The use of hydrological model output as targets in tree ring-based streamflow reconstructions

Tree-ring records have been used extensively to reconstruct past streamflow variability.  Annually resolved estimates for several centuries prior to observations, and in some cases millennia, have been produced from dendroclimatic proxies.  However, despite an often strong hydroclimatic signal embedded in the rings, some factors limit the skill of such reconstructions, including human interference with the hydrological cycle.  We examine the relationship between local output from a Precipitation-Runoff Modeling System and tree-ring chronologies from six U.S. regions of varying climate and ecology.  Strong positive correlations between tree growth and capillary zone moisture variability are recorded.  When compared to correlations with precipitation, the tree ring-capillary zone relationship is overall stronger with a more coherent seasonal focus within the respective regions.  The capillary zone moisture also appears to capture spatial variability at a finer resolution than gridded soil moisture products.  This difference could isolate uncertainties during model calibration but also help identify localities from which new proxies containing independent and unexplained variance may be collected.  Although the results are not equal across regions, we suggest that there are several potential advantages to be gained from using hydrological model parameters as the target in dendroclimatic reconstruction exercises.