-
Adjustment software – considering Star*Net
Hi,
I haven’t mined every thread in this discussion forum, so I apologize if I’m asking a duplicate question here.
The company I work for does a lot of bush surveying for resource industries – and I’ll note, we work in BC, Canada. In the last year, we’ve gotten larger and larger projects where we are, for example, surveying a remote access road 50km long with 20 bridge crossings throughout, where we need to fly in, set control, tie into other lidar/old survey data, do our designs, and get confidently set up for construction supervision/layout. This is a typical work flow for us, but on a much larger scale, with more companies involved and more on the line.
We survey with Leica total stations, Hemisphere RTK units, and Field Genius software. In the office, we use Civil 3D mostly, with RoadEng for road designs and a bit of lidar-processing support, Autodesk Recap just for processing LAZ/LAS files, and a basic version of Leica Infinity for when we need to investigate possible rod height errors etc.
My question is hard to sum up – I’m looking at Star*Net for a new addition in our work flow. I have spoken with MicroSurvey and done some online research, and having an ‘adjustment software’ seems really valuable to process our survey control and give us confidence in our results. I’ve heard that Star*Net is the most popular adjustment software in NA. Does anyone have feedback on –
– The application of Star*Net for our needs? Could I take my RTK data, some total station data, and combine it into a large topo map with solid survey control – with reports/numbers to verify our results? Currently, I stitch all survey data files together manually in Civil3D, often relying on best fits. This is time-consuming, has too much room for human error, and I don’t have a good way to quantify and display, at a glance, the confidence of my results.
– Other, better adjustment software programs? I spoke to a Leica rep to look into getting an expanded version of Leica Infinity, but was told it is a bit more finicky, and less popular, as they use a European-based adjustment method.
– Our survey process in general – do I not even NEED adjustment software? How do you ensure tight results when setting survey control and gathering lots of topo data? If you’re left with 3-4 shots on one base station/benchmark that was set the first day, then picked up on following days as checks, but they are all 1-5cm off horizontally/vertically, how do you decide where the ‘right’ location is and document this?
I really appreciate any feedback! I have a sneaking suspicion this is too broad of a question, but I need to start digging in somewhere. I also realize that our field surveying practices may need tweaking, but that is another post. We generally follow the Canada RTK surveying guide and use NRCan’s Precise Point Positioning website to post-process our RTK raw data.
Let me know if I can clarify! Thanks for reading.
IO
-
20 Replies
-
I have not used Starnet in many years. I mean 15 plus. But it is or was an awesome least squares program. When i used it our work flow was adjusting control and boundary marks gps and total station. I then used the adjusted control to re compute the topo shots etc. in the other softwares we had for going into sdrmap and autocad eagle point for map building. I don’t know if starnet you can make any maps but adjusting in least squares I think it’s a great software or it was. If i were not running Trimble equipment i would probably most definitely look hard at the starnet for sure. I would imagine it can take rtk total station data in from many manufacturers for least squares. That was the great thing about it back then it was not manufacturer dependent. Before I bought it i would confirm this and make sure your rtk vectors could be exported in a format that starnet could read along with total station data from field genius etc. Good luck.
-
As I sit here on Saturday morning it has been about 18 hours since I last used StarNet. I’m feeling withdrawal pains. With Kent McMillian inactive, I’m probably the number 1 StarNet advocate on this site. It has been a routine part of my workflow for 27 years. Back in the day I had several phone conversations with Ron Sawyer, the original developer of StarNet -including one time when he phoned me asking advise. Just so you know were I’m coming from. I run every project, even the smallest half day efforts, through StarNet. I can’t imagine doing anything really substantial without it. I can’t imagine why anyone would try.
I’ve also had some experience using Leica’s LS adjustment package. Also I little experimenting with Carlson’s SurvNet.
Could I take my RTK data, some total station data, and combine it into a large topo map with solid survey control – with reports/numbers to verify our results?
This is what StarNet does best – simultaneous adjustments of GPS vectors, RTK or otherwise, and total station data. It’s also really good at accepting and combining data from a very wide variety of sources (ie/data collector outputs form various manufacturers). It doesn’t map things, it just outputs reports and coordinate lists for import to the CAD of your choice. Leica’s software accepts Leica data and RINEX and nothing else.
Admittedly, Starnet’s approach to adjustments – on the grid – breaks down when dealing with very large areas. As in state/province wide areas. At least that is what I’ve been told as I’ve never had the opportunity to deal with such huge areas. So if you are regularly dealing with that sort of thing you might want go that direction. But for 99.999% of all work, StarNet is just fine. The largest project I’ve done was on the order of several miles and there was no problem.
Carlson’s SurvNet seemed to be a fair analogue to StarNet. I didn’t like the report format as much as I like StarNet’s but I could probably get used to it in time.
-
Add’l:
With StarNet you edit your input data in a text editor, save it, and then click RUN to recalculate the adjustment. If you forget to RUN, you can make all the changes to the data you want and nothing in the coordinate lists or report will change. That can be good, or that can be bad.
With Leica the program is reactive – it automatically saves the data and updates the adjustment with every keystroke. Except when, occasionally, it doesn’t. When it doesn’t there is no RUN button to force it to. That is, if you are clever enough to notice that it hasn’t. It’s been awhile, probably they have fixed this problem by now.
-
I’ll piggyback on Norman Oklahoma’s excellent synopsis and add that you can also add level data. MicroSurvey has a great online training course as-well.
-
Generally speaking, if you’re running a single-manufacturer solution, I would suggest going with that for post-processing. But you’re using Leica conventional gear and Hemisphere GNSS gear, and with FieldGenius, you’re already using MicroSurvey software. So StarNet would be a natural step, especially if you can export directly to native StarNet format from FieldGenius. I’m a Trimble fan myself, but I love StarNet.
I can’t imagine trying to stitch survey data together using Civil 3D, and I‘m with the other posters in that I adjust every project.
It sounds like your workflow is setting control and using it both for mapping and for construction. As a general rule, it is best practices to run and adjust control before anything else, and then use the adjusted values for subsequent work. That way you’ll know immediately if something is out of whack when you do your check observations throughout the mapping and staking process.
“…people will come to love their oppression, to adore the technologies that undo their capacities to think.” -Neil Postman -
“Admittedly, Starnet’s approach to adjustments – on the grid – breaks down when dealing with very large areas.”
As I recall, this only pertains to projects 100s of km in size.
-
One more thing about StarNet and Microsurvey – when I took this job at the city they didn’t have StarNet. Nobody here even knew what it was. But the guy who had been the city surveyor 20 years before me had recorded a control survey that had some tell-tale language and formatting in it that made me think that he had been using it. So I called up Ed in Kelowna and asked if the city owned a copy. He acknowledged that we did and that it had last been updated before Microsurvey acquired the property from Ron Sawyer. They fixed us up with a current copy and a new USB dongle for about a third of the price of a new license. They could have easily denied it and extracted an extra couple thousand out of us.
So kudos for honesty and fair play to Microsurvey.
-
I was with my new boss today and I have been working on a project for him troubleshooting and gathering all the raw data and performing a least squares on that project in TBC because we are a Trimble Shop. We were discussing Starnet as he is pushing for it as that is what he was use to. I told him as long as we are all Trimble I would just stick to TBC. He and I are about the only ones that believe in least squares. Most others do compass rule. Which is fine until you adjust one survey come back and come off of it adjust that one and keep doing that and when it gets so big and now all that error has been propagated and one side of the now larger project doesn’t fit the other and it just gets hairy. Also some route surveys crossing cause a headache as well. If I were not a Trimble Shop I imagine i would have already purchased starnet even if it had to be out of my own pocket. Least squares is just way to valuable in today’s world of working.
-
That’s pretty much the story at my workplace as well. We are required to use it for certain clients, so there’s no getting out of it there.
But there is also a vocal minority that seems to think that there’s something different about how StarNET adjusts data versus TBC. I’ve run many, many projects through both, and if the settings are the same, the results vary by maybe a few mm at most. Heck, I’ve written LSA programs and the results are pretty much the same. The fundamental equations don’t change.
Again, I like ’em both. StarNET has some things that I would like TBC to have, namely Pre-Analysis and the ability to use taped distances. Not that we do much taping at all any more…
But TBC is head & shoulders above StarNET when it comes to QC/QA. (Not that I would expect StarNET to be better, since it’s largely equipment-agnostic, and TBC is more limited in data intake.) It’s far, far easier to have a dynamic network to work with before adjustment, especially for moderate to large projects. TBC’s outlier testing is better as well.
“…people will come to love their oppression, to adore the technologies that undo their capacities to think.” -Neil Postman -
>something different about how StarNET adjusts data versus TBC
I don’t know what TBC does.
The difference between StarNet and some other software is that StarNet assumes the standard errors you give it are correct, so it uses Rayleigh statistics.
Some software, as illustrated in Wolf and Ghilani’s textbook, only uses your standard error values in proportion to each other, and estimates the overall scaling from the fit of the data, and uses F-distribution statistics.
This can result in different pass/fail results for the same data.
. -
I use both Star*Net and Geolab.
Star*Net for deformation surveys gives a lot of good info. I used to do everything in Geolab, but have transitioned to star*net for these types of surveys.
I do a lot of very large area GNSS surveys and for those I use Geolab. Some are over 20,000 sq miles, star*net does not handle networks that large correctly.
One thing about Star*Net that constantly aggravates me is that when you alt-tab to another program, say to get some coordinates to paste into star*net, and then alt-tab back, you have to click in the window, it won’t automatically go back to where you last were at. Almost all other programs that deal with text do this properly, including all of the programs that I write. I have complained to MicroSurvey but they have never changed this behavior. Simple to implement.
-
Thanks! I’ve confirmed that Starnet can take in data from different brands of gear, which is a good choice for us.
-
To all of those that have replied – Thank you! I will be taking this information to my employer. It supports my idea that we should be using this type of software in our work flow. The people I’ve spoken to at MicroSurvey have been great, and I may get to run some our own project data through a trial version of the program, which would be excellent.
-
Editorial Comment:
THIS is the sort of discussion I come to RPLS for.
Civil, informative, real life experience, practical.
Thank you for all the replies so far.I do have a math background that includes Least Squares adjustments, and a strong interest.
But I don’t have anything meaningful to add to this discussion. [Although if others want to discuss the (abstract) topic I would be more than willing to participate in a discussion.]
John Halleck. -
We’ve been using SALSA for a couple of years, it’s a pretty decent software package, gives you lots of information and you can export to Google Earth format for your client to view / interact with.
In Australia, we have AUSPOS, which allows processing up to 7 days of static observations, so we’ll start our work using RTK for coordinate references, with 3 static locations in good positions forming a triangle around the perimeter of the job with solar and battery, check the tripod daily for level and on point. We use an SX12 total station, a tilting level with micrometre and a Topcon GLS scanner, the observations from all are processed using SALSA. On the last job we engaged a subcontractor (Surveyor) with a NavVis VLX and set up a bunch of reflective targets for him, we used the SX12 and GLS to verify the VLX point cloud.
Since AUSPOS has already done some pretty comprehensive analysis, we fix the 3 static positions in SALSA and adjust to that. We use our RTK and total station coordinates in SALSA, but these aren’t fixed, the software adjusts their coordinates as it sees fit.
Regarding the tilting level, my dealer tells me they’ve got great digital levels now, but I’ve been caught in windy conditions where automatic and digital levels don’t work at all, and the wind can blow all week.
-
For those how are not familiar with SALSA, it is an open source program and library.
Your can get the manual (for the August 2022 version) for free at:
https://wwwext.arlut.utexas.edu/salsa/pdfs/SalsaUserManual_1150.pdfWritten at the University of Texas (at Austin). You can get a version for windows (for free),
and the sources are also free for under the GPL public license at:
https://wwwext.arlut.utexas.edu/salsa/
To run on your own machine.There is an active google group for questions, etc.
https://groups.google.com/g/salsa-least-squaresBottom line:
Good, pedantically correct Least Squares Adjustment.
Good documentation.
Free (GPL license), open source code.
Google group for problems.
John Halleck. -
Are there data format converters with SALSA? Or does the user have to manually enter raw data? If there are, which formats are supported? I’ve browsed the manual and don’t see a clear answer.
-
I don’t rightly know. My interest was in how it was handling the Adjustments. Mathematically it is a full implementation of proper least squares adjustments. For what I was looking at it for, it seems excellent.
My response was to the prior response that mentioned SALSA… mostly to point people at more information. The manual has a section on writing custom converters. I have no experience at doing so,
and I’m not a user of the program itself. Sorry. One of the prior responses was someone that has been a user of it… possibly he can enlighten us.
John Halleck. -
I don’t rightly know. My interest was in how it was handling the Adjustments. Mathematically it is a full implementation of proper least squares adjustment.
My response was to a response that mentioned mostly to point people at more information since it mentioned SALSA in passing.
John Halleck. -
There are built in importers, however my first experience was with manual entry, we wrote our own software that converts the observations from Trimble jobxml files and Topcon GLS2000 GLS2200 project directory structures read from the SD card.
Log in to reply.