Reconciling disparate trends in streamflow and precipitation networks in the Pacific Northwest United States

Precipitation gages in the historical climatology network (HCN2) show weaker declines over the last 60 years than observed in the stream gage network (HCDN). While such a difference could potentially be explained through increased evapotranspiration, there are several clues suggesting that is an unlikely cause. First, there would be an expectation based on Budyko theory that declines would be stronger in wet years compared to dry years, which is exactly the opposite of what is observed. Second, the magnitude of streamflow declines represents much greater increases in latent heat fluxes than could be balanced by increases in net radiation. Finally, the elevation distribution of precipitation gages is much lower than the elevation distribution of the watersheds, with elevations of the precipitation gages being consistent with the elevations of the mouths of the watersheds, in some cases on the order of 1000 m lower than the mean elevation of the nearest watersheds. The discrepancy in trends between the precipitation gage network and the streamflow network could most easily be explained by a trend in precipitation at higher elevations. The existing precipitation network can provide some support for that hypothesis, but a dearth of high elevation precipitation data makes direct analysis difficult. While numerous other trends in fires, tree mortality, snowpack, streamflow timing, and glacier extent in the Pacific Northwest are consistent with, and in some cases much more simply explained by, a trend in precipitation, they can also be explained by temperature trends. The putative stationarity in Pacific Northwest precipitation has been used to assign all of these changes to temperature trends, and a more realistic estimate of precipitation changes could improve estimates of the sensitivity of these ecosystem and water resources to shifting climate characteristics.