Some of the reasons RTK fails:
Multi-path: this is the one often discussed but not the only one.
Wrong point: this can get quite complicated, maybe there is a point close by, maybe the surveyor imputs #12 instead of #2, or more difficult to track, there may have been an adjustment to the control point and the pre adjusted point is still floating around the computer, maybe the original autonomous point is still imbedded in the file.
Bad projections: a file can have a bad projection for a number of reasons, I've seen files that the surveyor "thought" were on ground but were actually on the ellipsoid, or files that were calibrated over and over giving good "results" each time but were really messed up. I had one client using one of my projected files that he took hundreds of miles away, boy was that a mess to fix!!!!!
Geoid files not used or used improperly: I imagine many places it doesn't matter, but some areas have a very steep and undulating Geoid slope and no manner of calibration will work over anything but a small area.
Bad IH: Happens-you need to have error trapping to catch it, at the base and at the rover.
Not on the base point: If you are not over the base point just going out and surveying single vectors to RTK points will never catch this error. Had cattle push the tripod over before, still broadcasting, but........
Even when none of the above are present, yet another reason for bad RTK work is a lack of analysis of the actual results to actually check the relative positions of things positioned. In a thread below, for example, I demonstrated that while the screenshot from the Javad software that Shawn posted showed two points as having a relative accuracy of under 0.03 ft. and therefore easily passing the ALTA/ACSM Relative Positional Precision spec, in fact the data he posted showed that the work was right at the edge of failing the spec.
Downloading RTK vectors with full covariance data into a least squares adjustment and to either verify that the vectors are properly weighted or applying a proper scalar to the covariances is the common missing activity. Most commercial LSA software should perform error propagation and allow computation of the Relative Positional Precision of all points positioned by a survey as Star*net does.
Downloading RTK vectors with full covariance data into a least squares adjustment and to either verify that the vectors are properly weighted or applying a proper scalar to the covariances is the common missing activity. Most commercial LSA software should perform error propagation and allow computation of the Relative Positional Precision of all points positioned by a survey as Star*net does.
:good:
The first thing a Surveyor has to grasp to move into the crew chief slot is how to check his work. Every move up demands a new level of understanding of quality control and proper communication of results.
During my career I have watched the tools become more sophisticated. Proper use demands that our education adapt. In my opinion the underlying principles are no different. A braced quadrilateral is still the core of a good network. A check using a different method is still a great idea. A rigorous least squares adjustment still beats every other method. Adapting is still something requiring thought and effort.
Your comment about what software should do points to a major problem moving forward. The various packages available have several common faults. Not the least of these is a refusal to stop with proprietary quality indicators that abuse or ignore basic algebra and statistics. We demand simplification and they deliver.
I've used StarNet for a good while. It is the best I've seen for educating the user and producing sound results. That doesn't mean nothing else works or those who don't use it are inferior. In fact I'm encouraged to see surveyors push back and demand better. The next step is mentoring our way out of the black box hole we've dug for ourselves..
I agree with you that a surveyor checks his work and learns quality control.
I still remember what my dad told me...'I don't care what the data collector tells you..water doesn't flow uphill!' this was right after I staked out a retention basin backwards, I was bout 18 or so at the time and we just got these knew fangled things called data collectors....well I somehow managed to stake the thing backwards, never figured out how, but my point is I trusted the 'black box' and didn't do a simple check.
Lesson learned.