Generalization of multiple depths soil temperature estimation using LSTM and CNN
Bibliographic record
Abstract
Accurate estimation of soil temperature (ST) as a key factor in soil physical, hydrological, chemical, and biological processes is essential in agriculture and Earth system science. Different machine learning methods have been used in previous studies in ST estimation and have gained remarkable progress. However, such studies are not only rare due to the scarce multi-depth ST data, but also the used modeling approaches are prone to modeling errors nor generalization, especially at deeper depths. The main purpose of this study was to assess generalizability of trained deep learning model in multi-depth ST estimation in different land covers, climates and soils. In this regard, this study aimed to improve the accuracy of multiple depths ST estimation in multi-feature-temporal modeling of LSTM (Long Short-Term Memory) and variable-based modeling of CNN (Convolutional Neural Network). Applicability of trained DL model in multiple depths ST time series estimation of other regions with same condition was evaluated to assess its generalization potential. Daily time series of meteorological and soil data at depths of 5, 10, 20, 50 and 100 cm were gathered from 14 stations of U.S. Climate Reference Network with various land covers, climates and soils. The results indicated that estimation of ST with time series of meteorological parameters + ST (of each soil depth or upper depth) performed best with <5 % error in almost all regions and soil depths. Evaluation of the generalization potential revealed that in different land cover or climate of two stations with same soil, same range of air temperature of two stations led to accurate ST estimation in the monitoring station based on the trained model from the main station. LSTM model performed superior than CNN model at different soil depths with considerable improved accuracy in deeper depths. Therefore, LSTM model was highly recommended to estimate ST at different soil depths.
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 | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| 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".