Predictability of the anomaly pattern of summer extreme high-temperature days over southern China

Southern China (SC) has been increasingly affected by extreme high-temperature events during summer (July and August, JA) in recent decades. However, to what extent the anomaly pattern of the extreme high temperature days (EHDs) over SC is predictable remains largely unexplored. To address this issue, we build physics-based empirical model to predict the spatial distribution of JA EHDs over SC and examine its predictability. Two major modes of EHDs over SC are identified by empirical orthogonal function analysis based on the observational data during period of 1980–2009. The first mode is a homogeneous EHDs mode related to a prevailing low-level high-pressure anomaly over SC. The second mode of EHDs is a meridional dipole pattern associated with the Pacific-Japan pattern over East Asia/Western Pacific sector and Pacific Meridional Mode-like SSTA over North Pacific. Via tracking the boundary anomalies forcing of EHDs over SC, physically meaningful predictors are selected to establish a set of physics-based empirical (P-E) models for predicting the first two leading principal components (PCs). The two modes can be predicted reasonably well by the P-E models. Thus, they can be regarded as the predictable modes. Assuming that these two modes can be predicted perfectly, 66.1% of the total variance for the distribution of summer EHDs over SC are potentially predictable. Using the first two predicted PCs and corresponding observed spatial pattern of EOF modes, the prediction of anomalies pattern of EHDs over SC can be obtained. The independent forecast skill (2010–2019) of the domain-averaged temporal correlation coefficient is 0.37, which is close to that of 0.44 for maximum potential attainable prediction skill.