Yunji Zhang

Cloud-radiative feedback plays a critical role in climate and the formation of extreme-weather-producing systems such as tropical cyclones (TCs; e.g., hurricanes). Yet, the convective-scale processes by which this feedback promotes the upscale growth of tropical convection, including TC formation, remain poorly understood. Here, we leverage multiple numerical modeling frameworks to show that longwave warming by stratiform and anvil clouds promotes upscale development by suppressing downdrafts. As byproducts of deep convection, these cloud systems are responsible for the majority of evaporatively-cooled downdrafts in tropical mesoscale convective systems (MCSs), which help terminate convection by stabilizing the troposphere. We show, however, that their radiative warming is balanced by ascending motion that moistens the column and suppresses these downdrafts, therefore promoting convective maintenance and upscale growth by increasing precipitation efficiency (ε). We demonstrate this feedback through cloud-radiative feedback mechanism-denial tests in three distinct convection-permitting model frameworks, including TC case studies, an aquaplanet, and non-rotating radiative-convective equilibrium (RCE). When clouds and longwave radiation interact, moist convection exhibits greater positive trends in area coverage and mean convective-system size. This feedback promotes more frequent TC development across intensity thresholds in the aquaplanet framework. These findings underscore the active role of stratiform clouds in tropical convective evolution through the link between their longwave radiative warming and ε. They help explain why expanding stratiform raining area in nascent TCs can signal imminent genesis and intensification, and why model predictions of radiatively-driven tropical systems such as the Madden–Julian Oscillation strongly hinge on the representation of a robust stratiform regime.

Ice crystals in tropical cyclones (TCs) remain infrequently observed and difficult to model using Eulerian microphysics parameterizations. To better understand TC eyewall ice crystal processes, the Ice Crystal Trajectory Growth (ICTG) model, previously applied to an idealized squall line, was used to simulate Lagrangian ice crystal trajectories in a Weather Research and Forecasting (WRF) model simulation of Hurricane Harvey (2017). ICTG includes laboratory‐informed growth equations capable of simulating complex crystal morphologies and is used to assess how initial altitude, radius, and size influence crystal growth and advection. We introduce an initialization technique where the initial crystal locations and sizes best match WRF microphysical fields. Results showed that the initial crystal altitude was most important in determining final mass, shape, location, and likelihood of sublimation; initial size and radial distance played secondary but noticeable roles. The largest final masses began at 8 km altitude and grew predominantly in the dendritic growth zone (−12 to −18C) resulting in plate‐like crystals. Crystals that accumulated the most rime mass began between 7 and 9 km altitude, while crystals above 9 km experienced more radial outward transport often leading to sublimation particularly for smaller crystals. Results using a “super‐particle” method showed that most of the ice mass reached the melting level within or just beyond the eyewall with a large diversity of crystal shapes and sizes. The greatest radial extent of ice was 150 km, supporting recent research suggesting the eyewall is not a dominant source of TC outflow condensate.

Cloud‐radiative forcing (CRF) has been suggested to accelerate tropical cyclone (TC) genesis, but we do not yet understand the role of convective‐scale processes in this cloud‐radiative feedback. We use a convection‐permitting ensemble Weather Research and Forecasting model framework to examine the hypothesis that CRF within stratiform cloud regions weakens downdrafts, allowing the environment to moisten more easily. We specifically compare our control simulations (CTL) of TC development to sensitivity tests that exclude cloud‐radiative forcing (NCRF) either everywhere or just within specific cloud types. Our experiment and analysis indicate that CRF leads to fewer and weaker stratiform downdrafts and greater humidity and moist entropy in the developing TC core, implying suppressed ventilation, with stratiform and anvil CRF dominating this effect. This cloud‐radiative feedback accelerates TC development by promoting faster intensification of both the mid‐level vortex and surface cyclone.

Cloud-radiative forcing (CRF) has been suggested to accelerate tropical cyclone (TC) genesis, but we do not yet understand the role of convective-scale processes in this cloud-radiative feedback. We use a convection-permitting ensemble Weather Research and Forecasting (WRF) model framework to examine the hypothesis that CRF within stratiform cloud regions weakens downdrafts, allowing the environment to moisten more easily. We specifically compare our control simulations (CTL) of TC development to sensitivity tests that exclude cloud-radiative feedback (NCRF) either everywhere or just within specific cloud types. Our experiment and analysis indicate that CRF leads to fewer and weaker stratiform downdrafts and greater humidity and moist entropy in the developing TC core, implying suppressed ventilation, with stratiform and anvil CRF dominating this effect. This cloud-radiative feedback accelerates TC development by promoting faster intensification of both the midlevel vortex and surface cyclone.

This study uses the airborne Doppler radar data set, TC‐RADAR, and model simulations from the Penn State Ensemble Kalman Filter data assimilation system to examine the dynamics and evolution of tropical cyclone (TC) rainband convective updrafts. This examination utilizes an updraft selection algorithm and statistical analysis of convective updraft characteristics previously developed based on observations of TC rainband convective updrafts. The selected updrafts are collectively analyzed by their spatial frequency, radius, azimuthal location (relative to the environmental 200–850 hPa wind shear and environmental 850 hPa mean flow), structural characteristics, and secondary circulation (radial/vertical) flow pattern. The observed updrafts compare favorably with the simulated updrafts. A wavenumber‐1 asymmetry is found, showing that convective updrafts in the downshear quadrants of the TC are more frequent. The radial flow of updraft circulations aligns with the prevailing vortex‐scale radial flow, which is governed by the environmental shear or low‐level mean flow‐induced asymmetry. Convective‐scale circulations are hypothesized to be significantly influenced by vortex‐scale radial flow at the updraft base and top altitudes. The bottom‐up decay of aging convective updrafts, caused by increased low‐level downdrafts, affects the base altitude and the subsequent radial flow of the updraft circulation. The findings presented in this study support previous literature regarding observed convective‐scale patterns of organized rainband convection in a mature TC.

Satellites provide the majority of observations of tropical cyclones (TCs) when they are over the open ocean. These observations are under‐utilized in operational global and regional forecasting models as most of the cloudy and precipitation‐affected observations are discarded. Recent works have shown that assimilating all‐sky infrared (IR) radiances from geostationary satellites improves TC track, intensity, and rainfall forecasts. This study examines how two different cloud microphysical parametrization schemes impact all‐sky IR radiance assimilation and subsequent forecasts of Hurricane Harvey. Because model hydrometeor fields that are used to simulate IR radiances are strongly constrained by the model microphysics scheme, Harvey's dynamic and thermodynamic structure become very different when changing the microphysics schemes, which also leads to changes in subsequent intensity forecasts. Sensitivity forecasts suggest that intensity forecast errors in this study are primarily controlled by uncertainties in initial conditions, which are influenced by microphysics schemes as they simulate model priors differently when assimilating all‐sky IR radiance. On the other hand, the microphysics schemes used during the free forecasts generally have a smaller impact on Harvey's intensity forecast errors.

Satellite infrared (IR) and microwave (MW) radiances provide complementary information on tropical cyclones (TCs), with IR radiances informing on cloud locations, sizes, and cloud-top heights, and MW radiances conveying information on hydrometeor phase, amount, and location below cloud tops. This study utilizes real-data experiments to investigate the impact of assimilation of MW all-sky radiances, in addition to IR radiances, on four major TCs that impacted the U.S. during the 2017 Atlantic hurricane season. Furthermore, impacts of the microphysics parameterization scheme (MPS) are assessed with two schemes, the Weather and Research Forecast model single-moment-6-class (WSM6) MPS and the Thompson et al. (2008; THO) hybrid MPS. These MPSs lead to little difference in data assimilation analyses and forecasts when only conventional observations are assimilated. However, when satellite all-sky IR radiances are assimilated, a significant portion of analyzed TCs in the WSM6 experiments are notably stronger whereas most of the TCs in the THO experiments are weaker than best track. This discrepancy is closely linked to differences in cloud-top features produced by the two MPSs and the impact of these differences on the calculated IR radiances. However, regardless of the MPS used, the additional assimilation of MW radiances improves forecast performance, especially on intensity, compared to when they are not assimilated.

The structure of the planetary boundary layer (PBL) is important for the initiation, development, and organization of convection. High-spatiotemporal-resolution networks that directly observe the PBL structure are currently unavailable. Recent studies discovered that differential reflectivity ( Z _DR ) observations from dual-polarization Doppler weather radars in clear-air conditions can be used to characterize the top of the daytime PBL. Compared with other observational platforms that observe the PBL, these Z _DR -derived PBL depth observations have high temporal resolution and relatively dense and uniform distributions over the CONUS. Therefore, assimilating these observations could potentially improve the estimation and forecast skill of thermodynamic structures in the lower troposphere. This study examines the impact of assimilating Z _DR -derived PBL depth observations on the forecasts of the torrential rainfall and flash flood event in eastern Kentucky on 27–28 July 2022 using a strongly coupled land–atmosphere data assimilation system. The model configuration in the experiments mimics the operational HRRR. Results show that assimilating Z _DR -derived PBL depth observations leads to considerable changes in temperature and moisture in the lower troposphere. Soil conditions, including soil moisture and associated surface heat fluxes, are also modified. The assimilation of Z _DR -derived PBL depth observations contributes to a better match between model-diagnosed PBL depth with the observations. Subsequently, rainfall forecasts are statistically significantly improved using both gridwise and neighborhood metrics, especially for the most extreme rainfall. Sensitivity experiments also show that the assimilation frequency and the observation errors assigned to Z _DR -derived PBL depth observations influence the performance of the rainfall forecasts, which deserve future study. Significance Statement: This is the first study that presents a comprehensive investigation on the impact of assimilating planetary boundary layer (PBL) depth observations estimated from dual-polarization NEXRAD radars on the convection-allowing analyses and forecasts of severe weather events. The torrential rainfall and flash flood event that occurred in eastern Kentucky on 27–28 July 2022 is utilized as a case study. Assimilating radar-derived PBL depth observations leads to substantial changes in lower-tropospheric thermodynamic conditions, instability, surface heat fluxes, and soil states. As a result, rainfall forecasts of this event were significantly improved. Since the dual-polarization radar observations used to estimate PBL depth are available over most of the CONUS every 5–10 min, this study demonstrates the great value of these novel observations in data assimilation and numerical weather prediction practices.

In this study, we assimilated microwave (MW) all‐sky radiances from low‐Earth‐orbiting satellites and examined their impact on the analyses and forecasts of weather hazards associated with the 10 August 2020 Midwest derecho. Compared with the baseline that assimilated conventional surface and upper‐air observations and infrared (IR) all‐sky radiances from geostationary satellites, the addition of MW all‐sky radiances improved the analyzed and forecasted convection‐stratiform structures of the derecho. Results show that MW all‐sky radiances provided additional information, compared with IR radiances, on hydrometeors within the storm, leading to improved forecasts out to 2 hr with quantitatively more accurate surface gusts. This is the first study to assimilate MW all‐sky radiances for a severe weather event using a convection‐permitting numerical weather prediction model (our model resembles NOAA's High‐Resolution Rapid Refresh), and the results suggest promising avenues for improving severe weather forecasts worldwide in the future.