Explainable AI approach with original vegetation data classifies spatio-temporal nitrogen in flows from ungauged catchments to the Great Barrier Reef DOI Creative Commons

Cherie M. O’Sullivan,

Ravinesh C. Deo, Afshin Ghahramani

et al.

Scientific Reports, Journal Year: 2023, Volume and Issue: 13(1)

Published: Oct. 24, 2023

Transfer of processed data and parameters to ungauged catchments from the most similar gauged counterpart is a common technique in water quality modelling. But catchment similarities for Dissolved Inorganic Nitrogen (DIN) are ill posed, which affects predictive capability models reliant on such methods simulating DIN. Spatial proxies classify DIN responses demonstrated solution, yet their applicability unexplored. We adopted neural network pattern recognition model (ANN-PR) explainable artificial intelligence approach (SHAP-XAI) match all that flow Great Barrier Reef ones based proxy spatial data. Catchment suitability was verified using (ANN-WQ) simulator trained datasets, tested by matched unsupervised learning scenarios. show discriminating training regime benefits ANN-WQ simulation performance scenarios ( p< 0.05). This phenomenon demonstrates useful tool with regimes. Catchments lacking similarity identified as priority monitoring areas gain observed regimes Reef, Australia.

Language: Английский

Deep learning for water quality DOI
Wei Zhi, Alison P. Appling, Heather E. Golden

et al.

Nature Water, Journal Year: 2024, Volume and Issue: 2(3), P. 228 - 241

Published: March 12, 2024

Language: Английский

Citations

66
Improving River Routing Using a Differentiable Muskingum‐Cunge Model and Physics‐Informed Machine Learning DOI Creative Commons
Tadd Bindas, Wen‐Ping Tsai, Jiangtao Liu

et al.

Water Resources Research, Journal Year: 2024, Volume and Issue: 60(1)

Published: Jan. 1, 2024

Abstract Recently, rainfall‐runoff simulations in small headwater basins have been improved by methodological advances such as deep neural networks (NNs) and hybrid physics‐NN models—particularly, a genre called differentiable modeling that intermingles NNs with physics to learn relationships between variables. However, hydrologic routing simulations, necessary for simulating floods stem rivers downstream of large heterogeneous basins, had not yet benefited from these it was unclear if the process could be via coupled NNs. We present novel method ( δ MC‐Juniata‐hydroDL2) mimics classical Muskingum‐Cunge model over river network but embeds an NN infer parameterizations Manning's roughness n ) channel geometries raw reach‐scale attributes like catchment areas sinuosity. The trained solely on hydrographs. Synthetic experiments show while geometry parameter unidentifiable, can identified moderate precision. With real‐world data, produced more accurate long‐term results both training gage untrained inner gages larger subbasins (>2,000 km 2 than either machine learning assuming homogeneity, or simply using sum runoff subbasins. parameterization short periods gave high performance other periods, despite significant errors inputs. learned pattern consistent literature expectations, demonstrating framework's potential knowledge discovery, absolute values vary depending periods. traditional models improve national‐scale flood simulations.

Language: Английский

Citations

28
Remote sensing-enabled machine learning for river water quality modeling under multidimensional uncertainty DOI Creative Commons
Saiful Haque Rahat, Todd Steissberg, Won Chang

et al.

The Science of The Total Environment, Journal Year: 2023, Volume and Issue: 898, P. 165504 - 165504

Published: July 15, 2023

Language: Английский

Citations

30
A differentiable, physics-informed ecosystem modeling and learning framework for large-scale inverse problems: demonstration with photosynthesis simulations DOI Creative Commons
Doaa Aboelyazeed, Chonggang Xu, Forrest M. Hoffman

et al.

Biogeosciences, Journal Year: 2023, Volume and Issue: 20(13), P. 2671 - 2692

Published: July 6, 2023

Abstract. Photosynthesis plays an important role in carbon, nitrogen, and water cycles. Ecosystem models for photosynthesis are characterized by many parameters that obtained from limited situ measurements applied to the same plant types. Previous site-by-site calibration approaches could not leverage big data faced issues like overfitting or parameter non-uniqueness. Here we developed end-to-end programmatically differentiable (meaning gradients of outputs variables used model can be efficiently accurately) version process representation within Functionally Assembled Terrestrial Simulator (FATES) model. As a genre physics-informed machine learning (ML), couple physics-based formulations neural networks (NNs) learn parameterizations (and potentially processes) observations, here rates. We first demonstrated framework was able correctly recover multiple assumed values concurrently using synthetic training data. Then, real-world dataset consisting different functional types (PFTs), learned performed substantially better greatly reduced biases compared literature values. Further, allowed us gain insights at large scale. Our results showed carboxylation rate 25 ∘C (Vc,max25) more impactful than factor representing limitation, although tuning both helpful addressing with default This enable substantial improvement our capability reduce ecosystem modeling scales.

Language: Английский

Citations

26
A coupled model to improve river water quality prediction towards addressing non-stationarity and data limitation DOI
Shengyue Chen, Jinliang Huang, Peng Wang

et al.

Water Research, Journal Year: 2023, Volume and Issue: 248, P. 120895 - 120895

Published: Nov. 20, 2023

Language: Английский

Citations

25
HydroEcoLSTM: A Python package with graphical user interface for hydro-ecological modeling with long short-term memory neural network DOI Creative Commons
Van Tam Nguyen, Vinh Ngoc Tran, Hoang Tran

et al.

Ecological Informatics, Journal Year: 2025, Volume and Issue: 85, P. 102994 - 102994

Published: Jan. 5, 2025

Language: Английский

Citations

1
Metamorphic testing of machine learning and conceptual hydrologic models DOI Creative Commons
Peter Reichert, Kai Ma,

Marvin Höge

et al.

Hydrology and earth system sciences, Journal Year: 2024, Volume and Issue: 28(11), P. 2505 - 2529

Published: June 13, 2024

Abstract. Predicting the response of hydrologic systems to modified driving forces beyond patterns that have occurred in past is high importance for estimating climate change impacts or effect management measures. This kind prediction requires a model, but impossibility testing such predictions against observed data makes it difficult estimate their reliability. Metamorphic offers methodology assessing models validation with real data. It consists defining input changes which expected responses are assumed be known, at least qualitatively, and model behavior consistency these expectations. To increase gain information reduce subjectivity this approach, we extend multi-model approach include sensitivity analysis training calibration options. allows us quantitatively analyze differences between different structures options addition qualitative test In our case study, apply selected conceptual machine learning hydrological calibrated basins from CAMELS set. Our results confirm superiority over regarding quality fit during periods. However, also find inputs can deviate expectations magnitude, even sign depend on addition, cases all passed metamorphic test, there quantitative structures. demonstrates usual calibration–validation identify potential problems stimulate development improved models.

Language: Английский

Citations

6
Performance evaluation of deep learning based stream nitrate concentration prediction model to fill stream nitrate data gaps at low-frequency nitrate monitoring basins DOI Creative Commons
Gourab Saha, Chaopeng Shen, J. M. Duncan

et al.

Journal of Environmental Management, Journal Year: 2024, Volume and Issue: 357, P. 120721 - 120721

Published: April 1, 2024

Language: Английский

Citations

5
Deep dive into hydrologic simulations at global scale: harnessing the power of deep learning and physics-informed differentiable models (δHBV-globe1.0-hydroDL) DOI Creative Commons

Dapeng Feng,

Hylke E. Beck, Jens de Bruijn

et al.

Geoscientific model development, Journal Year: 2024, Volume and Issue: 17(18), P. 7181 - 7198

Published: Sept. 26, 2024

Abstract. Accurate hydrologic modeling is vital to characterizing how the terrestrial water cycle responds climate change. Pure deep learning (DL) models have been shown outperform process-based ones while remaining difficult interpret. More recently, differentiable physics-informed machine with a physical backbone can systematically integrate equations and DL, predicting untrained variables processes high performance. However, it unclear if such are competitive for global-scale applications simple backbone. Therefore, we use – first time at this scale (full name δHBV-globe1.0-hydroDL, shortened δHBV here) simulate rainfall–runoff 3753 basins around world. Moreover, compare purely data-driven long short-term memory (LSTM) model examine their strengths limitations. Both LSTM provide daily simulation capabilities in global basins, median Kling–Gupta efficiency values close or higher than 0.7 (and 0.78 subset of 1675 long-term discharge records), significantly outperforming traditional models. regionalized demonstrated stronger spatial generalization ability (median KGE 0.64) parameter regionalization approach 0.46) even ungauged region tests across continents. Nevertheless, relative LSTM, was hampered by structural deficiencies cold polar regions, highly arid significant human impacts. This study also sets benchmark estimates world builds foundation improving simulations.

Language: Английский

Citations

5
Identifying Structural Priors in a Hybrid Differentiable Model for Stream Water Temperature Modeling DOI Creative Commons
Farshid Rahmani, Alison P. Appling, Dapeng Feng

et al.

Water Resources Research, Journal Year: 2023, Volume and Issue: 59(12)

Published: Dec. 1, 2023

Abstract Although deep learning models for stream temperature ( T s ) have recently shown exceptional accuracy, they limited interpretability and cannot output untrained variables. With hybrid differentiable models, neural networks (NNs) can be connected to physically based equations (called structural priors) intermediate variables such as water source fractions (specifying what portion of is groundwater, subsurface, surface flow). However, it unclear if outputs are meaningful when only physics imposed, priors enough impacts identifiable from data. Here, we tested four alternative describing basin‐scale memory instream heat processes in a model where NNs freely estimate the fractions. We evaluated models’ abilities predict baseflow ratio. The exhibited noticeably different behaviors these two metrics their tradeoffs, with some dominating others. Therefore, better identified. Moreover, testing yielded valuable insights: having separate shallow subsurface flow component matches observations, recency‐weighted averaging past air calculating resulted prediction than traditionally employed simple averaging. also highlight limitations insufficient physical constraints implemented: internal (water fractions) may not adequately constrained by single target variable (stream temperature) alone. To ensure significance fluxes, one either employ multivariate data selection, or include more priors.

Language: Английский

Citations

11