Convolutional Neural Network‐Based Insights Into Extreme Precipitation Regional Dynamics Over Central Africa Using Moisture Flux Patterns
Bibliographic record
Abstract
Abstract Understanding the atmospheric drivers of extreme rainfall is essential for improving regional adaptation strategies. In Central Africa, previous studies widely emphasize large‐scale influences, often overlooking complex regional processes. In this study, we combined the capacity of vertically integrated moisture flux convergence (VIMFC) to capture information from the entire atmospheric column with the ability of convolutional neural networks to learn complex, nonlinear patterns from large and intricate data sets, thereby unlocking VIMFC's potential as an effective regional predictor. Our machine learning model successfully identifies 95% of observed extreme rainfall events using VIMFC as input. The CNN's predictions, interpreted using the layer‐wise relevance propagation method, highlight its ability to capture spatiotemporal distributions of strong moisture convergence and divergence linked to extreme rainfall and drought events. Over the past two decades, we have observed a clear increase in the frequency of extreme precipitation moisture flux patterns (EPMFPs), which aligns with rising extreme rainfall occurrences. On EPMFP days, we detect stronger low‐level moisture inflow from the Atlantic Ocean and enhanced moisture supply driven by a deeper and more intense Congo Basin convective cell. This is supported by evapotranspiration from the basin's dense vegetation, acting as a continental moisture reservoir. Midlevel analysis reveals more moisture retention over the region during EPMFPs, linked to the positioning and strength of the African easterly jets. While both EPMFP and non‐EPMFP composites show similar meridional moisture transport, distinct zonal moisture outflow and inflow patterns highlight the dynamical differences between the two regimes.
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) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 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) | 0.001 | 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".