Stats NZ’s 2 October 2026 temperature update reports evidence through 2025 from two complementary datasets. In the national seven-station series, annual temperature rose by 1.36°C from 1909 to 2025 according to a linear trend assessed at the 95% confidence level. The result describes the fitted average change across the full record, not a constant rise in every year or a forecast. Stats NZ cautions that the underlying path is nonlinear, so one full-period slope can understate or overstate warming rates in particular parts of the record.

The warmest year in this series was 2022, with an average temperature of 13.76°C. That was 1.02°C above the 1991–2020 climate baseline average of 12.74°C. Seven of the ten warmest years occurred since 2016; their annual averages were at least 13.37°C, and their anomalies ranged from 0.63°C to 1.02°C above that baseline. These comparisons use a defined 30-year normal. An anomaly is a difference from the baseline, not a separate measured temperature or a comparison with the previous year.

The national series is assembled from long records at seven climate stations: Auckland, Masterton, Wellington, Hokitika, Nelson, Lincoln, and Dunedin. Earth Sciences New Zealand combines local station records and adjusts for climatic differences and changes in exposure or instrumentation. This national long-run dataset is distinct from the separate 30-site analysis. For 1972–2025, average annual temperature was assessed as very likely increasing at 29 of 30 sites. The seasonal counts were 26 in spring, 25 in summer, 25 in autumn, and all 30 in winter. Overall warming was fastest in winter and slowest in spring.

A warm day is defined as a day when the maximum recorded temperature exceeds 25°C. From 1972 to 2025, warm days very likely increased at 22 of the 30 sites, and no site had a very likely decreasing trend. This is a count of days that crossed the specified threshold, not a measure of how far daily temperatures exceeded it. It also does not imply every year had more warm days than the one before: weather and climate variability remain within the longer-term record.

Stats NZ and the Ministry for the Environment use station data from Earth Sciences New Zealand’s National Climate Database. The 30 selected stations represent the local climates around their measurement sites and more closely reflect where many people live than the climate of each entire region. Some older records were extended using nearby former stations. Gridded estimates for 1972–2025 use spatial interpolation to estimate temperatures where observations are sparse or absent. Stats NZ says those maps show broad national patterns, not precise local conditions, and advises caution when comparing places or dates.

Trend direction and likelihood are assessed using the Mann–Kendall test, while Sen slope regression estimates average rates of change. “Very likely” means probability above 90%; “likely” is above 66%; below that threshold or without a determined direction is classed as indeterminate. Likelihood does not describe trend magnitude, and long-period trends can hide shorter variability, including climate oscillations. The datasets support monitoring and environmental reporting; they do not by themselves identify every cause of a particular year’s temperature or local weather event.