Intercomparisons of multiproxy-based gridded precipitation datasets in Monsoon Asia: cross-validation and spatial patterns with different phase combinations of multidecadal oscillations

Abstract

Proxy-based reconstructions are essential for investigating precipitation change in Asia on multidecadal to centennial timescales. Thus far, a total of four gridded precipitation/drought datasets for the past half millennium have been developed, including reconstructed Asian summer precipitation (RAP), May–September precipitation in China (IGGPRE), warm season precipitation for Asia (WSP), and Monsoon Asia Drought Atlas (MADA). This study uses some independent evidence to evaluate and compare the four reconstructions and then examine the spatial patterns of precipitation under different phase combinations of the Atlantic multidecadal oscillation (AMO) and Pacific decadal oscillation (PDO). The results suggest that RAP and IGGPRE capture most variance of observations in eastern China, WSP performs better in central Asia and India, while MADA shows skill over Mongolia and the eastern Tibet Plateau. When proxies are evenly distributed in space, increasing the sample size contributes more to the explained variance than increasing the signal-to-noise ratio. In terms of the responses to the multidecadal oscillations, the spatial correlation coefficients are all negative between pairs of patterns, with only one of the oscillations residing in opposite phase. This means that all four reconstructed datasets capture the reverse spatial pattern when the phase of multidecadal oscillation changes. The WSP is the only dataset that shows positively correlated spatial patterns when the AMO and the PDO reverse together. The RAP and IGGPRE are more sensitive to the AMO than to the PDO, because the negative correlations caused by the AMO exhibit higher confidence levels. The spatial pattern in history is similar to that in the twentieth century when the PDO and AMO are both in positive phase, which indicates that the mechanism in this situation is stable over the long-term period.