How WaterTableCheck Calculates a County's Water Table Trend
One well, fit to its own history
Each qualifying county is matched to one USGS National Groundwater Monitoring Network well: at least 15 years between its first and last reading, a reading within the last 10 years, and 6 or more usable depth-to-water measurements (parameter code 72019 from USGS's field-measurements API). An ordinary-least-squares line is fit through that well's full qualifying record; its slope, in feet per year, becomes the trend reported on the county page.
Why "mild / moderate / severe" are not universal thresholds
Among counties that are declining, this site splits the decline rate into thirds using this dataset's own distribution (mild below 1.24 ft/decade, moderate below 4.1, severe above). No published national standard defines a "severe" aquifer decline rate, so this is an honest empirical split of this dataset, not an external benchmark, and is labeled that way.
Coverage, stated as a number
1,449 of 3,222 US counties (45%) have a qualifying well today. The rest show a plain "no data" page rather than a trend borrowed from a neighboring county, because groundwater behavior is local and a neighboring county's aquifer can move in the opposite direction. See the fastest-moving counties guide for what this model looks like at its extremes, and the well-pump risk guide for what a declining figure actually means for a pump.
Three counties, three outcomes of the same model
Carroll County, IL shows how severe a fitted decline can get when a well has decades of consistent readings. DeKalb County, IL shows the same model applied to a well where the table is getting shallower, not deeper. San Diego County, CA shows what every page looks like when no well in that county clears the 15-year bar at all: a stated gap, not a number.