scholarly journals Development of a Deep Learning Emulator for a Distributed Groundwater–Surface Water Model: ParFlow-ML

Water ◽  
2021 ◽  
Vol 13 (23) ◽  
pp. 3393
Author(s):  
Hoang Tran ◽  
Elena Leonarduzzi ◽  
Luis De la Fuente ◽  
Robert Bruce Hull ◽  
Vineet Bansal ◽  
...  
Keyword(s):  
Deep Learning ◽  
Water Table ◽  
Overland Flow ◽  
Three Dimensional ◽  
Pressure Head ◽  
Hydrologic Model ◽  
Water Table Depth ◽  
Water Model ◽  
Physical Parameters ◽  
Integrated Hydrologic Model

Integrated hydrologic models solve coupled mathematical equations that represent natural processes, including groundwater, unsaturated, and overland flow. However, these models are computationally expensive. It has been recently shown that machine leaning (ML) and deep learning (DL) in particular could be used to emulate complex physical processes in the earth system. In this study, we demonstrate how a DL model can emulate transient, three-dimensional integrated hydrologic model simulations at a fraction of the computational expense. This emulator is based on a DL model previously used for modeling video dynamics, PredRNN. The emulator is trained based on physical parameters used in the original model, inputs such as hydraulic conductivity and topography, and produces spatially distributed outputs (e.g., pressure head) from which quantities such as streamflow and water table depth can be calculated. Simulation results from the emulator and ParFlow agree well with average relative biases of 0.070, 0.092, and 0.032 for streamflow, water table depth, and total water storage, respectively. Moreover, the emulator is up to 42 times faster than ParFlow. Given this promising proof of concept, our results open the door to future applications of full hydrologic model emulation, particularly at larger scales.

scholarly journals A national topographic dataset for hydrological modeling over the contiguous United States

2021 ◽  
Vol 13 (7) ◽  
pp. 3263-3279
Author(s):  
Jun Zhang ◽  
Laura E. Condon ◽  
Hoang Tran ◽  
Reed M. Maxwell
Keyword(s):  
United States ◽  
Hydrological Modeling ◽  
Overland Flow ◽  
R Package ◽  
The United States ◽  
Network Routing ◽  
Hydrologic Model ◽  
Water Model ◽  
Stream Network ◽  
Integrated Hydrologic Model

Abstract. Topography is a fundamental input to hydrologic models critical for generating realistic streamflow networks as well as infiltration and groundwater flow. Although there exist several national topographic datasets for the United States, they may not be compatible with gridded models that require hydrologically consistent digital elevation models (DEMs). Here, we present a national topographic dataset developed to support gridded hydrologic simulations at 1 km and 250 m spatial resolution over the contiguous United States. The workflow is described step by step in two parts: (a) DEM processing using a Priority Flood algorithm to ensure hydrologically consistent drainage networks and (b) slope calculation and smoothing to improve drainage performance. The accuracy of the derived stream network is evaluated by comparing the derived drainage area to drainage areas reported by the national stream gage network. The slope smoothing steps are evaluated using the runoff simulations with an integrated hydrologic model. Our DEM product started from the National Water Model DEM to ensure our final datasets will be as consistent as possible with this existing national framework. Our analysis shows that the additional processing we provide improves the consistency of simulated drainage areas and the runoff simulations that simulate gridded overland flow (as opposed to a network routing scheme). The workflow uses an open-source R package, and all output datasets and processing scripts are available and fully documented. All of the output datasets and scripts for processing are published through CyVerse at 250 m and 1 km resolution. The DOI link for the dataset is https://doi.org/10.25739/e1ps-qy48 (Zhang and Condon, 2020).

scholarly journals Development of a Coupled Land Surface Hydrologic Model and Evaluation at a Critical Zone Observatory

2013 ◽  
Vol 14 (5) ◽  
pp. 1401-1420 ◽  
Author(s):  
Yuning Shi ◽  
Kenneth J. Davis ◽  
Christopher J. Duffy ◽  
Xuan Yu
Keyword(s):  
Latent Heat ◽  
Water Table ◽  
Land Surface ◽  
Hydrologic Model ◽  
Heat Fluxes ◽  
Water Table Depth ◽  
Surface Model ◽  
Land Surface Scheme ◽  
Model Predictions ◽  
Surface Scheme

Abstract A fully coupled land surface hydrologic model, Flux-PIHM, is developed by incorporating a land surface scheme into the Penn State Integrated Hydrologic Model (PIHM). The land surface scheme is adapted from the Noah land surface model. Because PIHM is capable of simulating lateral water flow and deep groundwater at spatial resolutions sufficient to resolve upland stream networks, Flux-PIHM is able to represent heterogeneities due to topography and soils at high resolution, including spatial structure in the link between groundwater and the surface energy balance (SEB). Flux-PIHM has been implemented at the Shale Hills watershed (0.08 km2) in central Pennsylvania. Multistate observations of discharge, water table depth, soil moisture, soil temperature, and sensible and latent heat fluxes in June and July 2009 are used to manually calibrate Flux-PIHM at hourly temporal resolution. Model predictions from 1 March to 1 December 2009 are evaluated. Both hydrologic predictions and SEB predictions show good agreement with observations. Comparisons of model predictions between Flux-PIHM and the original PIHM show that the inclusion of the complex SEB simulation only brings slight improvement in hourly model discharge predictions. Flux-PIHM adds the ability of simulating SEB to PIHM and does improve the prediction of hourly evapotranspiration, the prediction of total runoff (discharge), and the predictions of some peak discharge events, especially after extended dry periods. Model results reveal that annual average sensible and latent heat fluxes are strongly correlated with water table depth, and the correlation is especially strong for the model grids near the stream.

scholarly journals Analysis of an extreme rainfall-runoff event at the Landscape Evolution Observatory by means of a three-dimensional physically-based hydrologic model

2013 ◽  
Vol 10 (10) ◽  
pp. 12615-12641 ◽  
Author(s):  
G.-Y. Niu ◽  
D. Pasetto ◽  
C. Scudeler ◽  
C. Paniconi ◽  
M. Putti ◽  
...  
Keyword(s):  
Size Distribution ◽  
Null Hypothesis ◽  
Overland Flow ◽  
Landscape Evolution ◽  
Alternative Hypothesis ◽  
Three Dimensional ◽  
Hydrologic Model ◽  
Distribution Analysis ◽  
Physically Based ◽  
Seepage Face

Abstract. We present a detailed analysis, by means of a three-dimensional physically-based hydrological model, of the first experiment conducted at the Biosphere 2 Landscape Evolution Observatory (LEO). The experiment was driven by an intense rainfall event and produced a hydrological response characterized predominantly by water outflow along the lower lateral boundary (seepage face) of LEO, together with overland flow that began 15 h after the start of rainfall and caused erosion of the superficial soil and formation of a small channel. The analysis is designed to test the null hypothesis that the soil is hydraulically homogenous, and an alternative hypothesis that the soil has developed some hydraulic heterogeneity in the downstream direction due to saturated soil compaction near the seepage face. More than 20 000 sensitivity simulations were run in a systematic search for optimal parameters to reproduce measurements of seepage face outflow and hillslope water storage. We varied the saturated hydraulic conductivity (Ksat) of the seepage face (18 values), Ksat in the rest of the LEO soil (30 values), and soil porosity (21 values), and we considered two values of the pore size distribution parameter (n) in the water retention characteristics, obtained from a particle size distribution analysis and from laboratory experiments on LEO soil samples. For both n values, the best simulations under the heterogeneous soil hypothesis produced smaller errors than the best runs under the null hypothesis. Moreover the heterogeneous runs yielded a higher probability of best realizations than the homogenous runs. These results support the hypothesis of localized incipient heterogeneity of the LEO soil.

scholarly journals Deep learning‐based carotid media‐adventitia and lumen‐intima boundary segmentation from three‐dimensional ultrasound images

2019 ◽  
Vol 46 (7) ◽  
pp. 3180-3193 ◽  
Author(s):  
Ran Zhou ◽  
Aaron Fenster ◽  
Yujiao Xia ◽  
J. David Spence ◽  
Mingyue Ding
Keyword(s):  
Deep Learning ◽  
Three Dimensional ◽  
Ultrasound Images

scholarly journals Temporal and Local Heterogeneities of Water Table Depth under Different Agricultural Water Management Conditions

Water ◽  
2021 ◽  
Vol 13 (16) ◽  
pp. 2148
Author(s):  
Jonathan A. Lafond ◽  
Silvio J. Gumiere ◽  
Virginie Vanlandeghem ◽  
Jacques Gallichand ◽  
Alain N. Rousseau ◽  
...  
Keyword(s):  
Water Management ◽  
Water Table ◽  
Cropping Systems ◽  
Irrigation Management ◽  
Sprinkler Irrigation ◽  
Water Table Depth ◽  
Management Systems ◽  
Integrated Water Management ◽  
Growing Seasons ◽  
Good Drainage

Integrated water management has become a priority for cropping systems where subirrigation is possible. Compared to conventional sprinkler irrigation, the controlling water table can lead to a substantial increase in yield and water use efficiency with less pumping energy requirements. Knowing the spatiotemporal distribution of water table depth (WTD) and soil properties should help perform intelligent, integrated water management. Observation wells were installed in cranberry fields with different water management systems: Bottom, with good drainage and controlled WTD management; Surface, with good drainage and sprinkler irrigation management; Natural, without drainage, or with imperfectly drained and conventional sprinkler irrigation. During the 2017–2020 growing seasons, WTD was monitored on an hourly basis, while precipitation was measured at each site. Multi-frequential periodogram analysis revealed a dominant periodic component of 40 days each year in WTD fluctuations for the Bottom and Surface systems; for the Natural system, periodicity was heterogeneous and ranged from 2 to 6 weeks. Temporal cross correlations with precipitation show that for almost all the sites, there is a 3 to 9 h lag before WTD rises; one exception is a subirrigation site. These results indicate that automatic water table management based on continuously updated knowledge could contribute to integrated water management systems, by using precipitation-based models to predict WTD.

scholarly journals Automated Segmentation of Infarct Lesions in T1-Weighted MRI Scans Using Variational Mode Decomposition and Deep Learning

Sensors ◽  
2021 ◽  
Vol 21 (6) ◽  
pp. 1952
Author(s):  
May Phu Paing ◽  
Supan Tungjitkusolmun ◽  
Toan Huy Bui ◽  
Sarinporn Visitsattapongse ◽  
Chuchart Pintavirooj
Keyword(s):  
Deep Learning ◽  
Brain Infarction ◽  
Three Dimensional ◽  
Dice Similarity Coefficient ◽  
Automated Segmentation ◽  
Variational Mode Decomposition ◽  
Brain Scans ◽  
Automated Method ◽  
Mode Decomposition ◽  
Segmentation Task

Automated segmentation methods are critical for early detection, prompt actions, and immediate treatments in reducing disability and death risks of brain infarction. This paper aims to develop a fully automated method to segment the infarct lesions from T1-weighted brain scans. As a key novelty, the proposed method combines variational mode decomposition and deep learning-based segmentation to take advantages of both methods and provide better results. There are three main technical contributions in this paper. First, variational mode decomposition is applied as a pre-processing to discriminate the infarct lesions from unwanted non-infarct tissues. Second, overlapped patches strategy is proposed to reduce the workload of the deep-learning-based segmentation task. Finally, a three-dimensional U-Net model is developed to perform patch-wise segmentation of infarct lesions. A total of 239 brain scans from a public dataset is utilized to develop and evaluate the proposed method. Empirical results reveal that the proposed automated segmentation can provide promising performances with an average dice similarity coefficient (DSC) of 0.6684, intersection over union (IoU) of 0.5022, and average symmetric surface distance (ASSD) of 0.3932, respectively.

Sign in / Sign up

Export Citation Format

Share Document