Hey everyone,
I've got a question about surveying and photogrammetry and I'm hoping someone here can help me out. My knowledge in land surveying is a bit out of date – I took a couple of courses a few years back and have been more into GIS stuff since then. So, I might get a few things wrong and I'm sorry about that in advance. We're thinking about using drones and photogrammetry for measuring stockpile volumes at my company, instead of getting survey companies to do it. I'm trying to figure out how accurate and precise our results would be.
Right now, I'm using a Trimble R8 receiver with a CORS network for collecting RTK checkpoints. The standard deviation for these points averages around 0.025m horizontally and 0.04m vertically. I'm wondering if I should be using the 95% confidence interval instead of the usual 68% they give us for precision? Also, am I correct in assuming that because we are only calculating volumes we need to worry to about absolute accuracy?
Next, I do a drone flight with a PPK solution. In post processing I compare the differences in distance/elevation between our RTK checkpoints and the drone checkpoints, and this gives me an error of about sigma 0.006m in X/Y and sigma 0.015cm in Z. Should I be mixing these standard deviations to work out how much the volume might vary? Is this even the right way to do it?
Also, how should I present the accuracy of what I'm measuring? We will still have licensed surveyors surveying stockpiles whose values I can compare to, but they only give the estimated volume in cubic meters, without any precision or accuracy details. Any advice would be awesome. Thanks a lot!
Also: I think most of the time I'll be doing these measurements without any specific base elevation values for the stockpiles. I'm basically assuming that the ground is flat. I'm planning to use PIX4D and its algorithms to calculate the cut and fill, relying on their automatically calculated base height.
Without a base surface before the stockpile was placed you will only be guessing at your volume and how accurate that volume is. I think you made a typo in the vertical difference between the photogrammetry and RTK shots. There is no way you are finding a difference of 0.00045'
I ran an aerial photogrammetric mapping firm in the 1970s. We did dozens of aerial stockpile surveys of construction materials, wood chip (pulp) piles, etc. At the time, such photogrammetric surveys of sites with control provided by licensed Land Surveyors (and Targeted) flown at 1500 ft with a 6-inch calibrated focal length camera was generally accepted to be accurate to within 2% of actual volume. In some areas of South Louisiana with poor soils, the weight of the materials caused significant settlement below (original) grade such that results were useless. We were paid, but there were no repeat calls for our services. I have no idea what the company finally did regarding their final sale of the site and volume of inventory. (Location was along the Industrial Canal in New Orleans East.)
You can take a look at the attached ASPRS standards for a general idea of what's needed. You have a steep learning curve ahead of you. There are many variables that influence the final accuracy such as flight speed, flight altitude, insolation, number of GCPs, geometry of GCPs etc.. A general rule of thumb is that the ground control should be two to three times as accurate as the final product. If you are not repeating shots on the same control points, then averaging them, it's doubtful you'll get better than half a foot accuracy and it's quite likely you'll make a large blunder without realizing it as you don't have a background in error propagation.
Assume your surface will be 0.7ft higher than a perfect measurement and calculate the volume of your likely error (e.g. 0.7ft x 200ft x 500ft / 27 = 2,593yd3) Maybe reconsider this venture if the value of the stockpiles is anywhere close to your likely error.
Best practices are always the same. Use your drone and compare it to a physical survey. The drone will have the advantage of capturing more data than the physical survey, but the physical survey is data touching the ground, always a plus. How they match will provide you with invaluable data.
