Advancing Tropical Maritime Convection Prediction: An Integrated Modeling and Observation Approach

Accurate Earth system prediction relies on both models and observations, yet both sources possess distinct uncertainties. Models are limited by initial condition errors and structural deficiencies in physical parameterizations. These deficiencies are especially magnified in tropical maritime environments, where physics parameterizations may misrepresent key processes given their primary development for mid-latitude continental systems. Concurrently, observations are subject to instrument limitations, retrieval algorithm assumptions, sampling biases. Consequently, standard merging techniques like statistical bias correction and data assimilation often fail to account for this full spectrum of uncertainty.

To address this, the NSF NCAR INFORM initiative facilitates a better integration of observational datasets with community models. In this invited talk, we present an observation-informed workflow that rigorously addresses uncertainties in both sources to diagnose and correct model bias. We focus on a tropical maritime squall line observed during the 2022 PRECIP field campaign.

First, we address the gap between model physics and observational reality. To account for uncertainties in observational quality control and sampling, we process model output through the CR-SIM forward operator and apply identical quality control to real SEA-POL radar observations. This observation-informed approach reveals that a microphysics scheme overestimates raindrop size due to a misrepresentation of the efficient breakup characteristic of tropical maritime convection. Targeted corrections to these processes substantially reduce model bias.

Second, we broaden this investigation by examining multiple microphysics schemes to quantify their impact on the representation of tropical maritime convection. We aim to leverage internal characteristics—specifically radar-observable microphysical properties—to diagnose and correct systematic biases. By constraining these internal processes, we evaluate how observation-informed physics improvements propagate upscale to enhance the simulation of convective organization and longevity. This analysis demonstrates how our workflow guides targeted corrections, ensuring the simulated storm structure aligns with the observed state.