Web Based Tools for Reproducible Preparation and Execution of an Energy Balance Snowmelt Model
Abstract
Snow modeling plays an important role in the prediction of seasonal runoff and water supply forecasting for water resources management in snow-fed river basins. Physically based modeling is generally assumed better suited to reproduce snow processes under changing conditions. However, challenges exist in the application of snowmelt models that make them hard to reproduce, in terms of (1) tracking the preparation of model inputs, (2) repeating the execution of model code to duplicate model outputs and (3) reproducing analyses used to report results. This presentation will describe cyberinfrastructure approaches that have been taken to address these challenges and improve the preparation of inputs for, execution, and analysis of results from the Utah Energy Balance (UEB) snowmelt model, a physically based model that we are working to apply and incorporate into water supply forecasts for test watersheds in the Colorado River Basin. For snow simulation, we prepared model inputs and executed the UEB model using HydroDS, a set of software tools deployed as web services that extract data from nationally available data sources to prepare inputs for UEB. HydroDS also generates a Python script that can be used to reproduce the input data preparation. A web based app (UEB app), linked to HydroShare provides an easy to use interface to these services. HydroShare is a web based hydrologic information system that supports the sharing of hydrologic data, model and analysis tools. All the data and the script from HydroDS are stored in HydroShare. For discharge simulation, we used snowmelt output from UEB to feed a distributed runoff model. Sciunit, which is software for creating self-contained and annotated containers that describe and package computational experiments, was used to create a container that enables reproduction of this model execution. Model outputs were analyzed using Python scripts that can be run and documented in a Jupyter Notebook in a JupyterHub server linked to HydroShare. This work illustrates how reproducibility of the complete modeling cycle can be enhanced using web based tools. The contents can be shared with other users in HydroShare to repeat or build on the work and can be permanently published to receive a digital object identifier for citation in papers to fulfill the open data mandate.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2018
- Bibcode:
- 2018AGUFM.C13J1256G
- Keywords:
-
- 0798 Modeling;
- CRYOSPHEREDE: 1805 Computational hydrology;
- HYDROLOGYDE: 1920 Emerging informatics technologies;
- INFORMATICSDE: 1978 Software re-use;
- INFORMATICS