arXiv subjects

Christoph Metzl

Publications and source records attributed to Christoph Metzl.

5 recordsLinked to original sources

arXiv2025

Physical Scales Matter: The Role of Receptive Fields and Advection in Satellite-Based Thunderstorm Nowcasting with Convolutional Neural Networks

The focus of nowcasting development is transitioning from physically motivated advection methods to purely data-driven Machine Learning (ML) approaches. Nevertheless, recent work indicates that incorporating advection into the ML value chain has improved skill for radar-based precipitation nowcasts. However, the generality of this approach and the underlying causes remain unexplored. This study investigates the generality by probing the approach on satellite-based thunderstorm nowcasts for the first time. Resorting to a scale argument, we then put forth an explanation when and why skill improvements can be expected. In essence, advection guarantees that thunderstorm patterns relevant for nowcasting are contained in the receptive field at long forecast times. To test our hypotheses, we train ResU-Nets solving segmentation tasks with lightning observations as ground truth. The input of the Baseline Neural Network (BNN) are short time series of multispectral satellite imagery and lightning observations, whereas the Advection-Informed Neural Network (AINN) additionally receives the Lagrangian persistence nowcast of all input channels at the desired forecast time. Overall, we find only a minor skill improvement of the AINN over the BNN when considering fully averaged scores. However, assessing skill conditioned on forecast time and advection speed, we demonstrate that our scale argument correctly predicts the onset of skill improvement of the AINN over the BNN after 2h forecast time. We confirm that, generally, advection becomes gradually more important with longer forecast times and higher advection speeds. Our work accentuates the importance of considering and incorporating the underlying physical scales when designing ML-based forecasting models.

physics.ao-ph↗

arXiv2025

Inferring Thunderstorm Occurrence from Vertical Profiles of Convection-Permitting Simulations: Physical Insights from a Physical Deep Learning Model

Thunderstorms have significant social and economic impacts due to heavy precipitation, hail, lightning, and strong winds, necessitating reliable forecasts. Thunderstorm forecasts based on numerical weather prediction (NWP) often rely on single-level surrogate predictors, like convective available potential energy and convective inhibition, derived from vertical profiles of three-dimensional atmospheric variables. In this study, we develop SALAMA 1D, a deep neural network which directly infers the probability of thunderstorm occurrence from vertical profiles of ten atmospheric variables, bypassing single-level predictors. By training the model on convection-permitting NWP forecasts, we allow SALAMA 1D to flexibly identify convective patterns, with the goal of enhancing forecast accuracy. The model's architecture is physically motivated: sparse connections encourage interactions at similar height levels while keeping model size and inference times computationally efficient, whereas a shuffling mechanism prevents the model from learning non-physical patterns tied to the vertical grid. SALAMA 1D is trained over Central Europe with lightning observations as the ground truth. Comparative analysis against a baseline machine learning model that uses single-level predictors shows SALAMA 1D's superior skill across various metrics and lead times of up to at least 11 hours. Moreover, expanding the archive of forecasts from which training examples are sampled improves skill, even when training set size remains constant. Finally, a sensitivity analysis using saliency maps indicates that our model relies on physically interpretable patterns consistent with established theoretical understanding, such as ice particle content near the tropopause, cloud cover, conditional instability, and low-level moisture.

physics.ao-ph↗

arXiv2025

Increasing NWP Thunderstorm Predictability Using Ensemble Data and Machine Learning

While numerical weather prediction (NWP) models are essential for forecasting thunderstorms hours in advance, NWP uncertainty, which increases with lead time, limits the predictability of thunderstorm occurrence. This study investigates how ensemble NWP data and machine learning (ML) can enhance the skill of thunderstorm forecasts. Using our recently introduced neural network model, SALAMA 1D, which identifies thunderstorm occurrence in operational forecasts of the convection-permitting ICON-D2-EPS model for Central Europe, we demonstrate that ensemble-averaging significantly improves forecast skill. Notably, an 11-hour ensemble forecast matches the skill level of a 5-hour deterministic forecast. To explain this improvement, we derive an analytic expression linking skill differences to correlations between ensemble members, which aligns with observed performance gains. This expression generalizes to any binary classification model that processes ensemble members individually. Additionally, we show that ML models like SALAMA 1D can identify patterns of thunderstorm occurrence which remain predictable for longer lead times compared to raw NWP output. Our findings quantitatively explain the benefits of ensemble-averaging and encourage the development of ML methods for thunderstorm forecasting and beyond.

physics.ao-ph↗

arXiv2024

A minimal model of the deep-convection lifecycle and its verification in remote-sensing observations

Deep convection is one of the most important atmospheric transport mechanisms and associated with various severe weather phenomena. Manifestations of deep convection in the atmosphere are composed of a recurring fundamental building block, called cell, which evolves through a characteristic lifecycle. Despite its importance, no simple, physically consistent quantitative lifecycle model exists that correctly reproduces remote-sensing observations. Based on the standard conceptual model, we develop an analytic minimal model of the convection lifecycle in the form of coupled reaction rules. This reaction scheme is equivalent to a system of coupled nonlinear differential equations that qualitatively agree with the empirically known dynamics. In order to demonstrate quantitative agreement of our convection-lifecycle model, we construct a representative random sample of satellite and radar observations of deep-convection events over Germany. In particular, we introduce a conditional sampling strategy based on an objective timescale criterion to remove events not covered by our model from the random sample. We show that by this procedure, a definite and recurrent temporal signature of the convection lifecycle can be identified. Our model closely reproduces the principal characteristics while essentially staying within one standard deviation of the mean signature. Next to the definite relevance of our model for nowcasting, the approach may be useful for convection parametrisations.

physics.ao-ph↗

arXiv2024

The role of mobility in epidemics near criticality

The general epidemic process (GEP), also known as susceptible-infected-recovered model (SIR), describes how an epidemic spreads within a population of susceptible individuals who acquire permanent immunization upon recovery. This model exhibits a second-order absorbing state phase transition, commonly studied assuming immobile healthy individuals. We investigate the impact of mobility on disease spreading near the extinction threshold by introducing two generalizations of GEP, where the mobility of susceptible and recovered individuals is examined independently. In both cases, including mobility violates GEP's rapidity reversal symmetry and alters the number of absorbing states. The critical dynamics of the models are analyzed through a perturbative renormalization group approach and large-scale stochastic simulations using a Gillespie algorithm. The renormalization group analysis predicts both models to belong to the same novel universality class describing the critical dynamics of epidemic spreading when the infected individuals interact with a diffusive species and gain immunization upon recovery. At the associated renormalization group fixed point, the immobile species decouples from the dynamics of the infected species, dominated by the coupling with the diffusive species. Numerical simulations in two dimensions affirm our renormalization group results by identifying the same set of critical exponents for both models. Violation of the rapidity reversal symmetry is confirmed by breaking the associated hyperscaling relation. Our study underscores the significance of mobility in shaping population spreading dynamics near the extinction threshold.

cond-mat.stat-mech↗