Improving Convection Trigger Functions in Deep Convective Parameterization Schemes Using Machine Learning

Abstract

Convection is critical to precipitation, heat and moisture transport, cloud amount and distribution, as well as to the global energy budget (Arakawa, 2004;Arakawa & Schubert, 1974). Properly representing convection is critical to successful numerical weather prediction (NWP) and climate simulations by general circulation models (GCMs). However, because current coarse-resolution models cannot explicitly resolve convection, it must be represented with convective parameterizations. Deficiencies in these parameterizations Abstract Deficiencies in convection trigger functions, used in deep convection parameterizations in General Circulation Models (GCMs), have critical impacts on climate simulations. A novel convection trigger function is developed using the machine learning (ML) classification model XGBoost. The large-scale environmental information associated with convective events is obtained from the longterm constrained variational analysis forcing data from the Atmospheric Radiation Measurement (ARM) program at its Southern Great Plains (SGP) and Manaus (MAO) sites representing, respectively, continental mid-latitude and tropical convection. The ML trigger is separately trained and evaluated per site, and jointly trained and evaluated at both sites as a unified trigger. The performance of the ML trigger is compared with four convective trigger functions commonly used in GCMs: dilute convective available potential energy (CAPE), undilute CAPE, dilute dynamic CAPE (dCAPE), and undilute dCAPE. The ML trigger substantially outperforms the four CAPE-based triggers in terms of the F1 score metric, widely used to estimate the performance of ML methods. The site-specific ML trigger functions can achieve, respectively, 91% and 93% F1 scores at SGP and MAO. The unified trigger also has a 91% F1 score, with virtually no degradation from the site-specific training, suggesting the potential of a global ML trigger function. The ML trigger alleviates a GCM deficiency regarding the overprediction of convection occurrence, offering a promising improvement to the simulation of the diurnal cycle of precipitation. Furthermore, to overcome the black box issue of the ML methods, insights derived from the ML model are discussed, which may be leveraged to improve traditional CAPE-based triggers.Plain Language Summary Deficiencies in convection trigger function, a set of conditions used to determine whether the convection will be activated at a given time in General Circulation Models (GCMs), have critical impacts on model simulated climate. This work presents a novel convection trigger function using a machine learning (ML) model. Environmental information on convective events are obtained from long-term data from the Atmospheric Radiation Measurement (ARM) program at its Southern Great Plains (SGP) and Manaus (MAO) sites, which represent two distinct convective regimes. The ML trigger is separately trained and evaluated per site, and jointly trained and evaluated at both sites as a unified trigger. The ML trigger substantially outperf...

ResU‐Deep: Improving the Trigger Function of Deep Convection in Tropical Regions With Deep Learning

“…However, this also increases the complexity of the existing convective trigger schemes, which are single‐column‐oriented. The machine learning methods, such as fully‐connected neural networks (NN) and tree‐based models, are suitable for simulating single‐column data and have shown promising improvement in predicting the deep convection in small regions or at a single point (Ukkonen & Mäkelä, 2019; P. Wang et al., 2022; T. Zhang et al., 2021).…”

“…Since the diurnal variation of deep convection has been extensively studied from rainfall observation, we use the precipitation rate obtained from TRMM 3B42 to estimate the occurrence of deep convection in tropical regions (Gray & Jacobson, 1977; Janowiak et al., 1994). When the precipitation rate is greater than or equal to 0.5 mm/hr at each grid, the grid is labeled as deep convection occurring at that time (Song & Zhang, 2017; Suhas & Zhang, 2014; T. Zhang et al., 2021). The occurrence or non‐occurrence of deep convection in a given time interval at each grid cell is the predictand of our model.…”

“…Recently, machine learning methods have shown great potential in the field of geoscience in simulating, modeling, predicting, and correction (Bonavita & Laloyaux, 2020; Dramsch, 2020; Ham et al., 2019; Ravuri et al., 2021; Shi et al., 2015; Weyn et al., 2019; T. Zhang et al., 2019). Since it can process volumes of data and model the complex non‐linear relationship between different variables, machine learning has been applied to the development of GCMs and the parameterization schemes for improving simulation accuracy (Gentine et al., 2018; Han et al., 2020; O'Gorman & Dwyer, 2018; Rasp et al., 2018; Scher, 2018; Ukkonen & Mäkelä, 2019; P. Wang et al., 2022; T. Zhang et al., 2021). Among them, Ukkonen and Mäkelä (2019) used the ECMWF atmospheric reanalysis (ERA5) and variables from climate models to train the convective trigger functions based on the logistic regression, decision tree and fully‐connected neural network in Europe and Sri Lanka.…”

Artificial Intelligence for Earth System Predictability (AI4ESP) (2021 Workshop Report)

“…Multiple studies have highlighted the importance and usefulness of interpretable AI (IAI), explainable AI (XAI), feature/signature detection, and causal inference techniques in climate and weather science Tao Zhang et al 2021). These methods can be used to identify indicator patterns of forced changes and emergent properties of the real and simulated climate system.…”

A Neural Network‐Based Scale‐Adaptive Cloud‐Fraction Scheme for GCMs

“…For example, the feedforward neural network (also called multiple‐layer perceptron, hereafter denoted as neural network for short) is good at emulating complex functions and can be used to replace certain modules by either accounting for poorly‐understood processes or saving computational cost. It has been used to parameterize processes such as cloud cover (Grundner et al., 2022), radiation (Krasnopolsky & Fox‐Rabinovitz, 2006), convection (Han et al., 2020; Rasp et al., 2018; T. Zhang et al., 2021), and boundary‐layer turbulence (J. Wang et al., 2019). All yield promising outcomes.…”