Mesoscale soil moisture heterogeneity can locally amplify humid heat

TL;DR

This study demonstrates that mesoscale soil moisture heterogeneity can locally amplify humid heat by about 1–4°C on 10–20 km scales, with a maximum when the soil-moisture length-scale is near a critical value λ_c ≈ 50 km. Using a convection-resolving, 500 m UM–JULES model with imposed circular wet patches (25–150 km), the authors uncover a soil-moisture–driven mesoscale circulation that reduces boundary-layer depth and concentrates warm, humid air over the wet area, yielding enhanced Twb and HI. The amplification is sensitive to background wind and the amplitude of the wet–dry contrast, and persists across different humid-heat metrics, indicating a robust mechanism linking soil moisture patterns to extreme humid heat. These findings suggest that high-resolution soil moisture observations (e.g., ASCAT) could improve near-term predictions of hazardous humid-heat hours at city and county scales in tropical regions, highlighting the need for models to resolve 10–100 km SM heterogeneity.

Abstract

Soil moisture is a key ingredient of humid heat through supplying moisture and modifying boundary layer properties. Soil moisture heterogeneity due to e.g., antecedent rainfall, can strongly influence weather patterns; yet, its effect on humid heat is poorly understood. Idealized numerical simulations are performed with a cloud-resolving ($Δx$=500 m), coupled land-atmosphere model wherein wet patches on length-scales $λ\in$ 25-150 km are prescribed. Compared to experiments with uniform soil moisture, humid heat is locally amplified by 1-4$^\circ$C, with maximum amplification for the critical soil moisture length-scale $λ_c=$ 50 km. Subsidence associated with a soil moisture-induced mesoscale circulation concentrates warm, humid air in a shallower boundary layer. The background wind and the magnitude of the wet-dry contrast control the relationship between $λ_c$ and the humid heat amplification. Based on observed soil moisture patterns, these results will help to predict extreme humid heat at city and county scales across the Tropics.

Figures (4)

• Figure 1: Daily mean (a) $T$, (b) $q$, (c) Twb, and (d) $z_i$ in P50. In (c) the daily mean 10-meter wind is shown with arrows. Insets show spatially averaged evolution of 3-hourly values: green, blue and black lines are domain averages for UDRY, UWET, and P50, respectively; orange line is wet patch average in P50 (see legend). The dashed green circle in (c), with radius $r=$10 km, indicates an area used in Section \ref{['sec:results']}.2.
• Figure 2: (a) Diurnal cycle of 3-hourly Twb in perturbed experiments (color lines, see legend), UDRY (dashed line), and UWET (dotted-dashed line). In P$\lambda$, wet patch averages are considered whereas in UDRY and UWET the domain average Twb is plotted. (b) Maximum 3-hourly Twb values averaged over circular areas centered at x$_0$=200 + $\lambda$/2 km and of radius $r$ (in km, horizontal axis) for all perturbed experiments and UWET (see legend in (a)).
• Figure 3: (a) Same as Fig. \ref{['fig:fig2']}a for $z_i$ (see legend in (b) for colors). (b) 3-hourly Twb at 1900 LT as a function of 1500--1800 LT average 3-hourly BL height in perturbed experiments (dots); both quantities are averaged over a circular area of radius $r$=10 km centered on x=x$_0$ + $\lambda/2$, as inferred in the previous section. The corresponding domain mean values in UWET are shown with dotted-dashed lines. The inset in the bottom right corner shows the R$^2$ of the linear regression between $\langle$Twb$_{max}\rangle_{10}$ sampled at each daytime hour and $\overline{z}$ averaged over the four preceding hours; significant regressions (at the 1% level) are indicated with circles. (c) Vertical cross section in the longitudinal direction at 1500 LT in P50. Potential wet-bulb temperature anomaly ($\theta_{wb}'$, in $^\circ$C) is shown with shading and wind anomaly (in m s$^{-1}$) with arrows (anomalies are calculated as P50-UWET). Solid lines denote domain-average $z_i$ in UDRY (green), UWET (blue), and P50 (black). (d) Near-surface $\theta_{wb}$ at 1500 LT, with colors similar to (a). Longitudinal cross section values (panels c and d) are averaged along a 20 km band centered at y$_0$=200 km.
• Figure 4: Wet patch averaged maximum Twb anomaly ($\langle \text{Twb'}_{max}\rangle _{\lambda}$, Twb' = Twb$_{\text{P}\lambda}$-$\text{Twb}_{\text{UWET}}$; y-axis) for different values of soil moisture length-scale ($\lambda$; x-axis) in various sensitivity experiments (see Table 1): (a) $U-$ and $U+$, (b) $\delta-$ and $\delta+$, (c) $\gamma_{\theta}-$ and $\gamma_{\theta}+$, and (d) $\gamma_{q}-$ and $\gamma_{q}+$. Within-patch spatial variability is shown with error bars corresponding to $\pm\sigma$ about the spatial mean. Average spatial variability across UWET experiments is shown with grey shading corresponding to $\pm\sigma$ about the zero line.