Upload README.md with huggingface_hub

README.md

@@ -20,15 +20,16 @@ pretty_name: "TorNet-Temporal: Temporal Dual-Pol NEXRAD Radar for Tornado Detect
 
 # TorNet-Temporal: Temporal Dual-Pol NEXRAD Radar for Tornado Detection
 
-A large-scale dataset of storm-centered NEXRAD WSR-88D radar sequences for tornado detection and prediction, featuring **24-channel dual-polarimetric** data across **12 temporal frames**.
+A large-scale dataset of storm-centered NEXRAD WSR-88D radar sequences for tornado detection and prediction, featuring **24-channel dual-polarimetric** data across **variable-length temporal sequences**.
 
 ## Dataset Summary
 
-- **24,861 storm events** from NEXRAD Level-II radar archives (2003-2023)
-- **12 consecutive radar scans** per event (~4-5 min cadence, ~1 hour total)
+- **24,862 storm events** from NEXRAD Level-II radar archives (2013-2022)
+- **8-22 consecutive radar scans** per event (~4-5 min cadence, ~45-90 min total; median 13 frames)
 - **24 channels**: 6 dual-pol radar products x 4 elevation angles
 - **128x128 spatial grid** at 1km resolution, storm-centered
-- **3 categories**: TOR (tornado, 7,301), WRN (tornado-warned but no tornado, 3,814), NUL (null/no severe, 13,746)
+- **3 categories**: TOR (tornado), WRN (tornado-warned but no tornado), NUL (null/no severe)
+- **130 unique NEXRAD radar sites** across CONUS
 
 ## Data Format
 
@@ -36,8 +37,8 @@ Each event is stored as a compressed NumPy archive (`sequence.npz`) containing:
 
 | Key | Shape | Type | Description |
 |-----|-------|------|-------------|
-| `data` | (12, 24, 128, 128) | float32 | Radar volume sequence |
-| `scan_times` | (12,) | str | UTC timestamps per frame |
+| `data` | (T, 24, 128, 128) | float32 | Radar volume sequence (T varies 8-22, median 13) |
+| `scan_times` | (T,) | str | UTC timestamps per frame |
 | `center_time` | scalar | str | Event center time |
 | `lat` | scalar | float64 | Storm center latitude |
 | `lon` | scalar | float64 | Storm center longitude |
@@ -67,15 +68,18 @@ The 24 channels represent 6 dual-polarimetric radar products at 4 elevation angl
 
 ## Recommended Splits
 
-Year-based splitting prevents data leakage from correlated storm environments:
+We recommend **year-based splitting** to prevent data leakage from correlated storm environments. Events from the same storm system can appear within hours of each other at the same radar site, so random splitting risks leaking information across splits.
 
-| Split | Years | Events | Purpose |
-|-------|-------|--------|---------|
-| Train | 2008-2021 | ~19,061 | Model training |
-| Validation | 2022 | ~2,117 | Hyperparameter tuning |
-| Test | 2023 | ~3,685 | Final evaluation |
+| Split | Years | Approx. Events | Purpose |
+|-------|-------|----------------|---------|
+| Train | 2013-2021 | ~21,800 | Model training |
+| Test | 2022 | ~3,040 | Final evaluation |
 
-**Note**: Pre-2008 data may have quality issues (constant fill values). We recommend filtering by standard deviation threshold.
+For train/validation splitting, we recommend holding out one earlier year (e.g., 2021) as a validation set.
+
+A `catalog.csv` is included with TorNet-compatible train/test assignments if you prefer to match TorNet's splitting methodology. Events not in the catalog can be assigned to either split.
+
+**Important**: All data is from 2013 onwards (post dual-pol upgrade), so there are no legacy single-pol-only events. All 24 channels contain real data in every event.
 
 ## Usage
 
@@ -86,12 +90,17 @@ from pathlib import Path
 
 # Load a single event
 event = np.load("tornet_1000855_TOR/sequence.npz")
-data = event["data"]          # (12, 24, 128, 128)
+data = event["data"]          # (T, 24, 128, 128), T varies per event
 label = int(event["label"])   # 1 for tornado
 category = str(event["category"])  # "TOR"
+print(f"Frames: {data.shape[0]}, Label: {label}, Category: {category}")
 
-# Select 8 consecutive frames for model input
-frames = data[2:10]           # (8, 24, 128, 128)
+# Select 8 consecutive frames for model input (pad/truncate as needed)
+T = data.shape[0]
+if T >= 8:
+    frames = data[:8]                         # take first 8
+else:
+    frames = np.pad(data, ((0, 8 - T), (0,0), (0,0), (0,0)))  # zero-pad
 
 # Convert to PyTorch tensor: (C, T, H, W) for 3D CNN
 tensor = torch.from_numpy(frames).float()  # (8, 24, 128, 128)
@@ -111,6 +120,14 @@ Using dual-head architecture (detection + prediction heads) on this dataset:
 
 Temporal context (8 frames vs 1 frame) provides a significant boost, especially for detection (+0.049 AUC).
 
+## Data Quality Notes
+
+- **Variable time steps**: Events contain 8-22 frames (not a fixed number). Your DataLoader should pad or truncate to a uniform length.
+- **KDP fill values**: KDP channels (20-23) have a slightly elevated fill rate (~6-7%) compared to other channels (<1%). This is normal -- KDP is a derived product that fails to compute in some conditions.
+- **26 corrupt files**: A small number of NUL-category events (~0.1%) have corrupted NPZ files. These should be caught by a try/except in your data loader.
+- **0.5-degree elevation gaps**: ~3% of events have missing VEL/SW data at the lowest elevation (0.5 deg) due to SAILS/MESO radar scan strategies. The model should learn to handle these naturally.
+- **No dual-pol leakage**: All data is post-2013 (after the NEXRAD dual-pol upgrade completed), so dual-pol channel availability cannot serve as a proxy for era or data quality.
+
 ## Data Source
 
 Raw radar data sourced from the [Unidata NEXRAD Level-II Archive](https://www.unidata.ucar.edu/data/radar.html) on AWS S3 (`s3://unidata-nexrad-level2`). Tornado reports from the [Storm Prediction Center](https://www.spc.noaa.gov/wcm/) storm reports database.