The launch of GRACE satellites has provided a new avenue for studying the terrestrial water storage anomalies (TWSA) with unprecedented accuracy. However, the coarse spatial resolution greatly limits its application in hydrology researches on local scales. To overcome this limitation, this study develops a machine learning-based fusion model to obtain high-resolution (0.25°) groundwater level anomalies (GWLA) by integrating GRACE observations in the North China Plain. Specifically, the fusion model consists of three modules, namely the downscaling module, the data fusion module, and the prediction module, respectively. In terms of the downscaling module, the GRACE-Noah model outperforms traditional data-driven models (multiple linear regression and gradient boosting decision tree (GBDT)) with the correlation coefficient (CC) values from 0.24 to 0.78. With respect to the data fusion module, the groundwater level from 12 monitoring wells is incorporated with climate variables (precipitation, runoff, and evapotranspiration) using the GBDT algorithm, achieving satisfactory performance (mean values: CC: 0.97, RMSE: 1.10 m, and MAE: 0.87 m). By merging the downscaled TWSA and fused groundwater level based on the GBDT algorithm, the prediction module can predict the water level in specified pixels. The predicted groundwater level is validated against 6 in-situ groundwater level data sets in the study area. Compare to the downscaling module, there is a significant improvement in terms of CC metrics, on average, from 0.43 to 0.71. This study provides a feasible and accurate fusion model for downscaling GRACE observations and predicting groundwater level with improved accuracy.
Latent heat flux (LE) plays an essential role in the hydrological cycle, surface energy balance, and climate change, but the spatial resolution of site-scale LE extremely limits its application potential over a regional scale. To overcome the limitation, five transfer learning models were constructed based on artificial neural networks (ANNs), random forests (RFs), extreme gradient boosting (XGBoost), support vector machine (SVM), and light gradient boosting machine (LightGBM) to upscale LE from site scale to regional scale in Heihe River basin (HRB). The instance-transfer approach that utilizes data samples outside of HRB was used in the transfer learning models. Moreover, the Bayesian-based three-cornered hat (BTCH) method was used to fuse the best three upscaling results from ANN, RF, and XGBoost models to improve the accuracy of the results. The results indicated that the transfer learning models perform best when the transfer ratio (the data samples ratio between external and HRB dataset) was 0.6. Specifically, the coefficient of determination (R2) and root mean squared errors (RMSE) of LE upscaled by ANN model was improved or reduced by 6% or 17% than the model without external data. Furthermore, the BTCH method can effectively improve the performance of single transfer learning model with the highest accuracy (R2 = 0.83, RMSE = 18.84 W/m2). Finally, the LE upscaling model based on transfer learning model demonstrated great potential in HRB, which may be applicable to similar research in other regions.
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.