The representation of Convectively Coupled Equatorial Waves and upscale energy transfer in models with explicit and parametrized convection

TL;DR

This study investigates how the representation of convection affects convectively coupled equatorial waves (CCEWs) and upscale energy transfer in km-scale tropical simulations. Using a local interscale energy transfer diagnostic based on Duchon–Robert theory and 2D surface filtering, the authors compare two Met Office UM configurations on a full tropical channel: GAL9 with convection parametrization and RAL3 convection permitting, for the DYAMOND Summer 2016 period. They find that convection-permitting RAL exhibits about $>1$–fold stronger upper-tropospheric upscale transfer, more coherent CCEWs, and higher correlations between interscale transfer and equatorial wave winds, with Kelvin and Rossby waves linked to transfers from scales $2$–$8$ times smaller than the dominant wavelength. The results indicate explicit convective representation markedly enhances scale coupling in the tropical atmosphere, with implications for improving tropical weather forecasting and the faithful simulation of convective coupling within CCEWs.

Abstract

Convectively Coupled Equatorial Waves (CCEWs) dominate atmospheric variability on timescales of 2--30 days in the Tropics, bringing episodes of widespread heavy precipitation. This study compares the representation of CCEWs and their connection to upscale energy transfer in two Met Office Unified Model simulations of the full tropical channel with identical km-scale resolution. The principal difference between the simulations is that one parametrizes convection (GAL), while the other (RAL) is convection permitting. This means GAL acts to remove vertical instability without explicitly representing the resolved-scale circulation associated with convective plumes. We present the first quantitative diagnosis of interscale energy transfer and its relation to CCEWs. This diagnosis is important because upscale energy transfer between convection and large-scale waves may influence accurate simulation and predictability of tropical weather systems. The average upper-tropospheric upscale transfer simulated by RAL is approximately 50\% higher than GAL. CCEWs are more coherent in RAL, with an average phase-speed variability 80\% higher than observations, compared with 166\% higher in GAL. RAL also simulates greater upscale energy transfer within waves than GAL, with a stronger correlation between the interscale energy transfer rate and equatorial wave winds. Simulated Kelvin and Rossby waves are associated with upscale energy transfer from scales 2--8 times smaller than the dominant wavelength, related to active deep convection within a particular sector of the wave phase. Our findings show that the explicit representation of convective motions has a significant impact on the simulation of upscale energy transfer, and is therefore very likely to be a significant factor in the faithful simulation of the convective coupling within CCEWs.

Figures (9)

• Figure 1: Snapshot of interscale energy transfer across a length scale of 440 km at 200hPa for the tropics 15S-20N from the RAL3 N2560 model field on 2016-08-02T0000Z. The three panels represent three different resolutions used in this study: N2560 is the native resolution of the model data; N1280 represents an interpolation to halve the horizontal resolution and quarter the computational expense; 0.5 deg represents the coarsest data used in the study.
• Figure 2: Hovmöller (longitude-time) diagrams of precipitation (shading) and equatorial wave winds at 850hPa - Kelvin (blue contours), $n=1$ equatorial Rossby (R1) (black contours), WMRG (red contours). Quantities averaged between 10S-10N: precipitation in mm h$^{-1}$, $u_{KW}$ at $\pm$ 1 m s$^{-1}$, $u_{R1}$ and $\upsilon_{WMRG}$ at $\pm$ 2 m s$^{-1}$. Dashed contour lines indicate negative values. GPM precipitation and UM analysis waves in (a) (the observational data set); GAL9 N2560 in (b), and RAL3 N2560 in (c). Quantities shown for the first 20 days of the DYAMOND Summer simulation.
• Figure 3: Longitude-height slice of wave standard deviation in time for Kelvin (a) - (c), R1 (d) - (f), and WMRG (g) - (i) waves for UM analysis (left column), GAL9 N2560 (middle column), and RAL3 N2560 (right hand column). The standard deviations are calculated on 10S-10N equatorially-averaged Kelvin wave $u$, R1 $u$ and WMRG $\upsilon$.
• Figure 4: Maps of time-mean $\mathcal{D}_L(\mathbf{u})$ for $L$ = 440 km averaged over the DYAMOND Summer simulation period for (a) GAL9 N2560 and (b) RAL3 N2560. Blue represents upscale energy transfer, and red downscale. The maps are for the full tropical channel from 20S to 20N. The average from 15S to 15N (indicated by the dotted lines) is on the top right of each panel, colour-coded to indicate whether the mean transfer in that region is upscale or downscale. Variance in $\mathcal{D}_L(\mathbf{u})$ over the same region indicated by the value in brackets. (c) mean GPM-IMERG precipitation for the same time period.
• Figure 5: Longitudinally averaged latitude-pressure plots of time-mean $\mathcal{D}_L(\mathbf{u})$ for $L$ = 440 km (shading, zero contour thick grey line) for the domain shown in Fig. \ref{['fig:Dlu_tm']}. Zonal mean zonal wind in black contours in intervals of 1 m s$^{-1}$, and zonal mean meridional and vertical velocities in vectors, with the reference arrow of 0.1 m s$^{-1}$ shown in the top right of the figure. (a): for the GAL9 N2560 model; (b): for RAL3 N2560. $w$ has been re-scaled by the aspect ratio of the domain for better visualisation of the time-mean flow.