Heat wave seasons stretched 8.71 days per decade across 45 years of global land data

The two ends of the season are not moving at the same speed. The last heat wave of the year is drifting later by 5.41 days per decade, and the first is arriving earlier by 3.29 days per decade ^[3][4]. Divide 5.41 by 8.71 and the later finish supplies about 62 percent of the lengthening ^[14]. The end-date trend runs roughly 1.64 times the onset trend ^[15].

Each season's first heat wave was sorted into rapid-, moderate- or slow-onset by where the temperature peak fell within the event ^[8]. The classification does not report a warming rate in degrees per day. An event that reaches its hottest point early in its own span counts as rapid whether it climbed two degrees or six.

The two halves of the paper also rest on different lengths of record. The timing trends come from 45 years of data, 1979 to 2023 ^[2]. The rise in rapid-onset first events was measured over 2001 to 2023, a 23-year window, about half as long ^[9][19]. The paper describes that increase as significant within heat wave-affected land areas ^[9].

Xu Wenfang of the South China Botanical Garden, the study's first author and one of its corresponding authors, said: "If heat waves arrive earlier and develop more rapidly, the time available for societies and ecosystems to prepare and adapt could become even shorter" ^[11]. The researchers wrote that onset and end timing, along with how quickly the season's first heat wave reaches its peak, could be considered alongside intensity and duration in heat monitoring and risk assessment ^[12].

The crop result is an exposure count. Alongside the increase during flowering and silking, heat wave exposure also extended into earlier and later stages of crop development ^[10]. The study does not report yields.

One check on the shape of the trend: 8.71 days per decade over 4.5 decades comes to 39.2 days, within a day of the total the paper reports for the full record, so the average lengthening is close to linear ^[16]. The spatial picture is less even. Longer seasons appear across 92.1 percent of global land and earlier starts across 72 percent, a gap of 20.1 percentage points ^[7][17]. In my view season length is the figure to plan against, because it covers more land area than either end-date trend on its own and it sets how many weeks of the calendar need cover. Changes were generally larger in drylands ^[7].