Shift of soil moisture-temperature coupling exacerbated 2022 compound hot-dry event in eastern China

The hourly temperature (2 m), dewpoint temperature (2 m), geopotential height (500 hPa), soil moisture (0–7 cm), wind speed (10 m), surface sensible heat flux, surface latent heat flux, surface net solar radiation, surface net thermal radiation, evaporation and potential evaporation (the amount of evaporation under existing atmospheric conditions from a surface of pure water, giving an indication of the maximum possible evaporation) data with a 0.25° × 0.25° spatial resolution for the period 1982–2022 were obtained from the ERA5 dataset provided by the European Centre for Medium Range Weather Forecasts (Hersbach et al 2020). The surface net radiation was calculated by subtracting the surface net thermal radiation from the surface net solar radiation (Guo et al 2016, Zhou and Wang 2016). All the hourly data were aggregated into averages of daily or longer time scales in this paper.

Daily CHDEs were defined by daily mean temperature and soil moisture (negative) simultaneously exceeding the corresponding historical 90th percentile thresholds (T90 and SM_n 90) (Ridder et al 2020, Weber et al 2020). For the calculation of T90 and SM_n 90 for each grid point, 62 d in July and August of each year within the 30 year baseline period (1982–2011) were used. In addition, the daily hot-only events (T > T90 & SM_n ⩽ SM_n 90) and dry-only events (T ⩽ T90 & SM_n > SM_n 90) were also defined as control groups.

To quantify the intensity of heatwaves in CHDEs, the daily heatwave magnitude (Russo et al 2015, Ma and Yuan 2023) of a given compound hot-dry day was calculated as:

$$\begin{equation*}{M_{\text{d}}}\left( {{T_{\text{d}}}} \right) = \frac{{{T_{\text{d}}} - {T_{25{\text{p}}}}}}{{{T_{75{\text{p}}}} - {T_{25{\text{p}}}}}}\end{equation*}$$

where$\,{T_{\text{d}}}$ is the daily mean air temperature of a compound hot-dry day, ${T_{25{\text{p}}}}$ and ${T_{75{\text{p}}}}$ are the 25th and 75th percentile values of the daily temperature of each grid cell within the baseline period (July and August 1982–2011), respectively. The annual heatwave intensity of CHDEs is the average of all the ${M_{\text{d}}}\left( {{T_{\text{d}}}} \right)$ in each year, and is zero if there was no compound hot-dry day at a given grid cell in a given year.

The strength of soil moisture-temperature coupling was assessed using the diagnostic π (Miralles et al 2012), which is based on two energy balances of evaporation and potential evaporation. This metric can be used to determine the related heating processes between land and atmosphere. Based on daily data, this method derives the soil moisture-temperature coupling on a daily scale, and the metric is defined as:

$$\begin{align*}\pi &= e{^{^{\prime}}} \times T{^{^{\prime}}}\end{align*}$$

$$\begin{align*}e^{\prime} &= H^{\prime} - {H_{\text{p}}}^{\prime} = {\left( {{R_n} - \lambda E} \right)^{\prime}} - {\left( {{R_n} - \lambda {E_{\text{p}}}} \right)^{\prime}}.\end{align*}$$

The primes represent daily anomalies of each variable expressed as the number of standard deviations relative to the climatological (1982–2001) mean. ${R_n}$ refers to the surface net radiation, $T$ is the daily temperature, $E$ and ${E_{\text{p}}}$ denote the actual and potential evaporation, respectively, and λ is the latent heat of vaporization.

The $e{^{^{\prime}}}$ is represented by $H^{\prime} - {H_{\text{p}}}^{\prime}$ and denotes the effect of soil moisture deficits on the surface energy balance. When soil moisture is sufficient to meet atmospheric demand (energy-limited region), the actual evaporation equals the potential evaporation, and $e{^{^{\prime}}}$ will be zero. As a soil moisture deficit gradually appears, the actual evaporation decreases significantly, but the potential evaporation remains constant with the atmospheric evaporation demand and $e{^{^{\prime}}}$ becomes positive. Hence, the magnitude of $\,e{^{^{\prime}}}$ can indicate the contribution of soil moisture deficiency to sensible heat flux. Only if the contribution of soil moisture deficiency to sensible heat ($e{^{^{\prime}}}$) is accompanied by large anomalous temperature values, the local energy balance may be controlling air temperature with large values of π. This diagnostic has been widely used in previous heatwave studies (Miralles et al 2014, Liu et al 2020, Geirinhas et al 2022, Seo and Ha 2022).

Figure 1 shows the spatial distributions of anomalies in temperature, soil moisture, and geopotential height during July and August of 2022. Under the control of the persistent WPSH, anomalously high temperature and exceptionally low soil moisture emerged over eastern China in July and peaked in August over the YZB (103°–123° E, 25°–35° N) (figures 1(a)–(c)). For instance, temperature anomalies exceeded 8 °C for nearly 10 d (9–18 August) in the Sichuan Basin, accompanied by extremely low soil moisture (figure S1). With the evolution of heat conditions, the anomalies of 3 d mean YZB temperature, soil moisture, and geopotential height in August 2022 exceeded 3, 5, and 2 standard deviations, respectively (figure 1(d)). Simultaneous heatwaves and droughts, which were typical CHDEs, covered nearly 80% of the study area in mid-August (figures S1 and S2). Notably, the geopotential height anomalies were relatively stable after the YZB-mean temperature became extremely high (exceeding 1.5σ) on 4 August, while temperature and soil moisture (negative) continued to increase for approximately 10 d and peaked in mid-August.

Zoom In Zoom Out Reset image size

High geopotential height, high net radiation, and dry soil are common drivers of extreme temperatures. Compared to temperature and soil moisture anomalies that exceeded their historical (1982–2021) ranges, the anomalies of geopotential height and net surface radiation anomalies during this event were relatively less extreme (figures S3(a) and (d)). Multiple linear regressions, incorporating the simultaneous anomalies of YZB-mean temperature against soil moisture and geopotential height or surface net radiation, show that nonlinear temperature amplifications maximized during the peak of the event (15–20 August) (figures S3(b) and (e)). The maximum temperature amplifications reached 1.6 °C–2 °C (2σ–3σ amplifications), which contributed to the total ∼5 °C temperature anomaly by ∼32%–40%. This indicated the nonlinear interactions between atmospheric circulation anomalies and land surface processes, which played an important role in amplifying this CHDE over the YZB.

To examine the interactions between atmospheric circulation anomalies and land surface processes, we analyzed the relationships between temperatures, soil moisture, and surface energy fluxes (figure 2). Generally, soil moisture has a negative correlation with temperature in the YZB (figure 2(a)). Although soil moisture continued to decrease from 29 July to 10 August, the temperatures were close to the climatology (figures 2(a)–(g)). However, an abrupt increase in temperature was observed during 10–13 August, and the soil stopped getting drier thereafter (figure 2(g)). In addition, the R² between temperature and net radiation was 0.789 in 2002–2021, which was better than that between temperature and soil moisture/geopotential height (figures 2(b) and S4). However, temperature continued to rise after net radiation anomalies peaked on 13 August and significantly deviated from the regression line. This evidence suggests that net radiation could not explain the variations in temperature and that soil moisture might play a dominant role during the CHDE period over YZB.

Zoom In Zoom Out Reset image size

Soil moisture significantly affects the land surface energy partitioning process. The surface net radiation can be distributed into latent heat flux (evaporative cooling) and sensible heat flux (heating near-surface air). Figure 2(f) shows that latent heat flux increased linearly with net radiation over the past four decades. Additionally, latent heat flux had a negative correlation with soil moisture (figure 2(e)), indicating that the YZB is a typical energy-limited region. However, in summer 2022, long-lasting temperature extremes continuously consumed soil water, leading to substantial water loss. Then, the regional mean latent heat flux decoupled from net radiation after early August, when soil moisture became extremely low (figure 2(f)). The decoupling between latent heat flux and net radiation suggested that the YZB experienced a shift from an energy-limited pattern to a water-limited pattern during the CHDE. This transition led to an immediate increase in sensible heat flux (figures 2(c) and (d)), which significantly heated the near-surface air and contributed to the unprecedented high temperatures in the following 10 d, coinciding with the severe drought in August over the YZB.

To quantify the local heating effect caused by the shift in soil moisture-temperature coupling, we applied $e{^{^{\prime}}}$ (see Methods) to describe the extra amount of heat flux transferred from latent heat to sensible heat due to enhanced soil moisture-temperature feedback. The increase in local sensible heat, which significantly contributes to local air temperature, is usually driven by the increase in net radiation or the increase in the proportion of net radiation allocated to sensible heat. The latter term can be reflected by $e{^{^{\prime}}}$. The magnitude of $e{^{^{\prime}}}$ is mainly determined by actual and potential evaporation. Potential evaporation is related to atmospheric evaporation demand, while actual evaporation is also limited by soil moisture availability (figure S5). Figure 3 stratifies different grid points of the YZB during the 2022 event into two soil moisture-temperature coupling modes. In the strong coupling regions (the grid cells with positive sensible heat anomalies and negative latent heat anomalies, orange dots), soil moisture had largely decreased and temperatures started to increase nonlinearly with soil moisture (figure 3(a)). While in the weak coupling regions (all other grid cells, blue dots), soil moisture decreased with slowly rising temperatures. Almost all dots with temperature anomalies higher than 8 °C were found in the strong coupling regions, and the regional mean $e{^{^{\prime}}}$ increased dramatically during 7–10 August just before the temperature peaked (figure S6). These results suggest that the excessive sensible heat induced by the shift in soil moisture-temperature coupling exacerbated the temperature extremes over the YZB.

Zoom In Zoom Out Reset image size

Moreover, during the 2022 CHDE, the extreme temperatures (temperature anomalies >8 °C) of different land grid cells had better consistency with $e{^{^{\prime}}}$ than with net radiation (figures 3(b) and (c)), which indicates that the spatial distribution of hot extremes was linked to the amount of excessive sensible heat produced by enhanced land–atmosphere interactions. Under the background of large-scale atmospheric circulation anomalies (figure 1(c)), almost the entire YZB experienced similar conditions of precipitation deficits and high net radiation (figure S7(e)). However, temperature anomalies in different regions varied greatly from less than 4 °C to over 8 °C in mid-August. This significant spatial heterogeneity of temperature extremes could not be adequately explained by circulation anomalies and net radiation anomalies but fit the spatial distribution of $e{^{^{\prime}}}$ very well (figures 3(d) and S7(f)–(i)). In addition, the previous local water storage played an important role in this process. Regions with insufficient water storage were more likely to be drained and shifted to a water-limited pattern with stronger land–atmosphere feedback during a prolonged CHDE, ending with more extreme temperatures (figures 3(a) and S7).

As a consequence of global warming, the PDFs of regional temperature and soil moisture show that the YZB climate in July and August tended to be hotter and drier in the latter 20 years (2002–2021) (figure 2). Seo and Ha (2022) noted that hotter and drier conditions can largely enhance local land–atmosphere interactions, raising the likelihood of CHDEs over northern East Asia, which is a typical water-limited region. While in the humid region YZB, to identify the impacts of background climate change on land–atmosphere interactions and CHDEs, we further examined the trends of temperature, precipitation, soil moisture, $e{^{^{\prime}}}$, and the frequency and heatwave intensity of CHDEs (see methods) during the past 40 years (1982–2021). Figures 4(a)–(c) show that temperature had a significant increasing trend and precipitation had a decreasing trend, accompanied by reductions in soil moisture over the YZB. The change in climatology led to an increase in CHDEs (figures S8(a) and (c)). However, there was still a significant increase in the frequency and heatwave intensity of CHDEs even if we have removed the background climate trends of temperature and soil moisture plotted in figures 4(a) and (c), which indicates the nonlinear interactions between temperature and soil moisture under extreme conditions (figures 4(e) and (f)). Moreover, the trends of CHDEs frequency and heatwave intensity calculated by the temperature and soil moisture without the climate trends shared similar spatial distributions with the trend of $e{^{^{\prime}}}$ (figure 4(c)), which implies that the enhanced land–atmosphere interaction may be related to the increasing frequency and intensity of CHDEs. Notably, the trends of these variables were significant in the Sichuan Basin, which was also a hotspot in the 2022 CHDE. Therefore, this area (104°−108° E, 28°−32° N, outlined by black rectangles in figures 4(a)–(f)) was chosen as a typical region to explore the relationship between land–atmosphere feedback and compound events.

Zoom In Zoom Out Reset image size

To further investigate the contribution of land–atmosphere feedback to CHDEs, we compared temperature, soil moisture, $e{^{^{\prime}}}$, and net radiation during CHDEs, hot-only events, and dry-only events in the typical region during July and August of 1982–2021 (figure S9). Figures 4(e)–(h) show that net radiation anomalies were similar during CHDEs and hot-only events. However, CHDEs had much higher temperatures and lower soil moisture. Almost all temperature anomalies higher than 8 °C were found in CHDEs. The extremely strong land–atmosphere feedback, represented by the highest $e{^{^{\prime}}}$ in CHDEs, might contribute to the amplification and prolongation of CHDEs. While dry-only events had the lowest net radiation and temperature, although their land–atmosphere feedback was relatively strong.