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

Modeling deep convection accurately in tropical regions is important. However, biases remain in current trigger functions. To alleviate the overestimation of frequency and wrong depiction of the diurnal cycle, we propose a deep convection trigger function, ResU‐Deep, based on the framework of U‐net with three modifications to better suit the problem of deep convection identification: (a) adding the upsampling process into the encoder part, (b) replacing the double convolution block with a residual‐convolutional block, and (c) adding a dynamic weight into the loss function. Thirty‐three environmental variables within tropical regions are used in ResU‐Deep, including 31 features from ECMWF atmospheric reanalysis (ERA5) data set, and two historical convection fields. Tropical Rainfall Measuring Mission 3B42 data set is used as the precipitation observation. Central America, North Africa, South and East Asia, and West Pacific Ocean within 0°∼30°N are selected as the study regions for the high frequency of deep convection activities. ResU‐Deep, incorporating the surrounding information, is separately trained and evaluated in four regions and has the F1‐scores of 58%, 53%, 60%, and 63% for the occurrence, outperforming the single‐column‐based machine learning methods. Also, a unified model has similar performance in four regions. Further comparisons are made with convective available potential energy‐based trigger functions in Southern Great Plains. Results show that ResU‐Deep can capture the trends and peaks of diurnal cycles on complex terrains in large regions. According to feature importance test, the contribution levels of environmental features are different in four regions, indicating the model can learn the mechanisms of deep convection in specific region, thus improving the prediction accuracy. Plain Language Summary: The deep convection trigger function is a widely‐used method to determine the occurrence of deep convection in general circulation models. However, many trigger functions face the overestimation of deep convections and cannot accurately capture the variations of occurrence frequency. We propose a convection trigger function based on deep learning, called ResU‐Deep, to improve the trigger function in tropical regions. Three modifications are made based on the framework of U‐net. Both the architecture and the loss function are modified to improve the model's performance and increase the training speed. Thirty‐one environmental variables from ERA5 data set and two historical deep convection occurrence information are used as the features. Central America, North Africa, South and East Asia, and West Pacific Ocean within the tropics have high deep convection frequency. The model is trained and tested independently on each site, reflecting the non‐uniform physical mechanisms for deep convection. Also, a unified model has similar performance. Besides, incorporating surrounding information in ResU‐Deep can help improve prediction accuracy. To compare with the convective available potential energy‐based trigger functions, we test the model on Southern Great Plains in central United States. Results show that deep learning is a promising solution to identify deep convection. Key Points: A location‐aware and deep‐learning‐based deep convection trigger function is proposed to improve the diurnal cycle simulation in tropicsResults show that terrain information, temporal dependence of convection, and water vapor content are essential for predicting convectionIncluding information from neighboring atmospheric columns can improve the performance of the deep convection trigger function [ABSTRACT FROM AUTHOR]