Contents

Known issues

This is a large data set, summarizing about 25 thousand model years (across all RCP, GCM, downscaling method and hydrologic model permutations) for nearly 400 locations. While we have done our best to ensure consistency and quality across all permutations and locations, errors and/or problems do pop up on occasion.

We keep a record of known issues and will report on these here. We will do our best to fix outstanding problems as soon as possible. Note that poor performance of the raw model simulations at some locations is unavoidable and something we cannot fix. However, spurious spikes in model out or in the bias-corrected files, negative numbers in the streamflow files other than blocks of -9999 in the dynamically-downscaled simulations, formatting errors, etc. are things we would like to hear about.

Please consult these pages before filing a report by sending an e-mail to orianac@uw.edu. Use the subject line Problem report for the CRCC streamflow data set. Please be as specific as possible when filing a report, at minimum clearly stating:

  1. the version of the data set in which you encountered the issue
  2. the specific files including the issue
  3. the specific issue

We will not be able to respond to generic queries and will not be able to support requests for help with data ingestion and formatting. Please use the suggested e-mail subject to ensure that the e-mail is not overlooked.

date filed dataset version with found issue date resolved dataset version with issue resolved issue notes action
2017-09-20 v1.0 2018-01-11 v1.1 Spurious spikes and -9999 values in bias-corrected streamflows for some models at COT and DOR locations This appears a result of multi-day sequences of 0 streamflow in the simulated streamflow for these locations Added a very small, but inconsequential, offset to the simulated streamflows to remove 0s from the record.
2017-12-05 v1.0 2018-01-11 v1.1 Spurious spikes and -9999 values in bias-corrected streamflows for some models at HAH and AUB locations This appears a result of the NRNI record for these two sites not extending the length of the period used for training the bias-correction. For these two sites, the streamflow bias-correction used a training period constrained between WY 1963 and 2002.