Anna L. Conlen Tara L. Morgan-King 20230330 spreadsheet https://doi.org/10.5066/P9YZ07JY Model archive summary (MAS) describing the development of a continuous 15-minute suspended-sediment concentration (SSC) time series regression model for the site: South Mokelumne River at New Hope Bridge near Walnut Grove, California, U.S. Geological Survey (USGS) station number 11336680. The SSC time series is computed from instream turbidity data that is managed by the USGS using a YSI 6-series multi-parameter water quality sonde. Time-series SSC data are needed to determine sediment transport in the Sacramento-San Joaquin River Delta, California. This data release includes the MAS, calibration dataset and link to the USGS turbidity data and resulting 15-minute SSC time series data. The MAS includes cited publications as well as all other supporting documentation, techniques and methods used to develop the SSC regression model. 20101220 20150323 ground condition Complete None planned -121.498189 -121.489906 38.225741 38.219706 ISO 19115 Topic Category environment USGS Thesaurus suspended material (water) Water Quality Geomorphology Hydrology Water Resources sediment transport ISO 7027 Turbidity USGS Metadata Identifier USGS:62794131d34e8d45aa6e3bdf Common geographic areas California San Joaquin County San Joaquin Delta Lower Cosumnes-Lower Mokelumne None. Please see 'Distribution Info' for details. None. Users are advised to read the dataset's metadata thoroughly to understand appropriate use and data limitations. USGS-authored or produced data and information are in the public domain from the U.S. Government and are freely circulated with proper metadata and source attribution. Please recognize and acknowledge the U.S. Geological Survey as the originator(s) of the dataset ad in the products derived from these data. Anna L. Conlen U.S. Geological Survey mailing and physical

3115 Ramco St, Suite 180

West Sacramento CA 95691 USA 916-591-0784 aconlen@usgs.gov Funding is provided by the U.S. Bureau of Reclamation and the Department of the Interior. QA/QCed turbidity data were produced using the National Water Information System (NWIS): https://nwis.waterdata.usgs.gov/usa/nwis/uv/?cb_00010=on&cb_00060=on&cb_00095=on&cb_63680=on&cb_63680=on&cb_72137=on&cb_72255=on&format=gif_default&site_no=11336680&legacy=1&period=&begin_date=2010-12-20&end_date=2015-03-23. The regression model development and data computation used a combination of the USGS Surrogate Analysis and Index Developer Tool (SAID) and Matlab and/or R software (see documentation and references in the MAS). Filename: "11336680_SMR_MAS.docx". USGS collected turbidity data from a YSI 6-series multi-parameter water quality sonde at South Mokelumne River at New Hope Bridge near Walnut Grove, California (SMR). Turbidity data were analyzed and approved by the USGS California Water Science Center per USGS guidelines, following Wagner and others (2006). Tidally-filtered discharge data were not used to compute SSC but were evaluated as a potential predictor; data were collected, computed, reviewed, and approved by the USGS California Water Science Center and retrieved from NWIS-TS. Methods to compute discharge follow Levesque and Oberg (2012). The methods for computing tidally-filtered discharge data follow USGS guidelines (U.S. Geological Survey, 2010). The 15-minute turbidity time-series data and hourly tidally-filtered discharge data are located at: https://waterdata.usgs.gov/nwis/uv?site_no=11336680. All suspended-sediment samples in the calibration dataset were collected following protocols and methods described in Edwards and Glysson (1999) and U.S. Geological Survey (variously dated) and are representative of the stream cross section. Multiple models were evaluated. Cross validation was used to assess how well the model would predict the independent dataset. Quality Assurance of sediment lab analyses are conducted by the USGS Quality Systems Branch (QSB) within the Sediment Laboratory Quality Assurance Project: https://qsb.usgs.gov/slqa/. The dataset is considered logically consistent and does not include duplicated or otherwise erroneous data. The 27 samples in the calibration dataset were collected over a range of turbidity values observed at the station. The dataset is considered complete for the information presented, as described in the abstract. Users are advised to read the rest of the metadata carefully for additional details. U.S. Geological Survey 2022 tabular digital data https://nwis.waterdata.usgs.gov/usa/nwis/uv/?cb_00010=on&cb_00060=on&cb_00095=on&cb_63680=on&cb_63680=on&cb_72137=on&cb_72255=on&format=gif_default&site_no=11336680&legacy=1&period=&begin_date=2010-12-20&end_date=2015-03-23 Digital and/or Hardcopy 20101220 20150323 ground condition USGS Turbidity Data Continuous 15-minute time-series turbidity data were retrieved from: https://nwis.waterdata.usgs.gov/usa/nwis/uv/?cb_00010=on&cb_00060=on&cb_00095=on&cb_63680=on&cb_63680=on&cb_72137=on&cb_72255=on&format=gif_default&site_no=11336680&legacy=1&period=&begin_date=2010-12-20&end_date=2015-03-23 Continuous 15-minute turbidity data were collected by the U.S. Geological Survey (USGS) California Water Science Center. Turbidity data were measured using a YSI 6-series sonde with a model 6136 sensor, reported in Formazin Nephelometric Turbidity Units (FNU). The QA/QCed data were obtained from the USGS National Water Information System (NWIS-TS). The sonde was cleaned and calibration checked during regular site visits, per USGS Techniques and Methods 1-D3. Turbidity data were analyzed and approved per USGS guidelines (Wagner and others, 2006). Turbidity data were loaded into the Surrogate Analysis and Index Developer (SAID) Tool to pair with the discrete sediment data using a ± 15-minute matching window. 2020 Sample Collection: Sediment sample collection was conducted by USGS personnel following USGS-approved and publicly available field methods for the collection of suspended sediment. Samples were collected during regular site visits and runoff events to target high sediment concentrations. Sample results and associated parameters are stored in the National Water Information System (NWIS) database: https://waterdata.usgs.gov/nwis. Technicians collected samples at SMR using either a D-74 or a D-96 suspended-sediment bag sampler approximately 1,400 feet downstream of New Hope Bridge. The D-74 sampler was used when sampling depths did not exceed 15 feet and the sampler was available (Edwards and Glysson, 1999). The channel cross section can approach 13 feet deep in the thalweg with a mean depth of approximately 8 feet. For the 27 samples analyzed for percent fines, the mean value was 88 percent fines. Any potential sampling bias due to non-isokinetic sampling is considered minimal. Discrete, boat-based sediment samples are representative of the stream cross-section. Either the Equal Discharge Increment (EDI) or Equal Width Increment (EWI) method was used to determine the locations of sampling verticals along the transect where 29 suspended-sediment samples were collected. Samples were predominantly collected using the EDI method as velocities are not always isokinetic due to the tidal nature of the site (Table 4-5 from U.S. Geological Survey, 2006). A boat-based discharge measurement was collected immediately before EDI sampling with an acoustic Doppler current profiler to determine the location of each vertical. The EWI method was used to collect four samples in the calibration dataset. 2020 Sample Analysis: Sediment samples were analyzed for SSC using USGS-approved laboratory methods. Samples were analyzed by the USGS Sediment Laboratory in Marina or Santa Cruz, California. All samples were analyzed for sediment concentration, in milligrams per liter (mg/L) by the filtration method and 27 samples were also analyzed for the percentage of fines (particle size less than 0.063 millimeters). The sand/fine break analysis can be used to identify dataset variability and potential outliers and shows that sediment at this station is composed of mostly fines - 88 percent fines on average. EWI samples and 9 EDI samples were composited before analysis. Each of the five EDI verticals from the remaining EDI samples were analyzed individually by the lab for quality control purposes. The set average SSC of the five verticals was computed and used in the calibration model dataset. Turbidity data was missing for two sediment samples: the sample collected on December 20, 2010 was collected before the turbidity time series began and turbidity data during the sample on December 3, 2013 were deleted due to fouling, leaving a total of 27 samples in the calibration dataset. 2020 Model Development: The model development process is documented in the MAS, including several diagnostic statistics and graphs. Suspended-sediment concentrations at this station were computed from a regression model between log10-transformed turbidity and SSC using USGS standardized approaches and guidelines to compute time-series SSC from in-stream turbidity-sensors (Rasmussen and others, 2009). Regression methods are also described in Helsel and others (2020); see references in the attached file "11336680_SMR_MAS.docx"). The USGS Surrogate Analysis and Index Developer Tool (SAID) was used to pair surrogate turbidity data with the discrete sediment data (Domanski and others, 2015). The complete model calibration dataset is described in the attached file "11336680_SMR_MAS.docx" and the digital csv file "11336680_SMR_CalibrationDataset.csv" can be downloaded. Multiple models were evaluated using Matlab and/or R including simple linear regression (SLR) and multiple linear regression (MLR). The most common estimation technique is SLR, but MLR is an alternate tool for computing SSC when the SLR model standard percentage error (MSPE) statistic is larger than 20 percent (Rasmussen and others, 2009). Multiple models were evaluated but a MLR model was only considered if the average of the upper and lower MSPE improved more than 10 percent compared to a SLR model. Models with either one explanatory variable (turbidity) and models with both turbidity and hourly, tidally-filtered discharge were evaluated. The best model was chosen based on residual plots, model standard error, R2, significance tests (p-values), correlation of explanatory variables, and PRESS (prediction error sum of squares) statistics. Values for the statistics and metrics were computed for various models. Refer to the MAS document for references. 2021 Final Model: The best SSC model is a log-log model with log10- transformed turbidity as the surrogate. This decision is based primarily on residual plots and model statistics. Linear and log-transformed models cannot be compared using the RMSE and PRESS diagnostic statistics, as log units are not directly comparable to mg/L. MSPE, which is RMSE expressed as a percentage, can be used to compare transformed and un-transformed models. Though the linear model has a slightly lower MSPE value than the log10-transformed model, the plot of residuals vs fitted values is more homoscedastic for the log model and the normal probability plot of residuals is approximately linear for the log model. The lack of random scatter in the residual plots for the linear SLR model violates model assumptions. Multiple transformations were evaluated, and the final model was peer reviewed and is documented in an electronic model archive verification following USGS policy. Model documentation was reviewed by USGS California Water Science Center Specialists and is summarized in the attached file "11336680_SMR_MAS.docx". Logarithmic transformations produce a bias when transformed back to original units, resulting in an underestimation of the mean response. This bias can be eliminated by multiplying the estimated response by a nonparametric-smearing bias correction factor, which was determined to be 1.042 for the chosen model (Duan, 1983). The final regression model is: log10(SSC) = 0.556 + 0.744 log10(Turbidity), where SSC is suspended sediment concentration in milligrams per liter and Turbidity is water turbidity in Formazin Nephelometric Units (FNU). The coefficient of determination (R²) is 0.876. Refer to the MAS document for references. 2021 SMR_XSection.jpg JPG map of station location and cross-section transect location. Producer Defined 11336680_SMR_CalibrationDataset.csv Comma Separated Value (CSV) file containing calibration data. Producer Defined Observation No. Numbered observations in dataset - 27 observations total. Producer Defined 0 27 Date/Time Date (MM/DD/YYYY) and time (hh:mm) suspended sediment sample data, in Pacific Standard Time. Producer Defined 12/30/2010 15:00 12/14/2014 14:37 logSSC log10 SSC - discrete sample data Producer Defined 0.699 1.91 logTurb log10 USGS turbidity data matched to discrete SSC samples Producer Defined 0.342 1.81 SSC Discrete suspended-sediment concentration lab analyzed result. Producer Defined 5 82 mg/L Turb USGS turbidity data (FNU), collected with a YSI 6-series sonde Producer Defined 2.2 64.0 FNU Percent fines Particle size less than 0.063 millimeters, in percent Producer Defined 63.5 96.8 percentage Computed logSSC log10 model estimated SSC values Producer Defined 0.811 1.9 Computed SSC Model estimated SSC values Producer Defined 7 83 mg/L Residual log10 model residuals Producer Defined -0.192 0.315 Normal Quantiles Normal quantiles from log-log model (SAID report) Producer Defined -2.01 2.01 Censored Values No samples were censored. Producer Defined Censored values from the SAID application. No samples were censored in this dataset. Model calibration dataset and station location image. A link to 15-minute turbidity data and resulting SSC time-series data can be found on the ScienceBase landing page. The entity and attribute information were generated by the individual and/or agency identified as the originator of the dataset. Please review the rest of the metadata record for additional details and information. GS ScienceBase U.S. Geological Survey mailing address

Denver Federal Center, Building 810, Mail Stop 302

Denver CO 80225 United States 1-888-275-8747 sciencebase@usgs.gov Unless otherwise stated, all data, metadata and related materials are considered to satisfy the quality standards relative to the purpose for which the data were collected. Although these data and associated metadata have been reviewed for accuracy and completeness and approved for release by the U.S. Geological Survey (USGS), no warranty expressed or implied is made regarding the display or utility of the data for other purposes, nor on all computer systems, nor shall the act of distribution constitute any such warranty. Digital Data https://doi.org/10.5066/P9YZ07JY None 20230330 Anna L Conlen U.S. Geological Survey mailing and physical

3115 Ramco St, Suite 180

West Sacramento CA 95691 USA 916-591-0784 aconlen@usgs.gov FGDC Content Standard for Digital Geospatial Metadata FGDC-STD-001-1998