Asymmetry of probabilistic prediction skills of the midsummer surface air temperature over the middle and lower reach of the Yangtze River valley

Using the prediction data from the Global Seasonal Forecast version 5 (GloSea5) during the 1993-2016 period, the probabilistic prediction skills of the midsummer (July and August) surface air temperature (SAT) over the middle and lower reach of the Yangtze River valley (MLYR) are evaluated by the ranked probabilistic skill score. We found that the GloSea5 better predicts below-normal (BN) events than above-normal (AN) events at a long lead time, as well as the associated western Pacific subtropical high (WPSH) and circumglobal teleconnection (CGT) anomaly. This result reveals the asymmetry of the probabilistic prediction skills of the midsummer MLYR SAT, which is related to the asymmetric effect of predictability sources: the warm midsummer ENSO favors BN events via promoting the eastward retreat of the WPSH, whereas the cold midsummer ENSO exhibits weak effect on AN events due to the induced southwestward expansion of the WPSH; the warm mid-latitude North Atlantic (MNA) promotes AN events via projecting its influence onto the atmospheric teleconnections such as the CGT, but not vice versa. The GloSea5 exhibits strong response to both two predictability sources, but fails to reproduce the asymmetric effect of predictability sources especially for ENSO, which limits the prediction skills of AN events. On the other hand, the worse (better) simulations of the warm MNA (warm midsummer ENSO) lead to the lower (higher) prediction skills of AN (BN) events. These results are useful for better understanding the predictability of the midsummer SAT over the MLYR.