Demo: Use the Global Watersheds API and Python to automatically delineate watersheds

It seems that a few people are discovering the mghydro watershed delineation API. Last Monday, the API handled requests for 9,679 watersheds over the course of a few hours. (Someone is very interested in the watersheds of France!)

One way to use the API is with GIS software, to fetch watershed boundaries and river centerlines from the web app and display them in your desktop GIS. Another use is to automate your requests, rather than using the online map interface. This could be useful if you want to delineate hundreds (or thousands!) of watersheds. However, remember that the results of automated delineation are very often wrong, so you should visually check each one.

Here is a demo in Python of how you could use the API to download watersheds for multiple outlet points. Click on the filename at the bottom to go to Github, where you can download the Jupyter notebook.

How to use Python to delineate watersheds in the continental US with USGS web services

I recently tried my hand at using NLDI, the new water data web service from the U.S. Geological Survey. The service provides access to water data from the National Hydrography Dataset (NHD), version 2. I was specifically interested in using the NHD geodata to delineate watersheds. This data was digitized from 1:100,000-scale maps. (There is newer, more detailed data, called NHDPlus HR, for high-resolution, but no web service for these data yet.)

Snake River watershed upstream of the Lower Granite Dam in eastern Washington state (just a location I picked at random for this demo!)

It took me a couple days of tinkering to get the results I wanted, but the results look impressive and seem accurate. You can run the code yourself by downloading one of these files from GitHub Gists. (Or follow along below.)