1997-98 Burnett County Yellow Lake/Gandy Dancer Level Project

Introduction

Burnett County has been involved in an ongoing project to upgrade the vertical control network in the county. The 1997 plan was to continue running high quality level lines from Siren west to the state line with Minnesota. In the Spring of 1997, a problem arose with the elevations that were used to maintain the level of Yellow Lake. It was decided to redirect the funds set aside for the 1997 plan and do some work help resolve the questions around Yellow Lake. A cooperative effort between Burnett County and Northwestern Wisconsin Electric Company (NWE) was undertaken to upgrade the quality of the vertical control in the area of the Danbury Dam and Yellow Lake. This upgraded control would serve to evaluate the operational procedures and would be used if a redefinition of the operating procedures is necessary.

Burnett County, represented by Kathy Swingle, County Surveyor, routinely contracts with surveyors in private practice for fieldwork and technical services. For this project, Norton Surveying, owned by Michael E. Norton, Spooner was hired to perform the fieldwork portion of the project. North Country Surveying, Inc. of Siren was hired to perform the calculations, compile the most complete level network possible and perform analysis of the resulting network.

This report was prepared by Douglas R. Crane of North Country Surveying, Inc. and will discuss the results of a least squares adjustment of the level network compiled for this project.
 

Project Planning

The project was planned by Douglas R. Crane and Michael E. Norton. All of the previous Burnett County Projects since 1991 were run using the same procedures. These could be described as using 2nd Order specifications with exception that an invar level rod was not used. Therefore, the work would be classified as 3rd Order. Level runs were run in chains. In general 6 to 10 setups were run out to a temporary benchmark and then looped back using the same turning points. In the computations, we actually input the two delta elevations between turning points as part of the least squares solution. We decided to continue with the same procedures for new work on this project based on our previous success and the high quality attained in the previous projects.

Our objectives were to:

The COE project was done in 1986-89. This was done for the purposes of performing a flood study to determine the regional flood elevations for numerous lakes and rivers around the county. We have the original field notes for this work. In 1991, Burnett County constructed high quality benchmarks to establish a more permanent vertical reference system to document the COE flood study elevations. Norton Surveying performed the level work to transfer elevations from the COE temporary benchmarks to the county benchmarks.

In 1993, Burnett County designed and performed a high precision geodetic control network. The monument locations were selected to be suitable for GPS (Global Positioning System) observations and were located to provide an approximate 3 mile spacing over the county. When possible, existing benchmarks were used as stations in this network or stations were located near benchmarks so that the ellipsoid heights generated in a GPS survey could be correlated to true elevations or geoidal heights. In 1994, level work was performed to make these connections to nearby stations.

We had a previous county project that ran levels from USGS Benchmark TT35B near the Minnesota state line on STH77 east to GPS Station Otten, continuing east into Danbury and south to TT32B and GPS Station Overlook South. This work was done in 1996 as part of the county project and consists of 12.6 miles of levels counting the chain loop effect used. Steve Geiger, previously employed by Burnett County, specified that this work be done and it is not clear why. I suspect that there must have been some question about the elevation on GPS Station Otten based on TT35B. The published number on TT35B does not fit the present monument.

We had previous county projects from 1994--96 that ran levels from the intersection of STH70 with the East Line of Burnett County and ran west into Siren. This project established the elevation of National Geodetic Survey (NGS) station Heidi at the Burnett County Airport. This project was very tight and yielded a standard error of +0.0075 foot on Station Heidi. This station was chosen to be the closing point for the new project. This network was used to control and verify another level network around Clam Lake where another problem existed regarding operation of the dam and the elevation to be maintained on the lake.

For the new project, we decided to recover temporary benches in Danbury and run south to the Danbury dam. We planned a run from the temporary benchmark at Gandy Dancer Trail and Hayden Lake Road to close on Station Heidi. This run was 11.8 miles one way or 23.6 miles looped. Side runs were done to connect to remnants of TT31B and temporary benches near the intersection of CTH U and N. Lake Road in Sec. 24-40-17. This 1991 run had previously connected into BM437-E at the Ulrich Bridge between Big and Little Yellow Lakes and was used to establish the elevation of BCBM#6 near the Anderson Bridge near the outlet of Little Yellow Lake.
 

Field Work

Preliminary reconnaissance work to recover benchmarks and temporary benchmarks was performed by Mike Norton prior to the start of observations. The location for new benchmark monuments was determined by Mike Norton. NWE constructed the new monuments using specifications provided by Burnett County.

The fieldwork, as planned above, was completed between October 9, 1997 and December 8, 1997. Norton used his Pentax AL-MC2 level for all of this new work.

The field notes for this work are filed with the Burnett County Land Information Office. Descriptions of all the benchmarks are in these notes.

I have included his two page final report to Burnett County in the Appendix.
 

Computations And Compilation of Network

The field notes for Norton's work both for previous projects and this project were compiled into a least squares level network adjustment. We used STAR*LEV by Starplus Software. Based on a statistical evaluation of the closures of Norton's work, the standard error assigned is 0.0032 feet per setup. For work of lower quality (having a higher standard error per setup), an equivalent setup number was determined using proper statistical procedures.

The objective of assigning realistic standard errors to the measurements is to weight the data with it's strength relative to data from another source. We want the highest quality data to have the strongest influence on the final answer. We want weaker data to have less influence.

Benchmark TT31B near the intersection of STH 35 and CTH U was destroyed in highway construction in 1985. Fortunately WisDOT contacted the county surveyor - Gerald Wagner - prior to destruction and two temporary benches were set to preserve the elevation. The COE work is tied to one of these temporary points and is also tied to BM437-E on Ulrich Bridge at the Yellow Lake Narrows.

The COE data from field notes around Yellow Lake was put into the adjustment using a standard error of +0.005 feet per setup. Water transfers around Yellow Lake were used in their procedure and we assigned a standard error of + 0.015 feet per water transfer setup.

NWE provided information that they had in their records. In 1954, Herman Hagestad, consulting engineer from River Falls, did a considerable amount of level work for NWE. It appears that prior to this time, the Wisconsin Public Service Commission (PSC), used an assumed datum for elevation work on the Yellow Lake area. Hagestad shot in some of these benches and gages around the dam and lake area. I have included the information from Hagestad in the Appendix. It appears that he had a different set of numbers on the USGS benchmarks in 1954 than is published today. I have compared his numbers to the presently published numbers and the numbers he used are 0.2 to 0.3 feet lower. This probably accounts for the published value of BM437-D at the dam being different than our measurements and calculations. It appears that Hagestad's work at the dam was based on TT33B which is destroyed. I have entered some delta elevations from Hagestad's information in the adjustment. Due to the fact that we do not have any field notes I have assigned standard errors based on distance from TT33B and an equivalent for a dumpy level.

NWE also provided field notes for a level run performed by Clarence Wagman of the NWE staff. This included a level run from TT34B down to the dam and a connection to BM437-D. I used the number of setups to key the assigned standard errors based on an instrument error of +0.005 ft. per setup.

I contacted John Haack of WisDNR to see if any original field data was available on any of the PSC benchmarks in this project. Apparently there is nothing on 437-E. I received fieldnotes that showed measurements at the dam between 437-C and 437-D. These observations were entered in the adjustment.

The published value on TT35B did not fit the measurements. I discussed this with Dixon Hoyle, Minnesota Advisor for the National Geodetic Survey. He checked and found that the Minnesota Highway Department had recently run 2nd Order levels from I35 at Hinckley east to TT35B. They also found that it did not fit the published value and renamed the station TT35B-RESET using their measurements. Dixon provided me with the elevation from their work.

All observations in the network were entered as delta (or difference in) elevations between points in the network. The software then allows us to run a series of adjustments holding any combination of benchmarks fixed (held to the published value) or free (to be computed based on the measurements). This provides tremendously powerful analysis tools in determining if a problem exists with a benchmark.

Approximately 18 miles of levels were run from TT35B-RESET to Heidi in one direction. The total distanced leveled is doubled due to the chained procedure that was used for new work in this project. To analyze this network, I have run the adjustment three different ways minimally constrained. Each one holds a different benchmark fixed. The adjustments are named as follows and the table shows the benchmark held fixed for each adjustment.

File Name BM Held Fixed
GANDYMCH Heidi
GANDYMCU TT31B
GANDYMCM TT35B-RESET
All benchmarks that are not held fixed in an adjustment are free and not constrained to their published value. A comparison is made between the computed value, resulting from the network, and the published value in a table (Adjustment Results) in the Appendix, for each of the three minimally constrained adjustments. For statistical analysis of the networks, the minimally constrained network passes the chi-square test and runs with an error factor of 1.03. This is an indication that the network checks well with itself and that the weighting scheme is realistic.

Basically, there is 0.15 foot of error between TT35B-RESET and Heidi. Considering the very large area covered, I feel that this is extremely precise. Examine the Adjustment Results table to see the variances from the published values in the various adjustments.

The only significant difference is at the dam. I believe that this is probably due to Hagestad using preliminary benchmark elevations from an earlier adjustment of the USGS benchmarks. When the final USGS adjustment was performed and new numbers were published, the change was not reflected in the benchmarks at the dam.
 

Contrained Adjustment

Based on examination of the results of the various minimally constrained adjustments, I did not reject any of the published values of the USGS benchmarks except TT35B, where I used the MNDOT value. I assigned standard errors to the benchmark values based on their published quality. This varied from +0.01 to +0.03 foot for the six benchmarks held partially constrained in the network. All other benchmarks in the network are based on measurements made off of the USGS benchmarks and therefore I left them free to float with the final solution. The Adjustment Results table in the Appendix shows the Standard Errors assigned to the USGS benchmarks along with the changes of all the benchmarks in the network from their published values.

The error factor of the constrained network rises from 1.03 to 1.05 and it still passes the chi-square test. This is an indication that there is no systematic error and that we fit the control points quite well.

The results are based on the NGVD29 published values.
 

Results And Diskette Contents

I have copied all the input and output files for the three minimally constrained adjustments and the constrained adjustment onto a diskette and enclosed it with this report. The files are ASCII text files and can be read with any text editor on a DOS machine. The adjustments are labeled with names as described above. The constrained adjustment is named "GANDY". The file extensions are defined below:
File Extension Explanation
*.dat Input Data File
*.prj Project Default File for STAR*LEV
*.lst Output Listing File - Shows Residuals and Error Propagation
*.pts Output Points Files - Shows Final Elevations for Each Point in the Network
I have included a print out of part of the constrained points file in the Appendix. This shows the major benchmarks and other points of interest. I have deleted the turning points from this print-out.
 

Conclusions

This project has been a large undertaking that has shown that the original USGS benchmarks in this part of Burnett County are very high quality. However in the entire run of level work discussed in this project only one USGS benchmark (TT32B) remains intact. All the others have been destroyed since the1950's. There are numberous others that have been destroyed or are missing that are not discussed in this report. Fortunately, we have been able to recover some records showing meaurements into these missing benchmarks and have been able to make connections from the network compiled for this project to some of these missing benchmarks. This has allowed us to use the benchmarks even though they are presently destroyed. This project has provided numerous high quality new benchmarks that can be used in the future.

It is important to make a serious attempt to preserve the remaining benchmarks and new benchmarks in order to preserve the vertical datum of Burnett County.

The resultant elevations derived from this project can be relied on as the basis for evaluation of the operating procedures of the Danbury dam and can be used to define a revision of those procedures if it is deemed necessary.

BM437-E at the bridge at the Narrows between Big and Little Yellow Lake did not change much from its published value no matter what adjustment you look at. In the final adjustment it changes only 0.013 feet from the published value, which is basically insignificant.

At the dam, BM437-C and D have changed significantly from their published values. This change varies from +0.17 to +0.35 feet in the minimally constrained adjustments. In the final adjustment, the change is +0.26 and +0.24 feet from the published values.

The elevations established on Burnett County Geodetic Control (GPS) Stations will be used to refine our abilities to use GPS equipment to perform high quality elevation work in the county. This will take some additional work to analyze the effect of this work. I believe that by combining this work with previous elevation work in the county, we will be able to use GPS observations to yield true elevations that are somewhat better than 0.1 foot. The value of this is that long distances can be covered in a single observation. This will significantly cut the cost of producing elevations for consumers in the future.

 

Final Adjusted Elevations

1997-98 Burnett County
Yellow Lake/Gandy Dancer
Level Project
Final Adjusted Elevations
1997-98 Burnett County
Yellow Lake/Gandy Dancer
Level Project