Large scale prediction of groundwater nitrate concentrations from spatial data using machine learning

Knoll, Lukas; Breuer, Lutz; Bach, Martin

doi:10.1016/j.scitotenv.2019.03.045

Cited by 176 publications

(75 citation statements)

References 49 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The redox conditions tend to be less anaerobic to intermediate in the Geest areas in contrast to the lowlands where strong anaerobic conditions prevail, which was also reported by Eschenbach et al (2018). Conversely, lowlands with sand and gravel deposits in the middle and southern part of Germany often show nitrate-polluted groundwater, due to a lack of cover layer, weakly anaerobic to intermediate redox conditions and intensive agricultural land use (Grimm-Strele et al 2008, Wendland et al 2008, Kuhr et al 2013, Knoll et al 2019. For the pore aquifers of the northern Alpine foothills, the RF NO3 model predicts large areas of nitrate-polluted groundwater, despite generally well-developed cover layers and the associated high protective function of the aquifers in this region.…”

Section: Discussionmentioning

confidence: 79%

“…However, the performance of the model can be regarded as good, especially with respect to the large scale of the application and the relatively low density of monitoring sites. Other studies on large-scale prediction of groundwater nitrate concentrations result in comparable model performances , Wheeler et al 2015, Ransom et al 2017, Knoll et al 2019. The predicted grid map of the NO 3 concentration in groundwater is shown in figure 6, a corresponding density plot can be found in figure S11.…”

Section: Spatial Distribution Of Groundwater Nitrate Concentrationmentioning

confidence: 71%

“…The redox condition is by far the most relevant predictor, followed by the hydrogeological units and the percentage of arable land. The high relevance of the hydrogeological units and the percentage of arable land has also been found in Knoll et al (2019). Ransom et al (2017) used the redox parameters manganese and oxygen for nitrate prediction, which in that case turned out to be the most relevant predictors.…”

Section: Spatial Distribution Of Groundwater Nitrate Concentrationmentioning

confidence: 82%

“…Since information on largescale groundwater flow conditions is limited, we used a simplified procedure to determine contributing areas for each monitoring site, i.e. a circular buffer of 1000 m radius results in best model performances (Knoll et al 2019). The spatial predictors are compiled within this buffer zone and are linked to the monitoring site.…”

Section: Buffer Analysismentioning

confidence: 99%

“…Comparisons of different machine learning techniques in terms of prediction performance have shown that tree-based models such as RF outperform others (Wang et al 2016, Knoll et al 2019. Such tree-based ML techniques have already been used in several studies on large-scale prediction of groundwater nitrate concentrations (Nolan et al 2014, Rodriguez-Galiano et al 2014, 2018, Wheeler et al 2015, Ransom et al 2017, Tesoriero et al 2017, Knoll et al 2019, Ouedraogo et al 2019. In terms of the estimation of groundwater redox conditions, Close et al (2016) and Wilson et al (2018) performed linear discriminant analyses.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Nation-wide estimation of groundwater redox conditions and nitrate concentrations through machine learning

Knoll

Bach

2020

Environ. Res. Lett.

Self Cite

View full text Add to dashboard Cite

The protection of water resources and development of mitigation strategies require large-scale information on water pollution such as nitrate. Machine learning techniques like random forest (RF) have proven their worth for estimating groundwater quality based on spatial environmental predictors. We investigate the potential of RF and quantile random forest (QRF) to estimate redox conditions and nitrate concentration in groundwater (1 km × 1 km resolution) using the European Water Framework Directive groundwater monitoring network as well as spatial environmental information available throughout Germany. The RF model for nitrate achieves a good predictive performance with an R 2 of 0.52. Dominant predictors are the redox conditions in the groundwater body, hydrogeological units and the percentage of arable land. An uncertainty assessment using QRF shows rather large uncertainties with a mean prediction interval (MPI) of 53.0 mg l −1 . This study represents the first nation-wide data-driven assessment of the spatial distribution of groundwater nitrate concentrations for Germany.

show abstract

Section: Discussionmentioning

confidence: 79%

Section: Spatial Distribution Of Groundwater Nitrate Concentrationmentioning

confidence: 71%

Section: Spatial Distribution Of Groundwater Nitrate Concentrationmentioning

confidence: 82%

Section: Buffer Analysismentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Nation-wide estimation of groundwater redox conditions and nitrate concentrations through machine learning

Knoll

Bach

2020

Environ. Res. Lett.

Self Cite

View full text Add to dashboard Cite

show abstract

Isotopic variation in groundwater across the conterminous United States – Insight into hydrologic processes

Stahl

Gehring

Jameel

2020

Hydrological Processes

View full text Add to dashboard Cite

Groundwater supplies a significant proportion of water use in the US and is critical to the maintenance of healthy ecosystems and environmental processes, thus characterizing aquifer hydrology is important to managing and preserving these resources. While groundwater isotopes provide insight into hydrologic and ecologic processes, their application is limited to where measurements exist. To help overcome this limitation we used the random forest algorithm to develop a predictive model for shallow groundwater isotopes in the conterminous US. Our model uses environmental variables (e.g. temperature, elevation, precipitation isotopes) as predictors. We used our model to develop the first shallow groundwater isoscape of δ 2 H and δ 18 O for the conterminous US. We describe the patterns in groundwater isotopes using both observations and our modeled isoscape. We find that throughout much of the Eastern US, groundwater isotopes are close to annual amount weighted precipitation, while groundwater isotopes are significantly depleted relative precipitation across much of the High Plains and Western US. Furthermore, we compare the observations compiled for this study to isotopes of precipitation, which allows us to determine the relative recharge efficiency (i.e. ratio of groundwater recharge to precipitation) between seasons and the proportion of annual recharge that occurs in a given season. Our findings suggest that winter recharge is generally more efficient than summer recharge; however, the dominant recharge season is more varied as it is the product of both seasonal recharge efficiency and the seasonal timing of precipitation. Parts of the central US have summer dominant recharge, which is likely the result of heavy summer precipitation/nocturnal summer precipitation. Interestingly, parts of coastal California appear to have summer dominant recharge, which we suggest could be due to recharge from fog-drip. Our results summarize spatial patterns in groundwater isotopes This article is protected by copyright. All rights reserved. This is the author manuscript accepted for publication and has undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as

show abstract

The potential for nitrate removal during infiltration: Mapping with machine learning informed by field and laboratory experiments

Gorski

Fisher

Dailey³

et al. 2022

Hydrological Processes

View full text Add to dashboard Cite

Managed aquifer recharge (MAR) can increase groundwater supply and, under some conditions, improve groundwater quality simultaneously. However, the conditions under which water quality improvements can be achieved during infiltration for MAR have not been systematically examined in a spatially explicit manner. This work aims to address that gap by synthesizing observations from laboratory tests, field experiments and operational facilities at four MAR sites within the Pajaro Valley, California, USA to develop a predictive model of nitrate removal during infiltration. We compare two machine learning approaches; random forest and boosted regression trees, to multiple linear regression. The preferred model uses boosted regression trees, with four predictor variables (initial nitrate concentration of infiltrating water, residence time in the soil, soil organic carbon content, and soil percent clay). We apply this model to simulate the spatial distribution of potential nitrate removal (NRp) across a heterogeneous and mixed‐use landscape, finding that >87% of the region has the potential to remove nitrate during infiltration. To link potential nitrate removal capacity to available water for MAR, we combine a map of NRp with independently simulated hillslope runoff, and find that potential load reduction is highest in urban areas (median = 18.8 kg‐N/yr) where large runoff volumes are collocated with soils having high nutrient cycling capacity compared to forested and agricultural areas (median = 1.6 and 3.5 kg‐N/yr, respectively). We analyse potential load reduction across dry, normal, and wet precipitation scenarios, which shows that urban areas have the potential to yield large load reductions (median = 11.3 kg‐N/yr) even under relatively dry conditions. However, much of the generation of elevated nitrate in runoff is associated with agricultural activity. These results suggest that the urban–agricultural interface represents a crucial link between hydrologic and biogeochemical cycles where water supply and quality goals may be met simultaneously. The technical approach used in this study is highly flexible and can help guide decisions in resource management and identify promising MAR sites. More broadly, this approach also elucidates heterogeneity in biogeochemical cycling across complex landscapes.

show abstract

Large scale prediction of groundwater nitrate concentrations from spatial data using machine learning

Cited by 176 publications

References 49 publications

Nation-wide estimation of groundwater redox conditions and nitrate concentrations through machine learning

Nation-wide estimation of groundwater redox conditions and nitrate concentrations through machine learning

Isotopic variation in groundwater across the conterminous United States – Insight into hydrologic processes

The potential for nitrate removal during infiltration: Mapping with machine learning informed by field and laboratory experiments

Contact Info

Product

Resources

About