Cookbook

UW Hydro Forecast Cookbook

Many of the steps that are necessary to create a forecast are automated and if things go as they should, then little manual intervention is needed to keep the forecasts running. However, the reality is that things do not always go as they should: Observations do not arrive on time, drives fails, the power goes out, someone messes up a crontab. The possibilities for failure are endless and you often need to provide some manual intervention to help the system catch up and recover. Why not provide fool-proof fail-over systems you may ask? Because this is a research system, not an operational system and the amount of code and effort needed to make the system fool-proof is difficult to justify. Rather than spending time to figure out how to successfully recover when a disk is failing, our time is better spent developing better forecasts, even though this means some manual intervention from time to time. Another situation in which you may need to intervene is when you need to set up a new forecasting system. In that case, there are some one-time actions that need to be performed.

To make this all a little more straightforward, this page collects recipes for how to do things and points you to the scripts you need to run and the checks you need to make.

Recipes

• Creating a retrospective run, with or without routing.