevation).
A simpler means of doing the above would be to measure as the diagram dictates below:
In the diagram above, a laser range finder could be laid on the road surface and intersected with the grade rods. In the comments you could write 'Road elevation set to zero; invert elevations measured from road surface user laser range finder'.
Because of how complicated this can be, and how specific you would need to be in the comments to make it clear, it might be better to skip elevations in such a scenario.
Let me know if more information is need.