Based on 2912 annual resolution proxy sequences mainly derived from tree rings and historical literature, we propose a set of standard precipitation index (SPI) reconstructions covering the entire Asia every year (November to October) since 1700, as well as standard precipitation index reconstructions for the rainy season (i.e. November to April in West Asia and May to October in other regions), with a spatial resolution of 2.5 ∘. In order to screen the best candidate proxy indicators for reconstructing SPI within each grid from available proxy indicators with homogeneous precipitation mechanisms and similar precipitation variability in their connected areas, we developed a new method that uses grid position dependency partitioning obtained from instrument SPI data. The verification results indicate that these reconstructions are effective for most regions in Asia. Compared with the measured precipitation before calibration time, the evaluation of data quality shows that, except for a few networks in western Russia, coastal areas of Southeast Asia, and northern Japan, our reconstruction has high quality and can display precipitation variability in most of the study areas.
The dataset includes four types of SPI reconstructions: (1) SPI reconstructions for the entire Asia from November to October, without using tree ring density chronology and width chronology that are negatively correlated with precipitation (November October SPIA version); (2) The SPI reconstruction of the entire Asia from November to October has added tree ring density chronology and width chronology negatively correlated with precipitation (SPI B version from November to October); (3) The reconstruction of rainy season SPI in non tropical regions of Asia (November April SPI in West Asia and May October SPI in other regions) did not use tree ring density chronology and width chronology negatively correlated with precipitation (rainy season SPIA version); (4) Reconstructing the rainy season SPI in extra tropical regions of Asia (November April SPI in West Asia and May October SPI in other regions), adding tree ring density chronology and width chronology negatively correlated with precipitation (rainy season SPI B version). Each version is stored in a NetCDF file (. nc) containing five three-dimensional (longitude x latitude x time) variables, including reconstructed SPI, adjusted decision coefficients (R2a), validation RE, validation CE, and the number of surrogate indicators used for construction (nPrx).
In the study, the grid size used for SPI reconstruction was set to 2.5 °× 2.5 °. The instrument data used for calibration is adjusted based on the 0.5 °× 0.5 ° grid monthly SPI data calculated from the National Oceanic and Atmospheric Administration's land precipitation products from 1948 to 2019, which can be downloaded from the IRI/LDEO Climate Data Library（ http://iridl.ldeo.columbia.edu/SOURCES/.IRI/.Analyses/.SPI/
There are a total of 2912 annual resolution proxy sequences available for reconstruction in Asia and adjacent land areas (Eastern Europe and Alaska), including 2792 from tree rings, 115 from historical literature, 4 from ice cores, and 1 from stalagmites. It is worth noting that all substitute series have at least 20 records overlapping with instrument time periods since 1948 to ensure sufficient sample size for calibration and validation, and over 30 records prior to 1948 for reconstruction. The tree ring data (2772) mainly comes from the International Tree Ring Database (ITRDB), which is compiled by the World Paleoclimatology Data Center (WDC-P), https://www.ncei.noaa.gov/products/paleoclimatology
Due to the diverse climate types of precipitation heterogeneity in the study area and the sensitivity of spatial representativeness of individual proxy indicators to location, we have developed a new method to determine the proxy indicator search area (hereinafter referred to as the "search area") to reconstruct the SPI in each grid. This method is developed based on the regional division of interannual precipitation variation consistency in the spatial model of the precipitation system, to ensure that the proxy indicators in the search area can effectively indicate the SPI changes in the target grid. We divide the regions based on the spatial pattern of the correlation coefficient (CC) between the SPI of each target grid used for SPI reconstruction and other grids within the study area, which is calculated from instrument SPI data. The definition of the search area is all connected grids around the target grid, whose CC exceeds the significance level of 0.05.
Compared with previous studies on the three reconstructions of summer precipitation (or PDSI) in the Asian monsoon region (JJA or May September), our May October SPI reconstruction showed a 10% increase in R2a during the calibration period from the southern Qinghai Tibet Plateau to the eastern Indian subcontinent, western Southeast Asia, and northwestern China, compared to the best of the three reconstructions. In addition, our reconstruction has slightly higher R2a than other reconstructions in parts of Mongolia, Central Asia, and eastern China. Especially in eastern China, the R2a we reconstructed is about 40% higher than the R2a reconstructed using only tree ring data. These improvements are not only due to the addition of more surrogate data (including DWI obtained from Chinese historical literature and recently published tree ring data), but also due to the development of reconstruction methods, which select surrogate data from the connection search area significantly positively correlated with the target grid SPI through the GLDD method.
In addition, the correlation graph between our rainy season SPI reconstruction results and the four reconstruction results previously studied in the Asian monsoon region shows that most grids passed the significance level of 0.01.
This work is licensed under
CC BY 4.0 (Creative Commons Attribution 4.0 International License).
https://cstr.cn/31253.11.sciencedb.01829
10.57760/sciencedb.01829
