Spatial And Temporal Variability Of Precipitation Over The Middle And Eastern Parts Of Northwest China During The Past 400 Years And Its Potential Driving Mechanisms

A large-scale drought event is usually composed by co-occurring droughts in several climate regions. Studying the temporal relationships of precipitation between different climate regions and its potential driving mechanisms is very important to understand the reasons of large-scale drought occurrences. In this study, two tree-ring widths based precipitation reconstructions, spanning about 1000 years(A.D. 1000~2000) and 400 years(A.D. 1615~2006), were developed for the middle and eastern parts of Northwest China, respectively. In order to understand the temporal relationships of the two reconstructions, a 41-year running correlation(RC41) was calculated over the past 400 years. Because they are located at the transition region of the mid-latitude westerlies(MLW) and the Asian summer monsoon(ASM), the following analyses will focus on the potential effects of the two climate systems on the precipitation relationships of the two study areas. Before conducting the reconstructions, we analyzed the relationships of 137 tree-ring width chronologies with precipitation and temperature first, which can help us understand whether tree growth of the middle and eastern parts of Northwest China responds to a similar climate.The major conclusions are as follows:(1) The tree-ring width chronologies from the middle to lower forest zones in not only the middle part but also the eastern part of Northwest China were mainly correlated positively with precipitation but negatively with temperature of the late growing season of the prior year and the early growing season of the current year, suggesting that tree growth at the low elevations over the two study areas is mainly limited by the similarly seasonal moisture. Further analyses showed that most of the chronologies of the two study areas had the highest correlations with seasonal precipitation in P7C6(from July of the prior year to June of the current year), P8C6, or P8C7. Moreover, the variations of the regional average chronologies of the two study areas were often similar to each other over the past centuries. This suggests that the tree-ring widths based precipitation or drought reconstructions of the middle and eastern parts of Northwest China may be comparable. Unlike to the middle to lower forest zones, the chronologies from the upper timberlines in the middle part of Northwest China correlated positively with both of precipitation and temperature, and the latter is much more significant than the former, suggesting that temperature is the main limiting factor to tree growth at the high elevations.(2) Because the sampling sites are highly clustered in the middle part of Northwest China,we averaged 17 selected site chronologies from the middle to lower forest zones into 6 regional chronologies according to the administrative divisions. The first principal component of these 6 regional chronologies(MPC1) was extracted to better represent the characteristics of the regional tree-ring width variability. Climate-growth correlation analyses showed that MPC1 mainly recorded P7C6 precipitation with a correlation of 0.74(p<0.01) during their common period of 1959~2000. Therefore, we used MPC1 to reconstruct local P7C6 precipitation variability over the past 1000 years. Spatial correlation analyses showed that the reconstruction were capable of representing P7C6 precipitation variability for a large-scale region over the middle part of Northwest China.(3) For the eastern part of Northwest China, we selected four site chronologies with more than 300 years, and calculated their first principal component(EPC1; 1708~2006) and the average of the two longest chronologies(EAVE; 1615~2006) to better represent the characteristics of the regional tree-ring width variability. Climate-growth correlation analyses showed that both of EPC1 and EAVE mainly recorded P8C7 precipitation, with correlations of 0.78(n=49, p<0.01) and 0.75(n=55, p<0.01) during their own common periods, respectively. Finally, we used EPC1 and EAVE to reconstruct local P8C7 precipitation variability over the past 299 years and 392 years, respectively. The two reconstructions were much high consistent with each other during the common period of 1708~2006, with a correlation of 0.88(n=299, P < 0.01). Further analyses showed both of the two reconstruction were capable of representing P8C7 precipitation variability for a large-scale region over the eastern part of Northwest China. Because the EAVE based precipitation reconstruction is longer than the EPC1 based precipitation reconstruction, the former was used in the following analyses.(4) In order to understand the temporal relationships of precipitation variability of the two study areas, RC41 between the MCP1 and EAVE based precipitation reconstructions was calculated over the common period of 1615~2000. RC41 has a large amplitude, with correlation coefficient from-0.05 to 0.63, and 41.0% of it is at the 95% significant level. During the common period of 1615~2000, the higher co-variability periods between the two reconstructions occurred around the 1615~1681, 1706~1757, 1792~1861, 1877~1933 and 1976~2000, and the lower co-variability periods were found around the 1677~1711, 1756~1781, 1843~1873 and 1934~1963. Droughts were much easier to occur simultaneously over the two study areas in the higher co-variability periods. We found that the higher co-variability periods often corresponded to the weak ASM and/or strong MLW, and the lower co-variability periods often correspond to the strong ASM and/or weak MLW. This may be because that when the ASM is weak both the NTP and NCC regions are controlled by the MLW since the ASM has retreated to its southern location, while the two regions are controlled by MLW and ASM, respectively, when the ASM is strong.