Kasra Khodkar , Ali Mirchi , Vahid Nourani , Afsaneh Kaghazchi , Jeffrey M. Sadler , Abubakarr Mansaray , Kevin Wagner , Phillip D. Alderman , Saleh Taghvaeian , Ryan T. Bailey
{"title":"数据稀缺地区的溪流盐度预测:迁移学习和不确定性量化的应用","authors":"Kasra Khodkar , Ali Mirchi , Vahid Nourani , Afsaneh Kaghazchi , Jeffrey M. Sadler , Abubakarr Mansaray , Kevin Wagner , Phillip D. Alderman , Saleh Taghvaeian , Ryan T. Bailey","doi":"10.1016/j.jconhyd.2024.104418","DOIUrl":null,"url":null,"abstract":"<div><p>Scarcity of stream salinity data poses a challenge to understanding salinity dynamics and its implications for water supply management in water-scarce salt-prone regions around the world. This paper introduces a framework for generating continuous daily stream salinity estimates using instance-based transfer learning (TL) and assessing the reliability of the synthetic salinity data through uncertainty quantification via prediction intervals (PIs). The framework was developed using two temporally distinct specific conductance (SC) datasets from the Upper Red River Basin (URRB) located in southwestern Oklahoma and Texas Panhandle, United States. The instance-based TL approach was implemented by calibrating Feedforward Neural Networks (FFNNs) on a source SC dataset of around 1200 instantaneous grab samples collected by United States Geological Survey (USGS) from 1959 to 1993. The trained FFNNs were subsequently tested on a target dataset (1998-present) of 220 instantaneous grab samples collected by the Oklahoma Water Resources Board (OWRB). The framework's generalizability was assessed in the data-rich Bird Creek watershed in Oklahoma by manipulating continuous SC data to simulate data-scarce conditions for training the models and using the complete Bird Creek dataset for model evaluation. The Lower Upper Bound Estimation (LUBE) method was used with FFNNs to estimate PIs for uncertainty quantification. Autoregressive SC prediction methods via FFNN were found to be reliable with Nash Sutcliffe Efficiency (NSE) values of 0.65 and 0.45 on in-sample and out-of-sample test data, respectively. The same modeling scenario resulted in an NSE of 0.54 for the Bird Creek data using a similar missing data ratio, whereas a higher ratio of observed data increased the accuracy (NSE = 0.84). The relatively narrow estimated PIs for the North Fork Red River in the URRB indicated satisfactory stream salinity predictions, showing an average width equivalent to 25 % of the observed range and a confidence level of 70 %.</p></div>","PeriodicalId":15530,"journal":{"name":"Journal of contaminant hydrology","volume":"266 ","pages":"Article 104418"},"PeriodicalIF":3.5000,"publicationDate":"2024-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Stream salinity prediction in data-scarce regions: Application of transfer learning and uncertainty quantification\",\"authors\":\"Kasra Khodkar , Ali Mirchi , Vahid Nourani , Afsaneh Kaghazchi , Jeffrey M. Sadler , Abubakarr Mansaray , Kevin Wagner , Phillip D. Alderman , Saleh Taghvaeian , Ryan T. Bailey\",\"doi\":\"10.1016/j.jconhyd.2024.104418\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<div><p>Scarcity of stream salinity data poses a challenge to understanding salinity dynamics and its implications for water supply management in water-scarce salt-prone regions around the world. This paper introduces a framework for generating continuous daily stream salinity estimates using instance-based transfer learning (TL) and assessing the reliability of the synthetic salinity data through uncertainty quantification via prediction intervals (PIs). The framework was developed using two temporally distinct specific conductance (SC) datasets from the Upper Red River Basin (URRB) located in southwestern Oklahoma and Texas Panhandle, United States. The instance-based TL approach was implemented by calibrating Feedforward Neural Networks (FFNNs) on a source SC dataset of around 1200 instantaneous grab samples collected by United States Geological Survey (USGS) from 1959 to 1993. The trained FFNNs were subsequently tested on a target dataset (1998-present) of 220 instantaneous grab samples collected by the Oklahoma Water Resources Board (OWRB). The framework's generalizability was assessed in the data-rich Bird Creek watershed in Oklahoma by manipulating continuous SC data to simulate data-scarce conditions for training the models and using the complete Bird Creek dataset for model evaluation. The Lower Upper Bound Estimation (LUBE) method was used with FFNNs to estimate PIs for uncertainty quantification. Autoregressive SC prediction methods via FFNN were found to be reliable with Nash Sutcliffe Efficiency (NSE) values of 0.65 and 0.45 on in-sample and out-of-sample test data, respectively. The same modeling scenario resulted in an NSE of 0.54 for the Bird Creek data using a similar missing data ratio, whereas a higher ratio of observed data increased the accuracy (NSE = 0.84). The relatively narrow estimated PIs for the North Fork Red River in the URRB indicated satisfactory stream salinity predictions, showing an average width equivalent to 25 % of the observed range and a confidence level of 70 %.</p></div>\",\"PeriodicalId\":15530,\"journal\":{\"name\":\"Journal of contaminant hydrology\",\"volume\":\"266 \",\"pages\":\"Article 104418\"},\"PeriodicalIF\":3.5000,\"publicationDate\":\"2024-09-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Journal of contaminant hydrology\",\"FirstCategoryId\":\"93\",\"ListUrlMain\":\"https://www.sciencedirect.com/science/article/pii/S0169772224001220\",\"RegionNum\":3,\"RegionCategory\":\"环境科学与生态学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q2\",\"JCRName\":\"ENVIRONMENTAL SCIENCES\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of contaminant hydrology","FirstCategoryId":"93","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0169772224001220","RegionNum":3,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"ENVIRONMENTAL SCIENCES","Score":null,"Total":0}
Stream salinity prediction in data-scarce regions: Application of transfer learning and uncertainty quantification
Scarcity of stream salinity data poses a challenge to understanding salinity dynamics and its implications for water supply management in water-scarce salt-prone regions around the world. This paper introduces a framework for generating continuous daily stream salinity estimates using instance-based transfer learning (TL) and assessing the reliability of the synthetic salinity data through uncertainty quantification via prediction intervals (PIs). The framework was developed using two temporally distinct specific conductance (SC) datasets from the Upper Red River Basin (URRB) located in southwestern Oklahoma and Texas Panhandle, United States. The instance-based TL approach was implemented by calibrating Feedforward Neural Networks (FFNNs) on a source SC dataset of around 1200 instantaneous grab samples collected by United States Geological Survey (USGS) from 1959 to 1993. The trained FFNNs were subsequently tested on a target dataset (1998-present) of 220 instantaneous grab samples collected by the Oklahoma Water Resources Board (OWRB). The framework's generalizability was assessed in the data-rich Bird Creek watershed in Oklahoma by manipulating continuous SC data to simulate data-scarce conditions for training the models and using the complete Bird Creek dataset for model evaluation. The Lower Upper Bound Estimation (LUBE) method was used with FFNNs to estimate PIs for uncertainty quantification. Autoregressive SC prediction methods via FFNN were found to be reliable with Nash Sutcliffe Efficiency (NSE) values of 0.65 and 0.45 on in-sample and out-of-sample test data, respectively. The same modeling scenario resulted in an NSE of 0.54 for the Bird Creek data using a similar missing data ratio, whereas a higher ratio of observed data increased the accuracy (NSE = 0.84). The relatively narrow estimated PIs for the North Fork Red River in the URRB indicated satisfactory stream salinity predictions, showing an average width equivalent to 25 % of the observed range and a confidence level of 70 %.
期刊介绍:
The Journal of Contaminant Hydrology is an international journal publishing scientific articles pertaining to the contamination of subsurface water resources. Emphasis is placed on investigations of the physical, chemical, and biological processes influencing the behavior and fate of organic and inorganic contaminants in the unsaturated (vadose) and saturated (groundwater) zones, as well as at groundwater-surface water interfaces. The ecological impacts of contaminants transported both from and to aquifers are of interest. Articles on contamination of surface water only, without a link to groundwater, are out of the scope. Broad latitude is allowed in identifying contaminants of interest, and include legacy and emerging pollutants, nutrients, nanoparticles, pathogenic microorganisms (e.g., bacteria, viruses, protozoa), microplastics, and various constituents associated with energy production (e.g., methane, carbon dioxide, hydrogen sulfide).
The journal''s scope embraces a wide range of topics including: experimental investigations of contaminant sorption, diffusion, transformation, volatilization and transport in the surface and subsurface; characterization of soil and aquifer properties only as they influence contaminant behavior; development and testing of mathematical models of contaminant behaviour; innovative techniques for restoration of contaminated sites; development of new tools or techniques for monitoring the extent of soil and groundwater contamination; transformation of contaminants in the hyporheic zone; effects of contaminants traversing the hyporheic zone on surface water and groundwater ecosystems; subsurface carbon sequestration and/or turnover; and migration of fluids associated with energy production into groundwater.