Evaluating Precipitation Features and Rainfall Characteristics in a Multi‐Scale Modeling Framework

Abstract

Cloud and precipitation systems are simulated with a multi‐scale modeling framework (MMF) and compared over the Tropics and Subtropics against the Tropical Rainfall Measuring Mission (TRMM) Radar‐defined Precipitation Features (RPFs) product. A methodology, in close analogy to the TRMM RPFs, is developed to produce simulated precipitation features (PFs) from the output of the embedded two‐dimensional (2D) cloud‐resolving models (CRMs) within an MMF. Despite the limitations of 2D CRMs, the simulated population distribution, horizontal and vertical structure of PFs, and the geographical location and local rainfall contribution of mesoscale convective systems (MCSs) are in good agreement with the TRMM observations. However, some model discrepancies are found and can be identified and quantified within the PF distributions. Using model biases in relative population and rainfall contributions, PFs can be characterized into four size categories: small, medium to large, very large, and extremely large. Four different major mechanisms might account for the model biases in each different category: (1) the two‐dimensionality of the CRMs, (2) a positive convection‐wind‐evaporation feedback loop, (3) an artificial dynamic constraint in a bounded CRM domain with cyclic boundaries, and (4) the limited CRM domain size. The second and fourth mechanisms tend to contribute to the excessive tropical precipitation biases commonly found in most MMFs, whereas the other mechanisms reduce rainfall contributions from small and very large PFs. MMF sensitivity experiments with various CRM domain sizes and grid spacings showed that larger domains (higher resolutions) tend to shift PF populations toward larger (smaller) sizes.

Representing Cloud Mesoscale Variability in Superparameterized Climate Models

“…So also in the case of using a LEM with prognostic cloud condensate, dedicated choices need to be made how to spatially distribute the advected liquid water into the local LEM. The Goddard multiscale modeling framework (Chern et al, 2016(Chern et al, , 2020Tao et al, 2009) for instance, is using a LEM with prognostic cloud condensate.…”

“…So also in the case of using a LEM with prognostic cloud condensate, dedicated choices need to be made how to spatially distribute the advected liquid water into the local LEM. The Goddard multiscale modeling framework (Chern et al., 2016, 2020; Tao et al., 2009) for instance, is using a LEM with prognostic cloud condensate. In their framework the large scale advected cloud condensate is only added to saturated grid points of the LEM and proportionally to the already existing cloud amount in each grid point, very similar to the humidity variance coupling proposed in this study.…”

“…SP should be viewed as an attempt for a better and model-informed parameterization, rather than a substitute for a global LEM. Chern et al (2020) evaluates biases in the size of rain events simulated in the Goddard MMF, caused by the size constraint imposed by the cloud-resolving model. On the other hand, other phenomena such as mesoscale convective systems and tropical cyclones have been shown to be able to grow across the scales of the two models when simulated with SP (Lin et al, 2022;Pritchard et al, 2011;Tulich, 2015).…”

Characteristics Analysis of Precipitation Observations and Climate Change Along the High-speed Railway in Jiangsu

“…We look for the characteristics of precipitation along the railway line from the air pressure, temperature and precipitation data monitored on the ground, and discuss the relationship between air pressure and temperature and the amount of precipitation produced by precipitation levels. As such, the production of extreme precipitation involves more multi-scale dynamics and microphysical processes (Chern et al, 2020;Endo and Kitoh, 2014). How to build a bridge between ground observations and physical processes requires further research.…”

Relating Vertical Velocity and Cloud/Precipitation Properties: A Numerical Cloud Ensemble Modeling Study of Tropical Convection

“…A simple hail size mapping (Tao et al., 2016) eliminates the need to select a hail intercept a priori and produced peak reflectivity profiles dominated by hail in good agreement with NEXRAD observations over a several‐hour period for an intense squall line observed over Oklahoma during the Midlatitude Continental Convective Clouds Experiment field campaign (Jensen et al., 2016). The 4ICE scheme has been used to generate the GCE database for the Goddard Convective‐Stratiform Heating algorithm (S. E. Lang & Tao, 2018) and Japan Spectral Latent Heating algorithm (Shige et al., 2009), study the land‐ocean contrast in tropical convection (Matsui et al., 2016), compared against the SBM scheme for intense convection (Matsui et al., 2019), and evaluated globally in the Goddard Multi‐scale Modeling Framework (Chern et al., 2016, 2020; Tao & Chern, 2017). Additional characteristics of Goddard 4ICE scheme are included in Text S1 in Supporting Information .…”