Skip to content

Latest commit

 

History

10 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 

Repository files navigation

Dataset for Fluctuation-Aware Attention Network (Flation) in Real-Time Runoff Forecasting

This dataset supports the research on real-time runoff forecasting (RRF) using the Flation model, focusing on a coastal watersheds study area with rainfall-driven runoff regimes. It is designed for training, validating, and testing data-driven RRF models, especially those emphasizing frequency-domain analysis and cross-channel interaction.

Dataset Overview

The dataset consists of high-resolution hourly hydrometeorological and hydrological data from small coastal watersheds, capturing distinct topographic, climatic, and runoff characteristics.

Coastal Watersheds Dataset

  • Study Area: 5 small coastal watersheds (Station IDs: 1015, 703, 708, 693, 626) on Calvert and Hecate Islands (British Columbia, Canada), part of the Northeast Pacific Coastal Temperate Rainforest (NPCTR).
  • Time Range: Continuous hourly observations from October 2013 to September 2019 (specific time ranges per station below).
  • Hydrological Features: Rainfall-dominated (pluvial) runoff, with varying lake regulation effects, terrain slopes (21.7%–40.3%), forest coverage (29.9%–79.8%), and elevation (59–1012 meters).

Station-Specific Time Details

Station ID Time Range Total Time Steps
1015 2014/7/31 15:00 – 2019/10/1 0:00 45296
708 2014/9/9 12:00 – 2019/10/1 0:00 44127
703 2014/9/9 12:00 – 2019/10/1 0:00 44319
693 2014/8/2 13:00 – 2019/10/1 0:00 45248
626 2014/8/2 13:00 – 2019/10/1 0:00 45252

Data Processing

All raw data have undergone standardized preprocessing to ensure reliability and suitability for deep-learning modeling:

  1. Quality Control: Two-tier flagging system (Accepted Value/AV, Suspicious Value-Caution/SVC, Suspicious Value-Discard/SVD) to filter sensor malfunctions, noise, and invalid records.
  2. Uncertainty Exclusion: Excluded watersheds with high data uncertainty (e.g., Watersheds 819 and 844 with ±32%–77% uncertainty) due to unstable monitoring conditions.
  3. Data Integration: Retained valid data from 5 watersheds, unified data formats, and aligned time steps to form a structured dataset for model training.

Data Access

Raw Data Source

Coastal Watersheds Raw Data (Oct 2013–Sep 2019): Hakai Institute Repository

Paper under review – full dataset and code coming soon.

About

No description, website, or topics provided.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors