Geology topics

Alexander Y. Sun

Publications and source records attributed to Alexander Y. Sun.

4 recordsLinked to original sources

USGS2024

Fair graph learning using constraint-aware priority adjustment and graph masking in river networks

Accurate prediction of water quality and quantity is crucial for sustainable development and human well-being. However, existing data-driven methods often suffer from spatial biases in model performance due to heterogeneous data, limited observations, and noisy sensor data. To overcome these challenges, we propose Fair-Graph, a novel graph-based recurrent neural network that leverages interrelated knowledge from multiple rivers to predict water flow and temperature within large-scale stream networks. Additionally, we introduce node-specific graph masks for information aggregation and adaptation to enhance prediction over heterogeneous river segments. To reduce performance disparities across river segments, we introduce a centralized coordination strategy that adjusts training priorities for segments. We evaluate the prediction of water temperature within the Delaware River Basin, and the prediction of streamflow using simulated data from U.S. National Water Model in the Houston River network. The results showcase improvements in predictive performance and highlight the proposed model's ability to maintain spatial fairness over different river segments.

Delaware River basin↗

USGS2023

Stream temperature prediction in a shifting environment: The influence of deep learning architecture

Stream temperature is a fundamental control on ecosystem health. Recent efforts incorporating process guidance into deep learning models for predicting stream temperature have been shown to outperform existing statistical and physical models. This performance is in part because deep learning architectures can actively learn spatiotemporal relationships that govern how water and energy propagate through a river network. However, exploration of how spatiotemporal awareness and process guidance influence a model's generalizability under shifting environmental conditions such as climate change is limited. Here, we use Explainable Artificial Intelligence (XAI) to interrogate how differing deep learning architectures affect a model's learned spatial and temporal dependencies, and how those learned dependencies affect a model's ability to maintain high accuracy when applied to unseen environmental conditions. Using the Delaware River Basin in the northeastern United States as a test case, we compare two spatiotemporally aware process-guided deep learning models for predicting stream temperature (a recurrent graph convolution network—RGCN, and a temporal convolution graph model—Graph WaveNet). Both models achieve equally high predictive performance when testing data are well represented in the training data (test root mean squared errors of 1.64°C and 1.65°C); however, Graph WaveNet significantly outperforms RGCN in 4 out of 5 experiments where test partitions represent different types of unseen environmental conditions. XAI results show that the architecture of Graph WaveNet leads to learned spatial relationships with greater fidelity to physical processes, and that this fidelity improves the generalizability of the model when applied to shifting and/or unseen environmental conditions.

Delaware River Basin↗

USGS2021

Effects of climate and irrigation on GRACE-based estimates of water storage changes in major US aquifers

Understanding climate and human impacts on water storage is critical for sustainable water-resources management. Here we assessed climate and human drivers of total water storage (TWS) variability from Gravity Recovery and Climate Experiment (GRACE) satellites compared with drought severity and irrigation water use in 14 major aquifers in the United States. Results show that long-term variability in TWS tracked by GRACE satellites is dominated by interannual variability in most of the 14 major US aquifers. Low TWS trends in the humid eastern U.S. are linked to low drought intensity. Although irrigation pumpage in the humid Mississippi Embayment aquifer exceeded that in the semi-arid California Central Valley, a surprising lack of TWS depletion in the Mississippi Embayment aquifer is attributed to extensive streamflow capture. Marked storage depletion in the semi-arid southwestern Central Valley and south-central High Plains totaled ∼90 km 3 , about three times greater than the capacity of Lake Mead, the largest U.S. reservoir. Depletion in the Central Valley was driven by long-term droughts (⩽5 yr) amplified by switching from mostly surface water to groundwater irrigation. Low or slightly rising TWS trends in the northwestern (Columbia and Snake Basins) US are attributed to dampening drought impacts by mostly surface water irrigation. GRACE satellite data highlight synergies between climate and irrigation, resulting in little impact on TWS in the humid east, amplified TWS depletion in the semi-arid southwest and southcentral US, and dampened TWS deletion in the northwest and north central US Sustainable groundwater management benefits from conjunctive use of surface water and groundwater, inefficient surface water irrigation promoting groundwater recharge, efficient groundwater irrigation minimizing depletion, and increasing managed aquifer recharge. This study has important implications for sustainable water development in many regions globally.

Environmental Research Letters↗

USGS2020

Comparison of groundwater storage changes from GRACE satellites with monitoring and modeling of major U.S. aquifers

GRACE satellite data are widely used to estimate groundwater storage (GWS) changes in aquifers globally; however, comparisons with GW monitoring and modeling data are limited. Here we compared GWS changes from GRACE over 15 yr (2002–2017) in 14 major U.S. aquifers with groundwater-level (GWL) monitoring data in ~23,000 wells and with regional and global hydrologic and land surface models. Results show declining GWS trends from GRACE data in the six southwestern and south-central U.S. aquifers, totaling −90 km 3 over 15 yr, related to long-term (5–15 yr) droughts, and exceeding Lake Mead volume by ~2.5×. GWS trends in most remaining aquifers were stable or slightly rising. GRACE-derived GWS changes agree with GWL monitoring data in most aquifers (correlation coefficients, R = 0.52–0.95), showing that GRACE satellites capture groundwater (GW) dynamics. Regional GW models (eight models) generally show similar or greater GWS trends than those from GRACE. Large discrepancies in the Mississippi Embayment aquifer, with modeled GWS decline approximately four times that of GRACE, may reflect uncertainties in model storage parameters, stream capture, pumpage, and/or recharge rates. Global hydrologic models (2003–2014), which include GW pumping, generally overestimate GRACE GWS depletion (total: approximately −172 to −186 km 3 ) in heavily exploited aquifers in southwestern and south-central U.S. by ~2.4× (GRACE: −74 km 3 ), underscoring needed modeling improvements relative to anthropogenic impacts. Global land surface models tend to track GRACE GWS dynamics better than global hydrologic models. Intercomparing remote sensing, monitoring, and modeling data underscores the importance of considering all data sources to constrain GWS uncertainties.

Water Resources Research↗