The statistical power to detect regional temporal trends in riverine contaminants in the Chesapeake Bay Watershed, USA

Wagner, Tyler; McLaughlin, Paul; Smalling, Kelly L.; Breitmeyer, Sara E.; Gordon, Stephanie; Noe, Gregory B.

doi:10.1016/j.scitotenv.2021.152435

Cited by 8 publications

(10 citation statements)

References 57 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Note that this two-stage machine learning model described above is not explicitly designed for MICX but can be applied to all leftcensored response variables, such as heavy metal and volatile organic pollutants. 2,4 We compared this two-stage model to a baseline model in which all data points below and above the detection limit are all used for regression. This is the most conventional and typical way of predicting MICX.…”

Section: Two-stage Machine Learning Modelmentioning

confidence: 99%

“…For example, Tobit regression, a most widely used model for left-censored variables, forces the predicted values to be the censoring point if it is predicted to be below that point. 7,15−18 Many other statistical models are also available for modeling left-censored variables, such as zero-inflated model, 19 robust regression, 7 Heckman correction for sample selection, 20,21 hierarchical Bayesian censored regression, 4,22 and hockey-stick model. 23 Machine learning models have many advantages over statistical models in making accurate predictions, 24 but unlike statistical models, few studies have addressed the issues of modeling left-censored variables with machine learning models.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A Two-Stage Interpretable Machine Learning Framework for Accurate Prediction of Trace Pollutants: With an Application to Microcystin

Wu,

Liang,

Qiu

2023

ACS EST Water

View full text Add to dashboard Cite

Trace pollutants are widely observed in aquatic ecosystems and can significantly impact human health and the environment. Accurate prediction of trace pollutants and understanding their response to environmental drivers are key to environmental management, yet these tasks remain challenging. An important reason for this challenge is that monitoring data for trace pollutants are often left-censored, leading to biased estimation and inaccurate response-driver relationships. Here we propose a novel two-stage interpretable machine learning framework applicable to left-censored trace pollutant data. The two stages include (1) a classifier to predict the presence of the pollutant and (2) a regressor to predict the pollutant concentration if present. The two stages were followed by a model interpretation to understand the contribution of drivers to the presence and concentration of the pollutant accordingly. We take the prediction of microcystin (MICX) in lakes across the United States as a case study. Applying this framework to MICX consistently improved prediction accuracy, including prediction of its occurrence and concentration regardless of the algorithms and performance metrics used. The best-performing algorithm using the two-stage framework, compared with the baseline model, improves classification performance by 48% to 290% and the regression performance by 11% to 33%, depending on the metric used to evaluate the performance. The interpretable machine learning model also successfully revealed the impacts of the most important drivers on the presence of MICX and its concentration. Our results showed the advantages of this framework, including its interpretability to understand the driver-response relationship, ability to handle nonlinearity, better prediction performance, differentiation between the underlying processes, and potential to be generalized to other pollutants. As such, we anticipate that the framework we propose will be a starting point for using state-of-the-art interpretable machine learning models for predicting trace pollutants.

show abstract

Section: Two-stage Machine Learning Modelmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

A Two-Stage Interpretable Machine Learning Framework for Accurate Prediction of Trace Pollutants: With an Application to Microcystin

Wu,

Liang,

Qiu

2023

ACS EST Water

View full text Add to dashboard Cite

show abstract

“…Due to the effective implementation of government policies, digital modeling and the application and development of remote sensing technology, significant progress has been made in solving the problem of agricultural nonpoint source pollution [12]. Studies on agricultural nonpoint source pollution at home and abroad have focused on pollutant monitoring and load calculation [13][14][15][16]; pollutant accounting and distribution characteristics [17]; and pollution prevention and control, emission reduction and impact factors [18,19]. Studies have shown that China's Taihu Lake Basin suffers the most serious agricultural nonpoint source pollution [20], while Chaohu Lake Basin shows an overall increasing trend of TN pollution and a small increasing trend of TP pollution [21].…”

Section: Literature Reviewmentioning

confidence: 99%

Decoupling agriculture pollution and carbon reduction from economic growth in the Yangtze River Delta, China

Yuan

Xu²,

Kong³

2023

PLoS ONE

View full text Add to dashboard Cite

Agriculture is the foundation of the national economy, and agricultural nonpoint source pollution and carbon emissions are the main environmental problems limiting the development of the agricultural economy. This study takes the Yangtze River Delta as the research object and measures agricultural carbon emissions and nonpoint source pollution in the study area from 2010 to 2020 respectively. The Tapio decoupling model is used to study types of decoupling between agricultural pollution and carbon reduction and economic growth in the Yangtze River Delta from 2010 to 2020, and the GM (1,1) model is used to predict the decoupling relationship between the agricultural environment and economic growth over the next ten years. The results show the following: (1) Agricultural COD emissions come mainly from livestock and poultry breeding, dropped from 1,130,120 tons in 2010 to 908,460 tons in 2020. Agricultural TN and TP emissions come mainly from plantations. Agricultural TN emissions dropped from 892,310 tons in 2010 to 788,020 tons in 2020. Agricultural TP emissions dropped from 149,590 tons in 2010 to130,770 tons in 2020. Agricultural carbon emissions dropped from 17,115,900 tons in 2010 to 15,786,600 tons in 2020, and come mainly from agricultural fertilizer and diesel fuel and pig breeding. (2) The decoupling effect of agricultural pollution reduction and carbon reduction in the Yangtze River Delta and economic growth has been in a long-term state, with negative decoupling occurring in a few regions, mainly in 2011, 2014 and 2020. (3) In the next ten years, except for 2021, when the coordination between agricultural pollution reduction and economic growth is poor, the two show good decoupling in the remaining years. Based on the results, this study makes recommendations on how to carry out comprehensive environmental management and promote green agricultural development.

show abstract

“…However, fewer studies considered the use of censuses conducted only in a sample of the target area (defined here as sample counts) in monitoring population trends and estimated the proportion of sites of the total area required to detect trends with similar power to total counts. Sample counts have been described as imprecise (Stoddard et al 1998, Yoccoz et al 2001), as it has been shown for instance that a limited number of sites generally enables detection of only strong declines in the population (Sewell et al 2012, Wagner et al 2022), unless the species is particularly abundant and easy to detect (Ficetola et al 2018). The efficacy of a particular sampling design, such as sample counts, in detecting population dynamics is closely intertwined with the study system (Weiser et al 2019, White 2019).…”

Section: Introductionmentioning

confidence: 99%

Monitoring wildlife population trends with sample counts: a case study on the Alpine ibex (Capra ibex)

Panaccio,

Brambilla,

Bassano

et al. 2023

Wildlife Biology

View full text Add to dashboard Cite

Monitoring population dynamics is of fundamental importance in conservation but assessing trends in abundance can be costly, especially in large and rough areas. Obtaining trend estimations from counts performed in only a portion of the total area (sample counts) can be a cost‐effective method to improve the monitoring and conservation of species difficult to count.We tested the effectiveness of sample counts in monitoring population trends of wild animals, using as a model population the Alpine ibex Capra ibex in the Gran Paradiso National Park (Italy), both with computer simulations and using historical count data collected over the last 65 years. Despite sample counts failed to correctly estimate the true population abundance, sampling half of the target area could reliably monitor the trend of the target population. In case of strong changes in abundance, an even lower proportion of the total area could be sufficient to identify the direction of the population trend. However, when there is a high yearly trend variability, the required number of samples increases and even counting in the entire area can be ineffective to detect population trends. The effect of other parameters, such as which portion of the area is sampled and detectability, was lower, but these should be tested case by case.Sample counts could therefore constitute a viable alternative to assess population trends, allowing for important, cost‐effective improvements in the monitoring of wild animals of conservation interest.

show abstract

The statistical power to detect regional temporal trends in riverine contaminants in the Chesapeake Bay Watershed, USA

Cited by 8 publications

References 57 publications

A Two-Stage Interpretable Machine Learning Framework for Accurate Prediction of Trace Pollutants: With an Application to Microcystin

A Two-Stage Interpretable Machine Learning Framework for Accurate Prediction of Trace Pollutants: With an Application to Microcystin

Decoupling agriculture pollution and carbon reduction from economic growth in the Yangtze River Delta, China

Monitoring wildlife population trends with sample counts: a case study on the Alpine ibex (Capra ibex)

Contact Info

Product

Resources

About