My previous newsletter highlighted a National Geodetic Survey (NGS) webinar held on April 25, 2025, titled “Design of Networks Using NOS NGS 92,” given by Dave Zenk, NGS northern plains regional advisory.
<p><em>[Authors note: Dave Zenk told me that he is retiring from the National Geodetic Survey on May 31, 2025. Dave’s presence will be deeply missed. His dedication and spirit have left a lasting impact on NGS’s products and services. I hope his retirement is filled with joy, relaxation, and new adventures.] </em></p> <p>In addition to Dave Zenk’s retirement, several other NGS Regional Geodetic Advisers have retired or left NGS employment over the past several months. <a href="https://geodesy.noaa.gov/ADVISORS/index.shtml" target="_blank" rel="noopener">Click here</a> for a list of the current advisors, along with the names of interim contacts handling inquiries for those advisors who have retired or departed from government service.</p> <p>As previously mentioned, Dave showed a well-presented outline of the tables that users need to be familiar with when using OPUS Projects to process and submit GNSS projects to NGS for publication. <em>It should be noted that users submitting data to NGS must follow the guidelines outlined in <a href="https://geodesy.noaa.gov/library/pdfs/NOAA_TM_NOS_NGS_0092.pdf" target="_blank" rel="noopener">NOS NGS 92</a>.</em></p> <p>I found the webinar to be very informative, and I would encourage all users of OPUS Projects to download the <a href="https://geodesy.noaa.gov/web/science_edu/webinar_series/2025-webinars.shtml" target="_blank" rel="noopener">presentation</a>. During the webinar, Zenk briefly mentioned three items that I believe deserve more explanation for anyone using OPUS Project. This newsletter will address the following topics in more detail:</p> <ul class="wp-block-list"> <li>The mark’s classification — primary, secondary, and local — will not be included on the NGS datasheet, but the local and network accuracy from the project will be provided on the datasheet. What does this mean to someone who’s using the mark in their project?</li> <li>OPUS Project uses the F-statistic test to determine if the appropriate constraints were imposed during the horizontally and vertically constrained adjustments. Why does OPUS Project use this statistic?</li> <li>The Constraint Ratio (CR) test, computed by OPUS Projects, provides a method for identifying which coordinates should be constrained and which should not be considered for constraints in the final horizontally and vertically constrained adjustments. What’s the best way to use this table?</li> </ul> <p>First, the presentation discussed the tables that described the procedures for establishing three different mark classifications — primary, secondary and local. It also mentioned that the classification will not be included on the NGS datasheet but the local and network accuracy from the project will be provided on the datasheet. See the image below.</p> <figure class="wp-block-image size-large is-resized"><img fetchpriority="high" decoding="async" width="1024" height="789" src="https://coordinates.net/wp-content/uploads/2025/06/Screenshot-2025-06-02-at-3.02.27 PM-1024x789-1.png" alt="Photo:" class="wp-image-110338" style="width:652px;height:auto" srcset="https://coordinates.net/wp-content/uploads/2025/06/Screenshot-2025-06-02-at-3.02.27 PM-1024x789-1.png 1024w, https://www.gpsworld.com/wp-content/uploads/2025/06/Screenshot-2025-06-02-at-3.02.27 PM-300x231.png 300w, https://www.gpsworld.com/wp-content/uploads/2025/06/Screenshot-2025-06-02-at-3.02.27 PM-245x189.png 245w, https://www.gpsworld.com/wp-content/uploads/2025/06/Screenshot-2025-06-02-at-3.02.27 PM-768x591.png 768w, https://www.gpsworld.com/wp-content/uploads/2025/06/Screenshot-2025-06-02-at-3.02.27 PM-1536x1183.png 1536w, https://www.gpsworld.com/wp-content/uploads/2025/06/Screenshot-2025-06-02-at-3.02.27 PM.png 2036w" sizes="(max-width: 1024px) 100vw, 1024px" /><figcaption class="wp-element-caption">Photo: NGS website </figcaption></figure> <p> What does this mean to someone who’s using the mark in their project? Since the NGS data sheet will provide the network and local accuracy from the project, users can determine if the accuracy value of the mark meets the requirements of their project. In my opinion, the network and local accuracy from the project provide a better indication and understanding of the level of trust of the published coordinate.</p> <p>As previously mentioned, anyone submitting a GNSS project to NGS for publication must adhere to the NOS NGS 92 guidelines. During the presentation, Zenk provided several examples that depicted correct network designs. I would encourage everyone to download the <a href="https://geodesy.noaa.gov/library/pdfs/NOAA_TM_NOS_NGS_0092.pdf" target="_blank" rel="noopener">NOS NGS 92</a> document and <a href="https://geodesy.noaa.gov/web/science_edu/webinar_series/2025-webinars.shtml" target="_blank" rel="noopener">Zenk’s presentation</a> to gain an understanding of the classifications and the network design requirements to meet a particular classification.</p> <figure class="wp-block-image size-full is-resized"><img decoding="async" width="775" height="549" src="https://coordinates.net/wp-content/uploads/2025/06/Screenshot-2025-06-02-at-4.14.28 PM.png" alt="Photo: NGS website " class="wp-image-110341" style="width:714px;height:auto" srcset="https://coordinates.net/wp-content/uploads/2025/06/Screenshot-2025-06-02-at-4.14.28 PM.png 775w, https://www.gpsworld.com/wp-content/uploads/2025/06/Screenshot-2025-06-02-at-4.14.28 PM-300x213.png 300w, https://www.gpsworld.com/wp-content/uploads/2025/06/Screenshot-2025-06-02-at-4.14.28 PM-245x174.png 245w, https://www.gpsworld.com/wp-content/uploads/2025/06/Screenshot-2025-06-02-at-4.14.28 PM-768x544.png 768w" sizes="(max-width: 775px) 100vw, 775px" /><figcaption class="wp-element-caption">Adhere to NOS NGS 92 guidelines (Photo: NGS website)</figcaption></figure> <p>Anyone who submits an OPUS Project to NGS for publication knows that the constrained adjustments must meet the requirements of the F-statistic test. So, what is this test, and why does OPUS Project require this statistic? Essentially, it is a method of verifying whether the appropriate constraints were applied during the horizontally and vertically constrained adjustments. The F-test evaluates the ratio of two variances; that is,</p> <figure class="wp-block-image size-full"><img decoding="async" width="582" height="93" src="https://coordinates.net/wp-content/uploads/2025/06/Screenshot-2025-06-02-at-4.16.22 PM.png" alt="Photo:" class="wp-image-110342" srcset="https://coordinates.net/wp-content/uploads/2025/06/Screenshot-2025-06-02-at-4.16.22 PM.png 582w, https://www.gpsworld.com/wp-content/uploads/2025/06/Screenshot-2025-06-02-at-4.16.22 PM-300x48.png 300w, https://www.gpsworld.com/wp-content/uploads/2025/06/Screenshot-2025-06-02-at-4.16.22 PM-245x39.png 245w" sizes="(max-width: 582px) 100vw, 582px" /></figure> <p>The F-test checks whether this ratio is significantly different from 1, which would suggest the models have significantly different fits to the data. The result is compared against the <a href="https://www.itl.nist.gov/div898/handbook/eda/section3/eda3673.htm" target="_blank" rel="noopener"><strong>critical value</strong> from the <strong>F-distribution</strong></a> based on the degrees of freedom from the constrained adjustment and the degrees of freedom from the minimally constrained adjustment, and a chosen <strong>significance level</strong> alpha (e.g., 0.01). NGS OPUS Project uses an alpha level of 0.01% or 99% confidence level.</p> <figure class="wp-block-image size-full is-resized"><img loading="lazy" decoding="async" width="548" height="692" src="https://coordinates.net/wp-content/uploads/2025/06/Screenshot-2025-06-02-at-4.19.02 PM.png" alt="Photo:" class="wp-image-110343" style="width:509px;height:auto" srcset="https://coordinates.net/wp-content/uploads/2025/06/Screenshot-2025-06-02-at-4.19.02 PM.png 548w, https://www.gpsworld.com/wp-content/uploads/2025/06/Screenshot-2025-06-02-at-4.19.02 PM-238x300.png 238w, https://www.gpsworld.com/wp-content/uploads/2025/06/Screenshot-2025-06-02-at-4.19.02 PM-166x210.png 166w" sizes="auto, (max-width: 548px) 100vw, 548px" /><figcaption class="wp-element-caption">Definition of the F-test (Photo: Wikipedia)<br><br> <br>The following is an excerpt on the F-test from <a href="https://geodesy.noaa.gov/OPUS-Projects/user_guide/user_guide.html" target="_blank" rel="noopener">NGS’s Online OPUS Project Guide</a> – (<a href="https://geodesy.noaa.gov/OPUS-Projects/user_guide/section12.7.html#analyzing-the-horizontal-constrained-adjustment" target="_blank" rel="noopener">Section 12.7.3.2. Analyzing the Horizontal Constrained Adjustment):</a> </figcaption></figure> <p><em>Once the adjustment has been deemed acceptable i.e. all shifts and residuals are reasonable, the F-test should pass. <strong>The F-test is a statistical test that helps determine if the variance (variance of unit weight) from a fully constrained adjustment is significantly different from the variance (variance of unit weight) of a minimally constrained adjustment. </strong>The variance of unit weight is a critical statistic and should be looked at carefully when evaluating adjustment results. If the fully constrained adjustment fits well with all selected control (the constraints), the value of the variance of unit weight should be close to 1.0. The F-test is performed using a 99% confidence level.</em></p> <p>So, if the constrained adjustment statistics differ significantly from the minimally constrained adjustment, then there could be an issue with the constraints. Of course, this is assuming that the minimally constrained adjustment variance of unit weight indicates that all data outliers have been eliminated. So, why are constraints important?</p> <p>OPUS Project first calculates GNSS coordinates in a minimally (free) network adjustment, which defines relative positions but not their absolute placement in space. Without constraints the entire network can float and/or rotate.</p> <p>Constraints are important in GNSS network adjustments because they:</p> <ul class="wp-block-list"> <li>Anchor the network in a geodetic datum; in this case, NAD 83 (2011), epoch 2010.0.</li> <li>Ensure a unique and stable solution that reflects the physical world.</li> <li>Make the network useful for engineering, mapping, and scientific purposes.</li> </ul> <p>Control point coordinates (from previous surveys or known datums) often have inherent errors or uncertainty. Constraining coordinates exactly assumes zero error, which is rarely true. Weighted constraints let you assign a realistic level of trust to known published coordinates by using error estimates. OPUS Project applies weighted constraints based on input error estimates (OPUS Project denotes these as sigmas of the coordinates), which allow for minor deviations in the constrained coordinates. The weighted constraint methodology provides flexibility to network adjustments by recognizing that published coordinates have some uncertainties and allows constraints to take on small corrections leading to more accurate and consistent network solutions. Although, it should be noted that the adjusted coordinates of the constraints from the final horizontally constrained adjustment are not updated in the NGS database even though there are minor deviations to their final adjusted values.</p> <p>After performing a least squares adjustment using the weighted constraints, the F-test helps to ensure that the appropriate coordinates are constrained with the appropriate error estimates. The following is an excerpt from the <a href="(https:/geodesy.noaa.gov/OPUS-Projects/user_guide/user_guide.html)">OPUS Project Online User Guide</a> <a>– (</a><a href="https://geodesy.noaa.gov/OPUS-Projects/user_guide/section12.7.html#analyzing-the-horizontal-constrained-adjustment" target="_blank" rel="noopener">Section 12.7.3.2. Analyzing the Horizontal Constrained Adjustment):</a></p> <p><em>If the F-test fails, it is due either to the errors (sigmas) of the constraints being overly optimistic (too small) or the constrained coordinates not agreeing with the observations (causing excessively large shifts of the constrained coordinates). <strong>Failure of the F-test does not automatically mean the constrained adjustment is bad. It is a flag that indicates there may be a problem with the constraints, and that they should be investigated. </strong>In addition, the F-test assumes of a normal (“bell-shaped”) probability distribution of the residuals. Networks with a distribution that is significantly non-normal may fail for that reason, even when a constrained adjustment is acceptable.</em></p> <p>if your adjustment fails the F-test, what do you do? How do you determine which constraint or constraints should be unconstrained? OPUS Project provides some information about the constraints that can be helpful in determining a bad constraint. The CR test, computed by OPUS Projects, provides a method for identifying which coordinates should be constrained and which should not be considered for constraints in the final horizontally and vertically constrained adjustments. What’s the best way to use this table? The box titled “Constraint Ratio” from <a href="https://geodesy.noaa.gov/OPUS-Projects/user_guide/section12.7.html#analyzing-the-horizontal-constrained-adjustment" target="_blank" rel="noopener">NGS’s Online OPUS Project User Guide</a> — (<a href="https://geodesy.noaa.gov/OPUS-Projects/user_guide/section12.7.html#analyzing-the-horizontal-constrained-adjustment" target="_blank" rel="noopener">Section 12.7.3.2. Analyzing the Horizontal Constrained Adjustment)</a> provides a good explanation with an example of using the constraint ratio table (12.7.3.2. Analyzing the Horizontal Constrained Adjustment). Basically, this statistic highlights coordinate shifts that are significantly larger than expected based on the sigma provided by the user. That is, coordinates that have a very small sigma should not be expected to change as much as coordinates with a very large sigma. The CR value is compared to a critical value of 3.0, which corresponds to a t-statistic at the 99% confidence level. Therefore, any constraint ratios greater than three should be investigated and are candidates to be unconstrained (see the box titled “Constraint Ratio”).</p> <hr class="wp-block-separator has-alpha-channel-opacity"/> <h3 class="wp-block-heading">Constraint Ratio</h3> <p>If the F-test fails, it is possible that some constraints need to be freed up. It might be the case where some of the shifts are too large. The CR test provides a way of identifying where the bad shift might be. The CR is essentially a Students T Test, with the absolute value of the shift between the adjusted, constrained coordinates and the published coordinates, divided by the sigma (σ, or standard deviation) used to constrain the station. It is computed for each component (north, east, and height):</p> <figure class="wp-block-image size-full is-resized"><img loading="lazy" decoding="async" width="470" height="58" src="https://coordinates.net/wp-content/uploads/2025/06/Screenshot-2025-06-02-at-4.22.59 PM.png" alt="Photo:" class="wp-image-110344" style="width:705px;height:auto" srcset="https://coordinates.net/wp-content/uploads/2025/06/Screenshot-2025-06-02-at-4.22.59 PM.png 470w, https://www.gpsworld.com/wp-content/uploads/2025/06/Screenshot-2025-06-02-at-4.22.59 PM-300x37.png 300w, https://www.gpsworld.com/wp-content/uploads/2025/06/Screenshot-2025-06-02-at-4.22.59 PM-245x30.png 245w" sizes="auto, (max-width: 470px) 100vw, 470px" /></figure> <p>OP provides the CR for all marks in the final table in the output summary given in the body of the email or in the Processing Report (.txt), as shown below in <a href="https://www.ngs.noaa.gov/OPUS-Projects/user_guide/section12.7.html#figure12-21-jpg" target="_blank" rel="noopener">Fig. 12.21</a>.</p> <figure class="wp-block-image size-full"><img loading="lazy" decoding="async" width="653" height="170" src="https://coordinates.net/wp-content/uploads/2025/06/Screenshot-2025-06-02-at-4.23.48 PM.png" alt="Fig. 12.21 Constraint Ratio Test as seen in the Processing Report of the Horizontal Constrained Adjustment. (Photo: NGS) " class="wp-image-110345" srcset="https://coordinates.net/wp-content/uploads/2025/06/Screenshot-2025-06-02-at-4.23.48 PM.png 653w, https://www.gpsworld.com/wp-content/uploads/2025/06/Screenshot-2025-06-02-at-4.23.48 PM-300x78.png 300w, https://www.gpsworld.com/wp-content/uploads/2025/06/Screenshot-2025-06-02-at-4.23.48 PM-245x64.png 245w" sizes="auto, (max-width: 653px) 100vw, 653px" /><figcaption class="wp-element-caption">Fig. 12.21 Constraint Ratio Test as seen in the Processing Report of the Horizontal Constrained Adjustment. (Photo: NGS) </figcaption></figure> <p>Computed CRs are compared to the critical value or 3.0, corresponding to a T-statistic at a 99% confidence level. If the value of CR is greater than 3.0 for any of the three components, that indicates that there may be a problem with the constrained station.</p> <hr class="wp-block-separator has-alpha-channel-opacity"/> <p>I find these statistics very helpful when determining which coordinates should be constrained in the final adjustments. I hear that some users select all possible constraints and then start releasing marks based on the CR table. That certainly is one way of doing it but could be time-consuming and confusing. That said, the first thing I do is compare the minimally constrained adjusted coordinates to the published coordinates to determine if there are any obvious outliers. This has been helpful to me in large GNSS projects located in subsidence regions such as the Harris-Galveston, Texas, region of the United States.</p> <h3 class="wp-block-heading">One final note on OPUS Project</h3> <p> On May 22, 2025, NGS issued a notice to users, announcing the implementation of the International Terrestrial Reference Frame 2020 (ITRF2020). The announcement provided the following information addressed to all Active OPUS Project Users.</p> <p><em>Active OPUS-Projects Users,</em></p> <p><em>In early June, NGS will implement the new International Terrestrial Reference Frame 2020 (ITRF2020) and IGS20 realizations in the U.S. National Spatial Reference System (NSRS) in order to maintain consistency with the International Earth Rotation and Reference System Service (IERS) and the International GNSS Service (IGS) reference frames. This results in updated North American Datum 1983 (NAD 83) coordinates for stations in the NOAA CORS Network (NCN), kept at epoch 2010.0. This update is called <a>the Multi-Year CORS Solution 3 (MYCS3), </a>and it follows NGS’s MYCS2 effort from 2018.</em></p> <p><em><strong>OPUS-Projects users with active projects are advised that open projects will need to be reprocessed from the beginning in ITRF2020. </strong></em></p> <p><em>If projects are close to completion, users have the option of submitting them to NGS before the transition using the currently published NAD83(2011/MA11/PA11) coordinates transformed from ITRF2014. <strong>The deadline for submissions is June 6, 2025 </strong>for those wishing to take this route.</em></p> <p><em>If you have questions, please contact </em><a href="mailto:ngs.opus@noaa.gov" target="_blank" rel="noreferrer noopener"><em>ngs.opus@noaa.gov</em></a><em>.</em></p> <p>Above, I bolded several sentences that will be important to users currently performing projects using OPUS Projects. That is, all projects <strong><em>not submitted</em></strong> by <strong>June 6, 2025</strong>, will need to be <strong>reprocessed </strong>from the beginning in ITRF2020. </p> <p>Users should continue to check <a href="https://www.ngs.noaa.gov/" target="_blank" rel="noopener">NGS’s website</a> for announcements regarding the transition from the alpha site to the beta site. Future newsletters will address the<a href="https://www.ngs.noaa.gov/web/science_edu/webinar_series/multi-years-cors-v3.shtml" target="_blank" rel="noopener"> Multi-Year CORS Solution 3 (MYCS3)</a> and will highlight the beta products as they are released.</p> <p><p>The post <a rel="nofollow" href="https://www.gpsworld.com/evaluating-your-constraints-in-a-gnns-network-adjustment/">Evaluating your constraints in a GNNS network adjustment</a> first appeared on <a rel="nofollow" href="https://www.gpsworld.com">GPS World</a>.</p></p>
