Work Description

Date: 24 Feb, 2020 Dataset Title: Monthly water balance estimates for the Laurentian Great Lakes from 1950 to 2019 Dataset Creators: H.X. Do, J.P. Smith, L. M. Fry, & A.D. Gronewold Dataset Contact: Hong X. Do hongdo@umich.edu To Cite Data: Do, H., Smith, J., Fry, L., Gronewold, A. Monthly water balance estimates for the Laurentian Great Lakes from 1950 to 2019 [Data set]. University of Michigan - Deep Blue. https://doi.org/10.7302/0rsp-v195 Key Points: - We present a new estimate of monthly water balance from 1950 to 2019 for the Laurentian Great Lakes, the largest freshwater system on Earth. - Lake storage changes and water balance components for each lake were estimated using a Bayesian framework that assimilated multiple independent data sources. - The source code as well as inputs of the statistical model are also made available. Research Overview: Water balance of the Laurentian Great Lakes, the largest freshwater system on Earth, is of great importance to the sustainable development of both the United States and Canada. In recent years, the water levels across all of the Laurentian Great Lakes have exhibited unprecedented dynamics, indicating potential implications of climate changes. To understand the factors driving changes in water levels, it is compulsory to have a reliable estimate of the components of the water balance (i.e. over-lake precipitation, over-lake evaporation, lateral tributary runoff, connecting channel flows, and diversions) that collectively contribute to changes in water budget of each lake. However, existing data sources that represent (each of) these variables are often developed independent with limited consideration to keep one water budget term consistent to the others. As a result, the water balance of the Great Lakes cannot be solved over multiple time periods using existing data alone. Here we present a new estimate of monthly water balance from 1950 to 2019 for the Laurentian Great Lakes. This dataset was developed using a Bayesian framework, encoded in the Large Lakes Statistical Water Balance Model (L2SWBM). The L2SWBM assimilated multiple independent data sources to infer feasible values of each water balance component. A conventional water balance model was also used to restrain new estimates, ensuring that the water balance can be reconciled over multiple time periods. The new estimates are useful to investigate changes in water availability, or benchmark new hydrological models and data products developed for the Laurentian Great Lakes Region. The source code as well as inputs of the L2SWBM model are also made available, and can be adapted to include new data sources for the Great Lakes, or to address water balance problems on other large lake systems. Methodology: The data are model output form the Large Lakes Statistical Water Balance Model (L2SWBM). Instrument and/or Software specifications: - The L2SWBM was developed in R programming environment (ver. 3.6.1). To execute the R-scripts available in this dataset, prospective users must install the following dependant libraries: `rjags` (ver. 4.9), `doParallel` (ver. 1.0.15), and `rlecuyer` (ver. 0.3.4). It is important to note that, `rjags` requires the JAGS (Just Another Gibbs Sampler; ver. 4.3.0) to be installed. This software is open-source and is available at `https://sourceforge.net/projects/mcmc-jags/files/JAGS`. - The `rootDir` variable in `0_main.r` is set to an absolute path, which causes an error if that path does not exist. Prospective users should change that working directory prior to executing the simulations. Files contained here: This data set contains the new estimate of the Great Lakes water balance together with the L2SWBM source code and inputs synthesized for this project (monthly data available up to December 2019 depending on variables). All of these files are compressed as a zip-archive. The zip file size is approximately 4 MB, and contains the below files: A. The L2SWBM source codes The model is organized in ten individual R-script files, and accompanied by a model configuration file (csv format; accompanied with a text file explaining the variables of this configuration file). The configuration file can be adjusted to include more data sources or focus on a different analysis period. B. Inputs of the L2SWBM Inputs of the L2SWBM were synthesized from a range of independent data sources. These independent data records were used to derive the prior distribution and likelihood functions for each of the variable. Data for each variable of a specific lake is stored in a separate csv file. All input files were contained in the input folder and accompanied with a readme file (pdf format). C. Outputs of the L2SWBM. The L2SWBM generated multiple outputs that are organized as three individual files and four separate folders within the "output" folder. The folders and individual files are described below: - File "GLWBData_config.csv" : A csv file records the configuration of the model parameters - File "GLWBData.bug.r" : The R-version of the BUGS (Bayesian inference Using Gibbs Sampling) model of the Bayesian inference framework - File "GLWBData_Closure.csv" : A csv file contains the water balance closure accounting over 1-, 6-, and 12-month periods - Folder "plot_prior" : PDF files contain plots of the prior probability distributions of the water balance components. Naming convention is <VAR>PriorCompare.pdf (e.g., "evapPriorCompare.pdf") - Folder "plot_preview" : PDF files contain plots of inputs over the analysis period. Each pdf file represent all data for a specific lake over one decade (from decade no. 0 to decade no. n-1, with n = no. of years/10). Naming convention is <LAKE>TS_Preview_<DECADE No.>_<PROJECTNAME>.pdf (e.g., "superiorTS_Preview_d0_GLWBData.pdf") - Folder "plot_posterior" : PDF files contain plots of inputs over the analysis period. Each pdf file represent all data for a specific lake over one decade (from decade no. 0 to decade no. n-1, with n = no. of years/10). Naming convention is <LAKE>TS_ALL_<DECADE No.>_<PROJECTNAME>.pdf (e.g., "miHuronTS_ALL_d5_GLWBData.pdf") - Folder "ts_posterior" : csv files contain monthly inference (2.5, 50 and 97.5 percentile of the MCMC iterations) of each water balance component across each lake over the analysis period. Naming convention is <LAKE>_Preview_<DECADE No.>_<PROJECTNAME>.pdf (e.g., "erieRunoff_GLWBData.csv") Related publication(s): Do, H.X., Smith, J., Fry, L.M., & Gronewold, A.D., Seventy-year long record of monthly water balance estimates for Earth�s largest lake system (pending for submission) Use and Access: This data set is made available under a Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0).