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.
The accuracy of these models has important implications for relevant government agencies and public stakeholders that place trust in them. The demand for accurate modeling capabilities will likely be on the rise due to increased risks of floods and droughts because of climate change (IPCC, 2021). Traditionally, regional hydrologic models describe not only streamflow but also other water stores in the hydrologic cycle (snow, surface ponding, soil moisture, and groundwater), as well as fluxes (evapotranspiration, surface runoff, subsurface runoff, and baseflow), whereas newer, data-driven machine learning approaches tend to focus on prediction of the variable on which it has been trained. The physical states (stores) and fluxes in traditional models help to provide a full narrative of the event, for example, high antecedent soil moisture or thawing snow primed the watershed for floods, which are important for communication with stakeholders.
Deep learning (DL) models trained on hydrologic observations can perform extraordinarily well, but they can inherit deficiencies of the training data, such as limited coverage of in situ data or low resolution/accuracy of satellite data. Here we propose a novel multiscale DL scheme learning simultaneously from satellite and in situ data to predict 9 km daily soil moisture (5 cm depth). Based on spatial cross‐validation over sites in the conterminous United States, the multiscale scheme obtained a median correlation of 0.901 and root‐mean‐square error of 0.034 m3/m3. It outperformed the Soil Moisture Active Passive satellite mission's 9 km product, DL models trained on in situ data alone, and land surface models. Our 9 km product showed better accuracy than previous 1 km satellite downscaling products, highlighting limited impacts of improving resolution. Not only is our product useful for planning against floods, droughts, and pests, our scheme is generically applicable to geoscientific domains with data on multiple scales, breaking the confines of individual data sets.
Satellite products can provide spatiotemporal data on precipitation in ungauged basins. It is essential and meaningful to assess and correct these products. In this study, the Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks-Climate Data Record (PERSIANN-CDR) product was evaluated and corrected using the successive correction method. A simple hydrological model was driven by the corrected PERSIANN-CDR data. The results showed that the accuracy of the original PERSIANN-CDR data was low on a daily scale, and the accuracy decreased gradually from the east to the west of the basin. With one correction step, the accuracy of the corrected PERSIANN-CDR data was significantly higher than that of the initial data. The correlation coefficient increased from 0.58 to 0.73, and the probability of detection (POD) value of the corrected product was 18.2% higher than the original product. The temporal-spatial resolution influenced the performance of the satellite product. As the resolution became coarser, the correlation coefficient between the corrected PERSIANN-CDR data and the gauged data gradually became lower. The Identification of unit Hydrographs and Component flows from Rainfall, Evapotranspiration, and Streamflow (IHACRES) model could be satisfactorily applied in the Lhasa River basin with corrected PERSIANN-CDR data. The successive correction method was an effective way to correct the bias of the PERSIANN-CDR product.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.