Assessing Land Memory in the GSWP2 Simulations and Association to Global Recycling Estimates

New model simulations and reanalyses allow investigation of the relationship between local precipitation recycling and the persistence of land surface moisture states. The Global Soil Wetness Project Phase 2 (GSWP2) baseline (B0) and precipitation sensitivity (P1, P2, and P3) simulations have been performed in “standalone mode” (i.e. prescribed atmospheric conditions) to provide global estimates of land states and fluxes at a 1°x1° resolution, spanning the years 1986-1995.  In this analysis, we present an evaluation of the land models' simulated memory of water storages, namely root-active soil moisture (RSM) and snow water equivalent (SWE). Our metrics of land memory are based on characterizing the features of auto-correlation decay. A timescale of “memory” is defined as the lag at which auto-correlation drops below a critical level. For this analysis, we consider these timescales resulting from three separate criterion values of: e-folding (i.e. critical auto-correlation value equals e^-1), and the auto-correlation value at the 1% and 5% significance levels. An additional metric is derived that characterizes the shape of the auto-correlation decay as either: exponential-like, linear, or Gaussian-like. For the SWE results, the GSWP2 simulations are compared against the recently released satellite-derived SWE estimates provided by the National Snow and Ice Data Center (NSIDC), which spans the entire GSWP2 simulation period.  The geographic and temporal features of SWE and RSM memory metrics are assessed separately as well as sequentially to determine the degree to which the models translate SWE memory into RSM memory through the spring ablation period. The RSM memory results are then also compared to evaporative sources and precipitation recycling estimates based on a back-trajectory diagnosis using Reanalysis-2 data. Overall, very distinct and co-varying features in the seasonally averaged fields are found, but special attention is drawn to regions and periods where strong RSM memory (in the GSWP2 results) and estimated precipitation recycling coincide.  Where possible, these strong coincidences are considered in light of recent model-based estimates of land-atmosphere coupling strengths.