Existing early and late fusion frameworks in multimodal learning are confronted with the fundamental challenge of modality imbalance, wherein disparities in representational capacities induce inter-modal competition during training. Current research methodologies primarily rely on modality-level contribution assessments to measure gaps in representational capabilities and enhance poorly learned modalities, overlooking the dynamic variations of modality contributions across individual samples. To address this, we propose a Causal-aware Modality valuation approach for Balanced multimodal learning (CMoB). We define a benefit function based on Shannon's theory of informational uncertainty to evaluate the changes in the importance of samples across different stages of multimodal training. Inspired by human cognitive science, we propose a causal-aware modality contribution quantification method from a causal perspective to capture fine-grained changes in modality contribution degrees within samples. In the iterative training of multimodal learning, we develop targeted modal enhancement strategies that dynamically select and optimize modalities based on real-time evaluation of their contribution variations across training samples. Our method enhances the discriminative ability of key modalities and the learning capacity of weak modalities while achieving fine-grained balance in multimodal learning. Extensive experiments on benchmark multimodal datasets and multimodal frameworks demonstrate the superiority of our CMoB approach for balanced multimodal learning.

This study addresses two key uncertainties in the fire radiative power (FRP) retrieval, which is essential for improving global top-down fire emission inventories. First, it proposes a novel FRP retrieval method by combining the ~4 and$\sim 8.6~\mu $m channels based on Monte Carlo simulation, which is verified using the Visible Infrared Imaging Radiometer Suite (VIIRS). The inclusion of the$\sim 8.6~\mu $m channel significantly improves the accuracy of FRP retrieval, especially for highly smoldering fires. Second, atmospheric correction is conducted using outputs from the state-of-the-art unified linearized vector radiative transfer model (UNL-VRTM). The importance of atmospheric correction is demonstrated through the single-channel ($\sim 4~\mu $m) FRP retrievals from the Moderate Resolution Imaging Spectroradiometer (MODIS) active fire (AF), VIIRS AF, and VIIRS second-generation fire light detection algorithm (FILDA-2) products. Post-correction results show effective mitigation of nighttime FRP angular dependency, achieved by considering the enhanced atmospheric attenuation due to longer path length off-nadir. However, a residual daytime FRP angular dependency remains, likely due to the angular dependency of the thresholds used for daytime fire detection. Additionally, an enhanced agreement is observed between the VIIRS FILDA-2 FRP retrievals from the Suomi National Polar-orbiting Partnership (NPP) and National Oceanic and Atmospheric Administration (NOAA)-20 satellites after correction. Lastly, a global FRP increase is noted across all three products, with VIIRS AF and VIIRS FILDA-2 showing more significant increases (65.8% and 62.5%, respectively) than MODIS AF (20.8%). These advancements in FRP retrievals may enhance the downstream fire emission products, which will benefit the air pollution modeling community.