Extended Data Fig. 6: Climatologies and trends of PNW temperature variability and land–atmosphere quantities.

Top row: 1981–2010 June–July climatologies (top panels) and 1979–2020 linear trends (bottom panels) of 2 m temperature (T2M), T2M variability (within-year standard deviation and skewness of daily anomalies), soil moisture (SM), and evaporative fraction (EF, calculated from daily latent heat flux [LHF] and sensible heat flux [SHF] as LHF/[LHF + SHF]). Bottom row: Climatologies and trends of four metrics of land–atmosphere coupling: the first three (correlations between LHF and SHF, LHF and SM, and EF and SM) represent the terrestrial component, while EF and T2M correlation represents the total feedback pathway. Correlation climatologies are created by correlating two variables (with June–July 1979–2020 trends removed) against each other throughout all June–July 1981–2010 days. Trends are between correlations within June–July of individual years (1979–2020). While SM and T2M are nearly everywhere anticorrelated, these metrics show where soil moisture deficit may causally affect T2M: LHF/SHF anticorrelation, LHF/SM correlation, EF/SM correlation, and EF/T2M anticorrelation indicate moisture-limited (versus energy-limited) regimes with potentially stronger land–atmosphere coupling, typical of transitional climate zones. If evapotranspiration is moisture-limited, under heating EF may decrease (SHF’s partition of flux increases), allowing for positive land–atmosphere feedbacks by further increasing T2M, decreasing SM, increasing SHF and decreasing LHF. Climatologically, such areas extend from the drier interior central West to the Columbia Plateau in eastern Washington and into interior British Columbia (bottom row, top panels). Trends indicate that much of the PNW has undergone strengthening in at least the terrestrial component of land–atmosphere coupling—most notably where soil moisture is climatologically moderate as opposed to extremely low, including much of BC’s Interior Plateau, much of the Cascade Range region (including near Portland and Seattle) and to the east of the Columbia Plateau. In some of these areas, T2M itself has become more coupled to EF, potentially signifying strengthened feedbacks—but such trends have not conclusively emerged overall. The spatial pattern of strengthening land–atmosphere coupling corresponds relatively well with warming, drying, and decreasing EF, and in some places with increasing T2M variability (areas of increasing T2M standard deviation and skewness correspond better to land–atmosphere correlation trends than to SM or EF trends alone).