Timescales and statistical definitions
Weather describes atmospheric conditions over days to weeks. A cold snap, a rainstorm, a heat wave, a tornado: these are weather events. Climate describes the statistics of weather over decades and longer, summarizing what happens on average and how much variability to expect. When a meteorologist forecasts next Wednesday, they are predicting weather. When a climatologist estimates global temperature in 2080, they are predicting climate.
This distinction reveals why a single cold winter does not refute global warming. Climate change is a shift in the distribution of outcomes, not the elimination of variability. A warming planet still experiences freezing temperatures and snowstorms; it simply makes extreme heat more frequent and extreme cold less frequent. The baseline distribution has shifted poleward and upward.
Distinguishing natural variation from forced change
Decades of data are necessary to separate climate signals from natural weather variability. Solar cycles, volcanic eruptions, and ocean circulation oscillations like ENSO and the Atlantic Multidecadal Oscillation create decadal-scale climate fluctuations that can mask or amplify underlying trends. A decade-long cooling from a volcanic eruption does not reverse a multi-decadal warming trend; it merely interrupts it temporarily.
Climate scientists use large ensembles of model simulations to estimate how much internal variability masks the forced response to greenhouse gas increases. The larger the forced signal relative to natural variability, the sooner the trend emerges clearly above the noise. This is why detecting climate change robustly required 30-50 years of data from the late 20th century onward.