Hotter Years, Hotter Days?

Research topic and question

• Can the Hottest Ten Days Reveal What Averages Hide?

Do a year’s ten hottest days reveal warming more clearly than its average temperature? Investigate New York’s Central Park using the latest available NOAA GHCN-Daily data (station USW00094728), starting in 1991. Download the data from https://www.ncei.noaa.gov/pub/data/ghcn/daily/ and use its documentation. Compare complete years and explain the evidence and limitations in English.

Research findings

The conclusion holds for Central Park during 1991–2025 after accounting for correlation between years and using a robust trend estimator.

Metric	Newey–West OLS slope	95% HAC CI	Theil–Sen slope	95% CI
Annual mean temperature	+0.305 °C/decade	+0.159 to +0.452	+0.323 °C/decade	+0.059 to +0.550
Mean of ten hottest maxima	−0.164 °C/decade	−0.451 to +0.123	−0.241 °C/decade	−0.736 to +0.400

Newey–West inference used three annual lags. The annual-mean trend remains clearly positive (p = 0.00005), whereas the hot-ten trend remains statistically indistinguishable from zero (p = 0.264). Theil–Sen estimates give the same interpretation. The negative hot-ten estimates are not persuasive evidence of cooling because their intervals are wide and cross zero.

All 35 years from 1991 through 2025 passed the completeness and quality-flag checks.