Code Monkey home page Code Monkey logo

kali's Introduction

KALI

Investigating the effect of the calibration data length on the performance of hydrological models

In the present repository, you can find the materials for the paper

Ayzel G., Heistermann M. The effect of calibration data length on the performance of conceptual versus data-driven hydrological models.

which was submitted to Computers & Geosciences.

Idea and workflow

In this study, we want to investigate the effect of calibration data length on the validation performance of different rainfall-runoff models. To this aim, we consequently increase the calibration data length from one to twenty calendar years and investigate how that affects the model skill on a hold-out (validation) period.



We use five models for runoff prediction at hourly temporal resolution:

one conceptual

  1. GR4H -- a conceptual hydrological model. It is a derivative from the GR4J model -- the version for runoff prediction at daily temporal resolution.

and four data-driven models which differ by the type of the computational layer used

  1. Multi-Layer Perceptron (MLP)
  2. Recurrent Neural Network (RNN)
  3. Long Short-Term Memory Network (LSTM)
  4. Gated Recurrent Units Network (GRU)

Code

The code is written in Python programming language (v3.6) using open-source software libraries, such as numpy, pandas, scipy, numba, tensorflow, and keras. The analysis of obtained results was done also using jupyter notebooks and matplotlib plotting library.

You can install all the required dependencies using conda -- an open-source package management system. First, install conda itself, then use the provided environment.yml file to create the isolated environment:

conda env create -f environment.yml

There are three files in the code directory:

  1. experiment.py

- describes the workflow for the main calibration/validation experiment.

  1. metrics.py

- calculates and aggregates the evaluation metrics based on obtained results of streamflow simulation DOI.

  1. analysis.ipynb

- represents the analysis of the effect of calibration data length on the performance of hydrological models.

Models

There are two files in the models directory:

  1. gr4h.py

- holds the code for the GR4H hydrological model.

  1. anns.py

- holds the code for generating data-driven hydrological models based on different architectures of artificial neural networks: MLP, RNN, LSTM, and GRU.

These files are used as modules in experiment.py.

Results

Two files aggregate evaluation metrics for the calibration and validation periods: summary_calibration.npy and summary_validation.npy, respectively. The figures subfolder consists of figures that were generated using the analysis.ipynb jupyter notebook.

Note on data availability

Unfortunately, we cannot directly provide the compiled dataset we use for the presented study due to license restrictions from data suppliers. Thus, the discharge data for Rimbaud River as well as rain gauge records for the area Real Collobrier should request it from the RECOVER research laboratory at INRAE (https://www6.paca.inrae.fr/recover). The data for the SAFRAN reanalysis is also available upon request from Meteo France.

Please, contact us in case you need additional details or support for retrieving the data.

kali's People

Contributors

hydrogo avatar

Stargazers

 avatar  avatar  avatar

Watchers

 avatar

Forkers

iforiv

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    ๐Ÿ–– Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. ๐Ÿ“Š๐Ÿ“ˆ๐ŸŽ‰

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google โค๏ธ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.