How predictable is the anomaly pattern of summer extreme high-temperature days over Central Asia?

Extreme high-temperature events pose huge threats on human health and ecological environment. Central Asia (CA), located in an arid region, experiences frequent occurrences of extreme high-temperature events with regional discrepancy. However, it is unknown to what extent the distribution of summer extreme high-temperature events over CA can be predicted. This study aims to investigate the dynamic origins of summer distribution of extreme high-temperature days (EHDs) over CA and estimate the predictability using Predictable Mode Analysis (PMA). Based on daily maximum temperature data from 1980 to 2010, two major EOF (Empirical Orthogonal Function) modes of EHDs over CA are identified. The first mode exhibits a homogeneous positive pattern, which is associated with a barotropic anticyclonic anomaly that covers the entire CA. The second mode features a meridional dipole pattern, corresponding to a north to south see-saw geopotential height anomaly pattern in CA. Based on the understanding of the simultaneous physical factors and tracing lower boundary anomalous forcing in the previous season, two physical predictors are selected for each principal component (PC), and a set of Physics-based Empirical (P-E) models is established. The temporal correlation coefficient (TCC) skill between observed and predicted PC1 (PC2) is 0.60 (0.74) during the independent forecast period (2010–2021). According to the criteria of PMA, the first two modes can be considered as predictable modes. If predictable modes can be perfectly predicted, 64.8% of the total observed variability of EHDs over CA is potentially predictable. Using the predicted values of the first two PCs and the corresponding observed EOF patterns, the predicted distribution of EHDs can be reconstructed. During the independent forecast period, the areal averaged TCC skill can reach 0.44, providing a reference for actual predictability.