From calibration to parameter learning: Harnessing the scaling effects of big data in geoscientific modeling
This addresses a critical bottleneck in geosciences like hydrology and ecosystem sciences by enabling more efficient and scalable model calibration, though it is an incremental advancement integrating deep learning with process-based models.
The paper tackles the inefficiency and non-unique solutions in calibrating spatially-varying parameters for geoscientific models by proposing a differentiable parameter learning (dPL) framework, which achieves better performance, physical coherence, and generalizability with orders-of-magnitude lower computational cost, as demonstrated by drastically outperforming existing methods or requiring only ~12.5% of training data for similar performance.
The behaviors and skills of models in many geosciences (e.g., hydrology and ecosystem sciences) strongly depend on spatially-varying parameters that need calibration. A well-calibrated model can reasonably propagate information from observations to unobserved variables via model physics, but traditional calibration is highly inefficient and results in non-unique solutions. Here we propose a novel differentiable parameter learning (dPL) framework that efficiently learns a global mapping between inputs (and optionally responses) and parameters. Crucially, dPL exhibits beneficial scaling curves not previously demonstrated to geoscientists: as training data increases, dPL achieves better performance, more physical coherence, and better generalizability (across space and uncalibrated variables), all with orders-of-magnitude lower computational cost. We demonstrate examples that learned from soil moisture and streamflow, where dPL drastically outperformed existing evolutionary and regionalization methods, or required only ~12.5% of the training data to achieve similar performance. The generic scheme promotes the integration of deep learning and process-based models, without mandating reimplementation.