Workshop: Application of Google Earth Engine for Hydrologically-Based Machine Learning Applications

This tutorial creates a simple machine learning model to predict streamflow using Google Earth Engine (GEE) accesssed National Land Data Assimilation System (NLDAS) temperature and precipitation data.

Objectives

Participants will learn how to:

• Virtual Environment. Create a virtual Python environment for accessing data on the GEE platform.
• Remote Data Access. Use a Jupyter environment to retrieve remote sensing data stored on the GEE platform, catchment information using StreamStats, and USGS streamguage information through the National Water Information System (NWIS).
• Data Processing. QA/QC real data, resample data to an apropriate temporal resolution, connect data to catchment attributes
• Machine Learning. Build and train a machine learning model using GEE provided remote sensing data to predict streamflow.
• Model Evaluation. Explore and understand different methods and metrics for evaluating the performance of a ML hydrological model.

Please click on the GEE_ML_Workflow folder to begin.