Volume Survey UncertaintyPosted by Millshot259 on July 21, 2019 at 10:18 am
I am wondering if anyone could me with the topic of ‘volume uncertainties’.
Basically, I have computed a relative uncertainty between two DTM surfaces derived from photogrammetry of +/- 0.10m. Now using an area, I can calculate a ‘thickness’ error between the two surfaces to compute a volume uncertainty.
But I have been led to believe as the survey contains thousands of points generated by photogrammetry there is as many points that are +0.10m as there are -0.10m and maybe cancel each other out and approach 0.
Is this where the ‘uncertainty of a mean’ is used? I am confused can someone shed some light on this. I am using approx. 25000 points and a relative uncertainty of +/- 0.10m over and area of approx. 4000m².
- 21 Replies
- MemberJuly 21, 2019 at 7:04 pm
Volume calculations are at most an approximate value, to begin with.
The elevations are not exact and no two gatherings to collect the required data will match.
In a nutshell, it is a tool to achieve the best value possible for an amount of dirt or other material that will change in size.
The entire volume is a variable.
- MemberJuly 21, 2019 at 7:39 pm
Is that 4,000 meters by 4,000 meters (3,950?ñ acres) or 207.5 meters by 207.5 meters (10.6?ñ acres) for us non-metric old fuddy-duddies?
- MemberJuly 21, 2019 at 8:14 pm
The problem is; there is no guarantee, the uncertainties, the random errors, will cancel each other out. That’s why it’s variable; plus or minus.
The report provided, along with volume calculation, should provide the method used to create the surfaces, and the calculations that derived the quantity. Assuring that the reader understands that is a high and low number for the actual size.
This reminds me of a project @Roadhand was working on. A big highway job, somewhere deep in the heart of Texas. He was wondering about the best way to check the accuracy of his equipment. It seems that the Bean Counters figured out that an eighth of an inch of concrete, over the entire project, was worth a million dollars; they wanted someone to provide a quantity, to that accuracy, so that an equitable pay out would be provided. I told him to tell them that a sixteenth of an inch was worth $500,000, why weren’t they chasing that? There were some nicer, less condescending answers, that quickly pointed out; it would cost a lot more than 1 million dollars to provide that level of accuracy.
Accurate results can be achieved, but it’s a tedious, time consuming task and as we all know; time is money.I hope everyone has a great day; I know I will!
- MemberJuly 21, 2019 at 9:38 pm
right, so you think its fine for me to leave my estimate uncertainty as i have currently calculated +/0.1m?
- MemberJuly 22, 2019 at 12:45 am
Why are you trying to determine this? Is it a part of the deliverable to the client?
- MemberJuly 22, 2019 at 3:41 am
For a site 40m x 100m, I would just get out the total station and check it. I don’t think it is necessary to check it every time but enough to get a good feel for the actual accuracy you are getting.
The thing I have a hard time getting across is the error for each layer compared to the thickness of the volume. If i have 0.10m error in the original layer, then 0.10m error in the final layer, but the vertical difference between the two layers is 10 meters, Then your error estimate would be 2% which is pretty good.
0.2 meter total error when the total average depth of the volume is 1 meter, then your error estimate would be 20%.
- MemberJuly 22, 2019 at 4:19 am
Well, 40 meters by 100 meters is about 328 feet by 131 feet, say 43,060± square feet. Not quite an acre.
0.1 meter is about 0.328 feet (4 inches), so ±0.33 feet over that almost-an-acre is 12,920± cubic feet.
That’s 478 cubic yards, the usual measure of volume here, or about 50 truckloads.
That close enough for ya?
- MemberJuly 22, 2019 at 2:33 pm
I’m with JaRo on this one.
SOP for all our photo topo jobs is to check points.
Unless this isn’t a job where something is getting built.
- MemberJuly 22, 2019 at 4:08 pm
To get an accurate check, you need to measure the site with a different procedure and different equipment (our spec’s call for equipment 3x more accurate than published accuracy), then calculate the volume with different software and COMPARE results. You will be surprised. I’d expect your quantities to vary by ?ñ4% or more.
At least run the field data through a second software. They all calc volumes slightly different.
- MemberJuly 22, 2019 at 4:31 pm
I think you should look at your scope of work and reassure yourself that the method used to collect the data and make the calculations meets the expectations of the client.
If it doesn’t, maybe you need to rethink your method.
DougieI hope everyone has a great day; I know I will!
- MemberJuly 22, 2019 at 5:49 pm
Agree with Lee. The only way to get a quantifiable standard error are independent measurements of a higher accuracy, per NSSDA specs.
However, methodology for quantifying the error aside, for the final volume you will be computing the combined error for a sum of quantities.
If there is a 0.1m vertical standard error in two 4000m2 surfaces, that is a 400m3 standard error in volume for each. The difference between them is a sum (subtract one from the other).
The combined error of a sum is the square root of the sum of the squares of the standard errors, which yields approximately 566m3.“…people will come to love their oppression, to adore the technologies that undo their capacities to think.” -Neil Postman
- MemberJuly 22, 2019 at 9:28 pm
Yeh so i have RTK obs on the surface which have a higher vertical accuracy which i will use for some checks.
Maybe I should of explained the situation further. I used photogrammetry to compute a stockpile survey. The one
survey was used to determine the base surface and the top surface therefore i have determine the two surfaces to have the same
uncertainty. The uncertainty in each surface is made up of the GCP uncertainty and the photo – software processing uncertainty. combined = +/-0.071m.
The relative uncertainty between the two surfaces as you have said is the square root of the sum of the squares which computes at +/-0.1m.
so with that said my question really was if people thought it would be okay to use the relative uncertainty between the two surfaces to compute a total volume uncertainty. which would look like
volume uncertainty =4000m2 * 0.1 = 400m3
as a percentage 400m3 / 12500m3(total volume) = 3.2%.
or is there another process such a uncertainty of a mean which is used where the mean is the relative uncertainty? as its a +/0.1 will there be points that will cancel each other out and approach 0 and if this was a problem is there a equation to compute the real average uncertainty??
- MemberJuly 23, 2019 at 4:58 pm
Don’t worry about it. No matter what you do some guy named Jed with one of these [attach]4898 is going to call you wrong based on his truck count. 🙂 That was typically my experience. Good Day, Jp
- MemberJuly 23, 2019 at 5:50 pm
Did you account for the dirt that left the site on the worker’s boots?
That’s what my first boss would ask then apply an arbitrary 20% shrinkage factor to the fill.
- MemberJuly 24, 2019 at 12:32 pm
Are you measuring dirt or enriched Uranium? Dirt purveyors in Florida sell it by the truckload which is supposed to weigh a certain amount per load. The trucks are weighed prior to leaving the supply source and if they are too light they just spray water on it until it reaches the desired poundage. They know every trick in the book on how to screw contractors needing mass quantities of fill.
With dirt quantities if you are within 20% yer good. ????
- MemberJuly 24, 2019 at 1:13 pm
Now another guy is doing it, but I used to have to do an inventory of the different rock and sand stockpiles at 3 asphalt plants monthly. Would measure, use the factors they came up with to convert to tons, then give the numbers to the bean counters. Usually there was something they didnt agree with, so I just started asking them what they wanted the numbrrs to be. Some piles were uncompacted, some were driven on, some were just a bunch of dumped loads next to each other. At one time we were going to do a check where they were going to unload a sand barge onto trucks, run them across the scales, the put it all in a separate pile and measure to see how it compared. After I measured it, turned out they had lost count of which trucks dumped in the pile.
- MemberJuly 24, 2019 at 1:30 pm
I kept shooting the man-sand at the hotmix plant and was being told there was supposed to be quite a bit more than I shot. One day I got a shovel and started digging. I gave up and got the guy with a backhoe to come dig a hole. They had been driving on 3 foot of man-sand. I had no idea where the bottom was, I was just doing like the original poster and using the toe of slope for the bottom. I have tried since then to get them to let me shoot the bottom before the first pile is delivered but it never happens.
Log in to reply.