资源环境科技发展态势分析平台

This study examines relationships between snowmelt and soil moisture (SM), in particular, the influence of snowmelt on soil moisture memory (SMM) and on near surface air temperature (T2M) over the extra-tropical northern hemisphere (ENH) using four state-of-the-art reanalysis products: ERA-Interim, ERA-Interim Land, MERRA-Land, and GLDAS, as well as using Canadian Seasonal and Interannual Prediction System (CanSIPS) seasonal hindcast data, over a 20 year period (1986-2005). We use correlation-based metrics along with a simple classification-based on when the top layer soil temperature () rises above freezing point during the annual freeze-thaw season, to evaluate the influence of snowmelt on SM. Our results show considerable differences across reanalyses as well as CanSIPS hindcasts regarding timing of maximum SWE () occurrences as well as the onset of thawing of the frozen soil. Correlation statistics indicate that strongly influences SM. As a measure of the persistence of this relationship, a decay time is defined by lag in days over which the correlation of SM with lagged decays to 1/e of its peak value. For a majority of grid cells over ENH this decay time is less than 45 days, which suggests does not strongly influence the SM beyond subseasonal time scales. The interannual autocorrelation of SM indicates strong persistence over subseasonal time scales, consistently across reanalyses as well as CanSIPS hindcasts. However, intra-seasonal autocorrelations of ERA-Interim and MERRA-Land SM over North America show anomalous sudden decline of SMM compared to the other products, likely due to the offline forcing of atmospheric variables which blocks the atmosphere's response to land feedbacks. One of the models used in CanSIPS, the Canadian Climate Model version 4 (CanCM4), also shows a sudden decline of intra-seasonal autocorrelation over Central Asia which is most likely due to weak land-atmosphere coupling over the region. Lag-lead correlation statistics between SM and T2M during the soil thaw period suggests that SM anomalies have measurable lagged influence on T2M with varying decay time over different regions and across different datasets.