Data sets

The main product of this project consists of streamflow time series for 396 locations. These come in both raw and bias-corrected forms for 190 of these locations.

Raw streamflow time series

Every one of the 396 locations throughout the domain includes 172 ensemble members each representing a distinct realization of both the 20th century and the future 21st century. These time series are each unique both in their representation of the control period (the 20th century) and the future period.

The time series have not undergone any post-processing and so reflect exactly what the given hydrologic model set-up produces. This is a key point: the raw streamflow time series are directly synced with the hydrologic states of the modeling set-ups. Also, all of the streamflow time series within a stream network are consistent (a.k.a. streamflows increase monotonically from any given location to a downstream location).

However, accordingly, using streamflow time series directly from the hydrologic model results in systemic biases in the modeling set-up are carried through to each of the time series. For example, if streamflow at a given location is systematically too high given a poor calibration or errors in meteorological forcing information, that systematic bias will likely occur across all simulations. In general, these systematic biases are more pronounced for smaller basins.


For a subset of locations (190), we have produced bias-corrected flows. For these sites, RMJOC supplied no-regulation, no-irrigation time series which provide a record of what streamflow would have been without human impacts on the system. These time series were used to adjust the modeled streamflow time series to account for systematic model biases.

It is important to note that this additional modeling step breaks the water balance of the entire model set-up: streamflow is added or subtracted systematically. Further, unlike in the raw simulations described above, there may be discrepancies between upstream and downstream locations along a stream network.