In the Northeast and Great Lakes regions of the United States, the influence of multiple lakes on overlying air can greatly affect lake-effect snowfall over downwind communities. To assess the impact of Lake Huron on snowfall downwind of Lake Erie, we simulated a lake-effect snow event which occurred from 1-6 January 2010 using the Regional Atmospheric Modeling System (RAMS). We found that the presence of Lake Huron enhances snowfall downwind of Lake Erie by almost 20\% and leads to much heavier local snowfall totals than when Lake Huron is not present. This increase in snowfall is due to a lake-to-lake (L2L) convective band, as secondary circulations associated with lake-effect convection form over Lake Huron and persist overland between the lakes before re-intensifying over Lake Erie. As these secondary circulations move over Lake Erie, low-level convergence from the secondary circulation induces mechanical lifting which accelerates the development of convection within the L2L band. Surface fluxes and convection over Lake Huron deepen the boundary layer, promoting deeper vertical development of this L2L band over Lake Erie. However, although this boundary layer modification strengthens the L2L band, we found that it actually reduced lake-effect snowfall produced by wind-parallel bands (WPB) in other parts of the lake. This indicates that boundary layer modification from upstream lakes may impact L2L bands differently than WPB which are not connected to upstream secondary circulations.

This dataset was produced by simulating a lake-effect snow event which occurred between 1-3 January 2010 with the Regional Atmospheric Modeling System (RAMS), an open-source, nonhydrostatic numerical weather model. The model was initiatlized and forced at lateral boundaries with ERA5 reanalysis data. Three simulations were conducted: A CONTROL simulation, without any modifications to the reanalysis, and which used initial water temperatures from a 1-degree horizontal resolution Reynolds-averaged global dataset, a NLH simulation in which the surface of Lake Huron was changed from water to mixed forest and in which initial soil and snow data over the former area of Lake Huron were adjusted to match those of neighboring Michigan, and a VARTEMP simulation which was identical to CONTROL except that data from the Great Lakes Environmental Research Laboratory (GLERL) were used for the initial water temperatures over the Great Lakes.

The model output files use a terrain-following sigma-z vertical grid structure and an Arakawa C-grid structure for vector quantities. To simplify analysis, we developed a post-processing routine for RAMS data to convert this grid structure to a standard Cartesian grid. We also output several derived variables in this post-processed output.

A PDF file describing the variables included in the RAMS analysis output files (those with extension .h5) as well as a .txt. file describing the variables present within the post-processed files (those with extension .nc) have also been uploaded here.

Finally, we have included the Buffalo sounding data used to construct Figure 4. This data comes from the National Oceanic and Atmospheric Administration (NOAA) Integrated Global Radiosonde Archive and is provided in .txt form. A product description PDF, as well as a README .txt file taken from NOAA which describes the structure, organization, and units within the bufsoundings_por.txt file, have also been uploaded here.

We should note that this collection is not intended for full data reproduction, but rather for those who would like to examine the raw and post-processed model data at a particular time of interest, in this case 2300 UTC on 2 January 2010. This is the time at which the bulk of the analysis and figures within the manuscript are focused, and therefore these data should allow for the reproduction of figures 7-8 and 10-14 as seen within the manuscript. For a guide to full data reproduction, please see the Zenodo repository at the following DOI: https://doi.org/10.5281/zenodo.10889801.