UCLA

Recent research has reported the great influence of springtime land surface temperature (LST) and subsurface temperature (SUBT) over the Tibetan Plateau (TP) on downstream region summer precipitation, indicating the potential application of LST/SUBT on subseasonal to seasonal (S2S) prediction. In this study, we employed both observational data and offline model simulation to explore the memory of surface and subsurface variables and assess the driving effects of snow/albedo and SUBT on the LST anomaly. Our composite analysis based on observation shows that the anomalous LST in the TP can sustain for seasons and is accompanied by persistent SUBT as well as snow and associated surface albedo anomalies. A multilayer frozen soil model reproduces the observed LST anomaly and shows surface albedo and middle-layer SUBT have 1–3 months memory, indicating the degree of persistence or dissipation of anomaly through time with more extended memory during spring. With simulated middle-layer SUBT as a predictor, the linear regression model produces (Formula presented.) of 0.44 and 0.26 for 1- and 2-month LST prediction, respectively. The predictability is higher during the spring. Our results also show February snowfall, May snowmelt, and aero in snow exert substantial impacts on springtime LST through snow albedo feedback. The long memory of SUBT allows it to preserve the surface thermal anomaly and release it gradually in the following months to seasons. Meanwhile, the sensitivity study indicates that the soil properties and soil column depth predominate SUBT memory, which suggests the key processes to improve LST/SUBT then downstream S2S prediction.