A comparative analysis of deep learning models for soil temperature prediction in cold climates
Bibliographic record
Abstract
Abstract Accurate soil temperature prediction in cold climates is crucial for optimizing agricultural practices, hydrological processes, water resource management, minimizing frost damage, and mitigating flood risks. The capacity of deep learning methods to capture intricate patterns and relationships in climate data enhances the accuracy of soil temperature predictions and offers substantial benefits for reducing climate change impacts. In the present study, a comparative analysis of different deep learning techniques, including long short-term memory (LSTM), convolutional neural network (CNN), and multi-layer perceptron (MLP), for predicting the soil temperature is provided. The study examined cold climate areas across Canada, from snowy regions to Arctic conditions. Input datasets were considered both as time series and shuffled order. To comprehensively evaluate the predictive approaches for soil temperature, four machine learning (ML) models—CNN, LSTM, MLP in time series, and MLP on shuffled data—were employed. The results showed ML models using input data as time series have struggled with accurate soil temperature prediction, especially in very cold and polar climates, likely due to the presence of ice layers on the soil, limiting fluctuations near the freezing point. The normalized RMSE (NRMSE) for the CNN, LSTM, and MLP was calculated to be 8.6%. 7.4%, and 6.9%, respectively, and the scatter index (SI) for CNN, LSTM, and MLP was calculated to be 1.0%, 0.9%, and 0.9%, respectively. On the other hand, MLP-shuffled that employs shuffled input data outperformed others with an NRMSE of 5.4% and an SI of 0.7%, by creating a generalized data representation, free from presentation sequence bias. This study showed that predicting soil temperature in very cold climates poses a challenge for machine learning, yet the MLP-shuffled model excels, attaining superior accuracy through the creation of a generalized data representation independent of the sample sequence.
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 distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 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 teacher head, 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".