Can I use custom model parameters?

Yes, for complex models, the skill allows users to input model parameters via a standardized JSON configuration file.

How are flood events identified?

The skill includes an internal module that uses R-based logic to divide time-series data into specific flood events for detailed analysis.

What models are supported by this skill?

It supports a variety of models including conceptual ones like MarrMot and XAJ, and neural network architectures like LSTM, TCN, and KAN.

Does this skill require specific Julia packages?

Yes, it primarily utilizes HydroTools, DataFrames, and RTableTools for its operations and data management.

What metrics are used for model evaluation?

The skill evaluates performance based on KGE (Kling-Gupta Efficiency), NSE (Nash-Sutcliffe Efficiency), and RMSE, focusing on peak flow and timing accuracy.

Hydrological Forecasting Suite

Name: Hydrological Forecasting Suite
Author: kongdd

bykongdd

•

Data Science & ML

Performs hydrological modeling and streamflow forecasting using Julia-based classical and machine learning models.

About

The hydro-forecast skill provides a comprehensive framework for hydrological modeling in Julia, integrating traditional conceptual models like MarrMot and Xinanjiang (XAJ) with advanced deep learning architectures such as LSTM, TCN, and KAN. It automates the end-to-end workflow of rainfall-runoff simulation, including flood event identification, dataset partitioning into training and validation sets, and model calibration using metrics like NSE and KGE. Designed for water resource engineers and researchers, it generates detailed evaluation reports and visualizations to assess model accuracy on flood peaks and timing.

Key Features

Integration with HydroTools and DataFrames for seamless data handling
Detailed output reporting for flood peak and timing accuracy
Support for hybrid models including MarrMot, XAJ, TCN, LSTM, and KAN
Comprehensive evaluation using KGE, NSE, and RMSE metrics
1 GitHub stars
Automated flood event division and dataset partitioning

Use Cases

Comparing performance between conceptual and deep learning hydrological models
Automating the identification and analysis of historical flood events
Simulating rainfall-runoff processes for watershed management

About

Key Features

Integration with HydroTools and DataFrames for seamless data handling
Detailed output reporting for flood peak and timing accuracy
Support for hybrid models including MarrMot, XAJ, TCN, LSTM, and KAN
Comprehensive evaluation using KGE, NSE, and RMSE metrics
1 GitHub stars
Automated flood event division and dataset partitioning

Use Cases

Comparing performance between conceptual and deep learning hydrological models
Automating the identification and analysis of historical flood events
Simulating rainfall-runoff processes for watershed management