Supplement to "Proof of concept for Bayesian inference of dynamic rating curve uncertainty" (v1.0.1)

1. Summary

This public repository contains the data and scripts associated with “Proof of concept for Bayesian inference of dynamic rating curve uncertainty,” which is currently under review in Hydrological Sciences Journal.

1. Code, Data, and Attribute Descriptions

Uploaded are ten directories containing input data and various outputs of the analyses performed as a test case for the proposed Modified GesDyn—FlowAM—BaRatin method for rating curve uncertainty estimation (Morlot et al. 2014; Manfreda et al. 2020; Le Coz et al. 2014):

●      Lbn_R contains (1) Lbn_RC_rev.R, the R script file and (2) Lbn_R_glbenv.RData, which stores objects generated from the script. The script consists of admittedly too many lines of code (e.g., for data wrangling, parsing the BaRatin outputs, and producing figures used in the manuscript) that should have been split into multiple .R files, so please be guided by the outline and the comments and kindly reach out lest issues with running the code arise.  

●      Lbn_CSVs contains the following CSV files and a sub-directory:

○      R12_Lbn_rated.csv contains the benchmark discharge time series, which is used as input to the R script. This file is an input to the R script.

○      Lbn_DMs-csv.csv contains the following information about the available discharge measurements (DMs) or “gaugings” conducted by DPWH hydrographers on the Libungan River (Grebona, Cotabato, Philippines): gauging ID [“GaugingID”]; date of gauging (YYYY-MM-DD) [“Date”]; method of gauging [“Method”]; gauging conditions (qualitative, e.g., “Poor”) [“Gng_cond”]; location of the gauging cross-section with respect to the staff gauge, expressed as distance (m), positive if upstream and negative if downstream [“Loc_gage”]; stages (m) rectified to a uniform datum [“H_m”]; discharge  [“Q”]; maximum depth (m) relative to the water surface [“MaxDepth”]; water surface elevation (m) [“WSE”]; and minimum streambed elevation (m) [“MINSBE”]. This file is an input to the R script.

○      Lbn_DMs-aux.csv contains the same information as in Lbn_DMs-csv.csv, plus the following ancillary information: mean flow velocity (m/s) [“Velocity (mps)”]; channel width (m) from the left water edge to the right water edge [“Width (m)”]; mean flow depth (m) [“MeanDepth (m)”]; and wetted area (m²) [“Wet_Area (sqm)”].

○      Lbn-XS-csv.csv contains most of the same data as the above files, although expanded row-wise for each gauging vertical to enumerate or calculate the following cross-sectional geometric properties: stage at the gauging vertical [“H_rec_vert (m)”]; elevation at the gauging vertical [“Elev_H_rec_vert (m)”]; area (m²) of the gauging vertical [“A_vert (sqm)”]; wetted area (m²) or total cross-sectional area of flow, i.e., sum of the areas of the gauging verticals [“A_flow (sqm)”]; wetted perimeter (m) at the gauging vertical, i.e., length of segment touching the riverbed [“P_hyp (m)”]; total wetted perimeter (m) [“P_wet (m)”]; hydraulic radius [“Hrad (m)”]; and section factor [“Z_m”]. 

○      Lbn_H_m-csv.csv contains the continuous stage record, specifically the following information: hydraulic analog set to which the dates belong [“HAS”]; date of stage record (YYYY-MM-DD) [“Date”]; stage recorded in the morning (m) [“H_am”]; stage recorded at noon (m) [“H_nn”]; stage recorded in the afternoon or evening (m) [“H_pm”]; mean daily stage (m) [“H_bar”]; standard deviation of the mean daily stage (m) [“H_std”]; 95% expanded uncertainty of the mean stage (m) [“H_confint95”]; weather conditions in the morning upon stage reading (qualitative, e.g., “cloudy”) [“AM_con]; and weather conditions in the afternoon upon stage reading (qualitative, e.g., “moderate rain”) [“PM_con”]. Extra stage readings were made at high flows or in inclement weather (as indicated by the columns “H_ext_1”, … , “H_ext_5”); the intensity (e.g., “moderate”) and duration (e.g., “30 min”) of rainfall were also noted. This file is an input to the R script.

○      Lbn_FlowAM_pars.csv contains the fitting parameters for the decoupled Flow Area Models (following Manfreda 2018) for each hydraulic analog set; these are also reported in the appendix of the revised manuscript. This file can be exported from the R script.

○      Lbn_priors-csv.csv contains the hydraulic priors set for each physical or inferred rating curve parameter for each hydraulic analog set [“HAS”] from a chronologically homogeneous sample (CHS) [“Hubert”]. The prior estimates are given as the means and their associated uncertainties, expressed either as the 95% expanded uncertainty [“_95unc”] or the standard deviation [“_sd”] or both for a given parameter. These priors are also reported in the appendix of the revised manuscript. This CSV file is an input to the R script.

○      Lbn_XS_gaugings is a sub-directory containing CSVs with data on the gauging transects (totaling 77): width (m) of the gauging vertical [“Width_vert..m.”]; depth (m) relative to the water surface [“Depth..m.”]; and stage (m) at the gauging vertical [“H_m_vert..m.”]. All these files are inputs to the R script.

●      Lbn_HASs_CSVs_rev contains all 77 hydraulic analog sets, i.e., sub-samples of gaugings (original plus proxy) to be used as input to BaRatinAGE, with the following columns: stage (m) [“H”]; gauged stage uncertainty [“uH”]; gauged discharge (m³/s) [“Q”]; gauged discharge uncertainty [“uQ”]; and a column indicating whether the gaugings are active [“Active”]. These files were generated and exported from the R script.

●      Lbn_Hm_CSVs contains all sub-series of the available stage record to which the rating curves computed for each hydraulic analog set would be applied in BaRatinAGE.

●      Lbn_BaRatin_files contains the BaRatin projects, which were created and can be opened in BaRatinAGE, the graphic user interface for implementing the BaRatin framework and whose latest open-source version is available in Renard et al. (2022). Each project file covers a CHS, i.e., spanning a stability period. In the case of CHS 3, however, the sample was too large (comprising 44 hydraulic analog sets), and so it was subdivided, for convenience, into four project files.

○      Lbn_BaRatin_CHS_1.bar.zip

○      Lbn_BaRatin_CHS_2.bar.zip

○      Lbn_BaRatin_CHS_3a.bar.zip

○      Lbn_BaRatin_CHS_3b.bar.zip

○      Lbn_BaRatin_CHS_3c.bar.zip

○      Lbn_BaRatin_CHS_3d.bar.zip

○      Lbn_BaRatin_CHS_4.bar.zip

○      Lbn_BaRatin_CHS_5.bar.zip

●      Lbn_BaRatin_Q_rated_CSVs_rev contain all the flow sub-series computed (“reconstructed”) for each HAS, as computed in and exported from BaRatinAGE. These files are inputs to the R script.

●      Lbn_RCs_CSVs contains five sub-directories, each dedicated to a specific CHS. These sub-directories contain CSV files of the posterior rating curves and their associated uncertainties (i.e., with intervals for parametric and total uncertainties), as computed in and exported from BaRatinAGE. These files are inputs to the R script.

●      Lbn_Params_CSVs contains five sub-directories, each dedicated to a specific CHS. These sub-directories contain CSV files of the MCMC-simulated parameter estimates, as exported from BaRatinAGE, for each hydraulic analog set. These files are inputs to the R script.

●      Lbn_prior_CSVs contains prior rating curve envelopes for two sample hydraulic analog sets (HAS 9 and 41, specifically). The CSV files contain the same information as the .txt files therein and are used as inputs to the R script.

●      Lbn_posterior_CSVs contains BaRatin-derived posterior rating curve envelopes for the same hydraulic analog sets as above, although in two contrasting cases: (i) those simulated if the calibration data contained proxy gaugings vs. (ii) those simulated if only the original gaugings were used for calibration. The four CSV files in this folder are used as inputs to the R script.

1. References

Le Coz, J., Renard, B., Bonnifait, L., Branger, F. & Le Boursicaud, R. (2014) Combining hydraulic knowledge and uncertain gaugings in the estimation of hydrometric rating curves: A Bayesian approach. Journal of Hydrology 509, 573–587. doi:10.1016/j.jhydrol.2013.11.016

Manfreda, S. (2018) On the derivation of flow rating curves in data-scarce environments. Journal of Hydrology 562, 151–154. doi:10.1016/j.jhydrol.2018.04.058

Manfreda, S., Pizarro, A., Moramarco, T., Cimorelli, L., Pianese, D. & Barbetta, S. (2020) Potential advantages of flow-area rating curves compared to classic stage-discharge-relations. Journal of Hydrology 585, 124752. doi:10.1016/j.jhydrol.2020.124752

Morlot, T., Perret, C., Favre, A. C. & Jalbert, J. (2014) Dynamic rating curve assessment for hydrometric stations and computation of the associated uncertainties: Quality and station management indicators. Journal of Hydrology 517, 173–186. Elsevier. doi:10.1016/J.JHYDROL.2014.05.007

Renard, B., Horner, I., Coz, J. L. & Vigneau, S. (2022, December 20) BaRatin-tools/BaRatinAGE: BaRatinAGE 2.2.1. [object Object]. doi:10.5281/ZENODO.7463233