nakas/parsimonious_snow_snotel_era5_data_fetcher

🏔️ Snow Prediction Data Fetcher

An interactive system for fetching and fusing SNOTEL snow measurements with ERA5 reanalysis data to create training datasets for AI snow prediction models.

Features

📊 Data Sources

• —SNOTEL Stations: Historical snow depth, snow water equivalent, temperature, and precipitation from USDA NRCS network
• —ERA5 Reanalysis: Weather variables aligned with ECMWF Open Data for real-time forecast compatibility
• —Data Fusion: Spatial and temporal alignment with advanced feature engineering

🎯 Key Capabilities

• —Interactive Interface: Three-tab Gradio web app for data exploration and visualization
• —Feature Engineering: 90+ features including lags, rolling averages, and derived meteorological variables
• —Quality Control: Automated data filtering and validation
• —ML-Ready Output: Normalized datasets ready for machine learning training

🌟 ECMWF Alignment

Variables are specifically chosen to match ECMWF Open Data real-time forecasts:

• —2m Temperature & Dewpoint
• —Surface & Sea Level Pressure
• —Wind Components (10m U/V)
• —Total Precipitation
• —Cloud Cover & Boundary Layer Height
• —Skin Temperature

Usage

Quick Start

1. 1.SNOTEL Data Tab: Select stations and date ranges to fetch snow measurements
2. 2.ERA5 Data Tab: Define geographic regions to fetch weather reanalysis data
3. 3.Data Fusion Tab: Combine data sources to create ML training datasets

Sample Output

• —5,000+ training samples from multiple Colorado SNOTEL stations
• —90 engineered features optimized for snow prediction
• —4 years of data (2020-2023) with seasonal coverage
• —High-quality fusion with 39.6% snow coverage and realistic patterns

Technical Details

Architecture

• —SNOTEL Fetcher: Downloads station data with caching and error handling
• —ERA5 Fetcher: Processes reanalysis data with daily aggregation
• —Data Fusion: Advanced temporal/spatial alignment and feature engineering
• —Gradio Interface: Interactive visualization and dataset creation

Data Pipeline

1. 1.Fetch SNOTEL snow measurements from multiple stations
2. 2.Download ERA5 reanalysis data for matching locations/times
3. 3.Align data spatially (nearest grid points) and temporally (daily)
4. 4.Engineer features: lags, rolling windows, derived variables
5. 5.Apply quality filters and normalization
6. 6.Output training-ready datasets

Applications

• —Snow Forecasting: Train models to predict snowfall using weather forecasts
• —Avalanche Risk: Assess snow stability and accumulation patterns
• —Water Resources: Predict snowpack for water supply management
• —Climate Research: Analyze snow-weather relationships and trends

Data Quality

The system produces high-quality training data with:

• —Strong correlations: Temperature variables show 0.79-0.82 correlation with snow depth
• —Seasonal patterns: Realistic snow accumulation/melt cycles
• —Geographic diversity: Multiple elevation zones and climate conditions
• —Temporal coverage: Multi-year datasets spanning various weather conditions

Getting Started

Simply select your parameters in the interface:

1. 1.Choose SNOTEL stations of interest
2. 2.Set date ranges for historical analysis
3. 3.Configure geographic bounds for ERA5 data
4. 4.Generate fused datasets for model training

Perfect for researchers, meteorologists, and data scientists working on snow prediction and climate modeling applications.