Select the state of interest, then the county or counties of interest. You may select up to five unique counties to compare.
Introduction
This tool provides county-level data and trends for unique minimum temperature (Tmin) time series across the conterminous United States from 1951–near present. The tool was developed to complement our Temperature Trends Dashboard (https://scipprisa.shinyapps.io/SCIPP_temp_dash/) but focuses on minimum temperatures, which, for many locations, are changing more than local average (Tavg) and maximum (Tmax) temperature. This tool uses a gridded dataset (nClimGrid-Daily) because gridded fields tend to be more representative of spatial patterns and produce more accurate regional or county averages when aggregated spatially.
Features
Four variables are available, summarized at the county level on an annual (1 January-31 December) or cool-season (1 July-30 June) basis. First, Freeze Days, defined as the count of cool-season days in which a county experienced temperatures at or below 32°F (0°C). Second, the Coldest Minimum Temperature, defined as the lowest county-wide minimum temperature during each cool season (e.g., one day each cool season). Third, the Warmest Minimum Temperature, defined as the warmest county-wide minimum temperature during each year (e.g., one day per year). Finally, a user-defined Minimum Temperature Threshold lets users select a Tmin value and view the annual count of days a county experienced temperatures either above or below that threshold. For example, selecting ‘≥ 65°F’ for East Baton Rouge Parish, Louisiana, generates a time series showing the number of days per year when the county-wide average Tmin was at or above 65°F. For more information on the variables, data, and methods, please visit our Documentation Tab.
Instructions
To get started, click the "Tool" tab, select a state, and then choose one or more counties (up to five, which may come from different states). Below the time series plot, you can export the results as an image or CSV. To reset the plot, use the "Clear Plot" option beneath the download options.
Source Data
Minimum temperature data are from nClimGrid-Daily (Durre et al., 2022). nClimGrid-D is a gridded dataset created by combining four station-based networks (Cooperative Observer Network, Automated Surface Observing Systems, Snow Telemetry, and Remote Automatic Weather Stations) with a resolution of approximately 0.0417° (nominally 5x5 km) and was designed for climate monitoring (Durre et al., 2022) by the National Centers for Environmental Information (NCEI). For more information on how nClimGrid-Daily is derived, please see Durre et al. (2022).
Durre, I., Arguez, A., Schreck III, C.J., Squires, M.F. and Vose, R.S., 2022. Daily high-resolution temperature and precipitation fields for the contiguous United States from 1951 to present. Journal of Atmospheric and Oceanic Technology, 39(12), pp.1837-1855.
The County-level shapefile was obtained from the U.S. Census Bureau via the U.S. County Boundary Shapefile dataset (TIGER/Line; U.S. Census Bureau, 2024).
TIGER/Line Shapefiles, (machinereadable data files) / prepared by the U.S. Census Bureau, 2024. https://www2.census.gov/geo/tiger/TIGER2024/COUNTY/
Methodology
To create unique, county-level time series, the County Boundary shapefile was converted from vector format to grid, matching the size and extent of nClimGrid-D. County boundaries often extended into areas where nClimGrid-D data were unavailable, such as the areas over large bodies of water. When converting the County Boundary shapefile to a grid, areas where nClimGrid-D data did not exist were ignored. For example, county time series around the Great Lakes (e.g., Ottawa County, MI), whose border extends over water in the Census Bureau’s shapefile, only represent non-water portions of the county. Areas where the resulting grid cell overlapped more than one county were assigned to the county in which the center of the grid cell was located. Data were processed in Python using the ArcPy module for geospatial analysis and the Pandas module for tabular analysis.
To test for trends in the above-mentioned series, Ordinary Least Square (OLS) regression is used. OLS regression provides parameter estimates that quantify how variables change during a period when the independent variable is time (year). Each parameter estimate is multiplied by ten to show the average rate of change per decade over the period of record (1951-near present).
Freeze Days
For a freeze day to occur, at least 50% of the grid cells in a county on a given day had to record Tmin ≤ 0 °C. All qualifying days within each cool season (July 1–June 30) were then summed to generate the county-level time series. Because the cool season spans two calendar years, the cool season is assigned to the year of January. For example, the first season covers July 1, 1951–June 30, 1952, and is labeled Cool Season 1952.
Coldest and Warmest Minimum Temperature
The Coldest Minimum Temperature (Tmin) is calculated by averaging Tmin across all grid cells within a county each day during the cool season (July 1–June 30) and then identifying the single lowest daily value. This represents the coldest average minimum temperature for the county in each cool season.
The Warmest Minimum Temperature (Tmin) is calculated in the same way, but over the calendar year (January 1–December 31). This represents the warmest average minimum temperature for the county in each year.
Minimum Temperature Threshold
This feature allows users to set a custom minimum temperature (°F) and view the annual number of days when a county’s average Tmin (mean of all grid cells) was above or below that threshold. Unlike Freeze Days or the Coldest/Warmest Tmin variables, this option is fully user-defined. To use it, first select either the ≥ or ≤ condition, then enter the minimum temperature of interest (you can also adjust the value with the arrows in the ‘Enter Tmin Threshold (°F)’ box). Finally, select one or more counties to generate the annual time series. Note, these series are summarized annually (1 January-31 December).
Caveats
Gridded data like nClimGrid-Daily are not synonymous with station-based data. For example, station-based data are highly localized and do not represent areas well beyond their location. This is a key difference between this tool and our Temperature Trends Dashboard Tool. Additionally, station-based data generally have more variability (higher and lower extremes) than a grid. Grids are spatially averaged, which reduces variability. Tests were conducted to ascertain the strength of the association between stations within counties and derived county time series developed in this tool. Results showed, for most counties, a high degree of association; however, larger differences were found between stations west of the Rocky Mountains and stations near unique features (e.g., water bodies, topographic features, etc.). Please see Durre et al. (2022) for a more detailed explanation of potential differences between station and gridded data.
