Phenological shifts in lake ice cover across the Northern Hemisphere: A glimpse into the past, present, and the future of lake ice phenology

<b>Abstract:</b> Long-term ice phenology records enumerate the effects of climate change on Northern Hemisphere lakes. This study uses lake ice phenological records for a gradient of small to large lakes (0.1 to 31967.8 km 2 in lake surface area) obtained from community science networks where we compiled in situ ice phenological records for 2499 lakes across 15 countries for an average of 30 years. These data revealed that for the last 50 years (1971-2020), the annual mean duration of lake ice cover decreased at a rate of 7.9 days per decade, with a global acceleration in lake ice loss following the late 1980s. We forecasted that ice duration at the end of the century (2070-2099) will decrease on average 29 days when compared to the historical time period (1971-2000) for the Shared Socioeconomic Pathway (SSP) 1-2.6 climate scenario (SSP126), 43 days for SSP370, and 49 days for the SSP585 scenario. The widespread loss of ice portends ecological,cultural, and economic consequences for ecosystems and communities that depend on lake ice cover.The NH_plusCLIC_lakes_ts_version1.1.csv dataset contains the following variables:<b>LAKEID: </b>Key ID column for this dataset and "NH_plusCLIC_lakes_hydrolakes_version1.1.csv" dataset.<b>lakename: </b>Name of the lakes.<b>year: </b>Year of the observation. For example, 1990 is the year value for the winter season 1990-1991.<b>froze: </b>Data about whether the lake froze that year or not. values are Y= yes, and N= no.<b>latitude: </b>Lake latitude coordinate<b>longitude: </b>Lake longitude coordinate<b>ice_on: </b>Ice-on date<b>ice_on_doy: </b>Ice-on day of the year calculated against distance from 31st December.<b>ice_off: </b>Ice-off date<b>ice_off_doy: </b>Ice-off day of the year calculated against distance from 31st December.<b>country: </b>Country name<b>season: Winter season years </b>(e.g., the value will be 1990-1991 for the winter season between years 1990 and 1991)<b>PPT01-12: </b>Precipitation value (mm/month) for 12 months where PPT06-12 corresponds to precipitation values for June-December averages of the previous year (e.g., 1990) and PPT01-05 corresponds to precipitation values for January-May averages of the next year (e.g., 1991).<b>Tave01-12: </b>Temperature value (Celsius/month) for 12 months where Tave06-12 corresponds to temperature values for June-December averages of the previous year (e.g., 1990) and Tave01-05 corresponds to temperature values for January-May averages of the next year (e.g., 1991).The NH_plusCLIC_lakes_hydrolakes_version1.1.csv dataset contains the following variables:<b>LAKEID: </b>Key ID column for this dataset and "NH_plusCLIC_lakes_ts_version1.1.csv" dataset.<b>LAKENAME: </b>Name of the lakes.<b>lat:</b> latitude coordinate<b>lon: </b>longitude coordinate<b>Hylak_id:</b> HydroLAKE IDs<b>Country:</b> Country names<b>Continent:</b> Continent names<b>Lake_area:</b> surface area of the lake (km²)<b>Shoreline:</b> Lake shoreline length (km)<b>Shore_dev:</b> Shoreline development (unitless)<b>Vol_total:</b> Total lake volume (million cubic-meter)<b>Depth_avg:</b> mean lake depth (meter)<b>Elevation:</b> Lake elevation (meter)<b>Wshd_area:</b> Lake watershed area (million cubic-meter)<b>Dmax_use_m:</b> Lake maximum depth (meter)The NH_coordinates.csv dataset contains the following variables:<b>Hylak_id:</b> HydroLAKE IDs of 1.24 million lakes projected<b>long:</b> longitudes<b>lat:</b> latitudesThe hydrolakes_morpho.csv dataset contains the same morphological variables as NH_plusCLIC_lakes_hydrolakes_version1.1.csv (for 1.24 million hydro-lakes used for prediction)Folders contain <b>.rds</b> format model files. The subfolder '<b>Used Random Forest Models'</b> contains the models used for projections, and the subfolder<b> 'Compared Models'</b> includes the initial comparison models.