Transferable soft-sensors for predicting nitrate in diverse watersheds
Bibliographic record
Abstract
• PR-TR predicts nitrate using minimal data and low-cost surrogate sensors. • Achieved NSE ∼0.51 with only 15 samples when site difference was <100%. • Outperformed ANN and MLR that required >40 samples for similar accuracy. • Identified suitable models without needing nitrate data from new sites. Understanding the spatial and temporal dynamics of nitrates is crucial to mitigate pollution that causes eutrophication and poor aquatic health. However, in-situ sensors for direct nitrate detection are often limited by high costs, frequent maintenance requirements, and low sensitivity. Soft-sensing has emerged as a promising alternative, where nitrates are predicted using surrogate sensors using models or machine learning. This study addresses a central challenge with soft-sensors: their transferability to sites with limited or no training data. We propose a transferable framework that predicts nitrate concentrations using only a small number of training data points together with simple, low-cost sensors such as electrical conductivity, temperature, and turbidity. The approach selects a pre-trained model (PR-TR) from a large model library using only historical surrogate data, with site similarity determined through Euclidean distance and a relative difference metric. For sites with relative differences below 100%, the PR-TR model achieved an average NSE of 0.51 using only 15 data points. For more dissimilar sites, higher data requirements and careful tuning of the learning rate (0.01) were essential, yet PR-TR still outperformed traditional approaches. Compared with artificial neural networks (ANN) and multiple linear regression (MLR), which required more than 40 data points to reach similar performance, PR-TR delivered robust and efficient predictions using significantly fewer data samples. The model selection process identified suitable PR-TR models capable of achieving positive NSE values even without nitrate data from the validation site. These findings demonstrate that PR-TR offers a practical, data-efficient method for reliable water quality monitoring.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".