Informatics Theme: Reproducible Modeling Environment

Reproducible model development and evaluation environment.
WHAT

Problem Statement:What is a suitable model development and evaluation for both existing NWM-subset(s) and new/novel model formulation investigation(s)?

WHY
Goal: Work with NWM-subset(s) and model formulation investigation(s) to design and demonstrate a base-line development and testing environment for reproducible model development and evaluation.

HOW
Approach: Review literature and community practices for reproducible model development and evaluation. Start with existing NWM containerization scheme. Determine appropriate abstraction to encapsulate a variety of hydrologic models and formulations. Implement example in cooperation with other themes.
Student Learning Goals: Students learn and demonstrate modern best practices for reproducible model formulation and software development.
Links to other projects: If completed successfully, outcomes of projects working with both subsets of them NWM and newly developed formulations should work with the approach implemented here. A final demonstration could compare new formulations to existing NWM using this same environment and evaluation architecture.
Training opportunities: This project will be part of the Scaling theme: students who participate in this theme will have 3 days of intensive training at the SI, including creating NWM subsets for CZO watersheds, running the NWM using jupyter notebooks/docker containers, using hydrologic data to evaluate hydrologic processes, using and analyzing CZO data.
Supplementary Materials:
Explore research toward community model development. https://doi.org/10.1002/2015WR017910 https://doi.org/10.1002/2016WR019285
Install and work through basic tutorials with Docker. https://docs.docker.com/get-started/ https://docker-curriculum.com/

DATA - What (or what types of) input data will be required?
Model Data: CUAHSI tool for NWM cutouts and forcings.
Observed Data: Readily available NWIS data.

可复现模型开发与评估环境。

问题陈述：针对现有的NWM子集及其新/创新模型公式研究，何种模型开发与评估方法为适宜？

目标：与NWM子集及其模型公式研究协作，设计并展示一个适用于可复现模型开发与评估的基础开发和测试环境。

方法：回顾可复现模型开发与评估的相关文献及社区实践。从现有的NWM容器化方案出发，确定适当的抽象层以封装各类水文模型及其公式。与相关主题合作实施示例。

学生学习目标：学生将学习和展示现代可复现模型公式及软件开发的最佳实践。

与其他项目的联系：若项目成功完成，与NWM子集及其新开发公式协作的项目成果应与此处实施的方法兼容。最终演示可以比较新公式与现有NWM在该相同环境和评估架构下的表现。

培训机会：本项目将纳入扩展主题：参与该主题的学生将在SI进行为期3天的密集培训，包括为CZO流域创建NWM子集、使用jupyter笔记本/容器运行NWM、利用水文数据评估水文过程、使用和分析CZO数据。

补充材料：

探索社区模型开发的科研进展。

安装并完成基本教程以使用Docker。

数据 - 需要何种（或何种类型）的输入数据？

模型数据：CUAHSI工具用于NWM裁剪和强迫数据。

观测数据：易于获取的NWIS数据。