Little Ice-Age cooling in the Hengduan Mountains, China: A 600-year warm-season temperature reconstruction from tree-rings

Description

The distributions of forest, ice and snow in the Hengduan Mountains of China have undergone significant changes due to ongoing climatic warming. To better understand the spatiotemporal pattern of temperature changes in the Hengduan Mountains we used tree-ring cores from multiple individuals of Larix speciosa Cheng et Law at five sites to develop a regional chronology and to establish the relationship between tree-ring radial growth and warm-season (May–September) mean temperature. The regional chronology accounts for 46.1% of the observed variance in the warm season and was used to reconstruct regional temperature levels back to 1420. Four cool intervals (1490 – 1570, 1590 – 1660, 1700 – 1790, and 1800 – 1880) indicate that the Hengduan Mountains experienced the Little Ice Age, and the changes were synchronous with cooling in the Tibetan Plateau and the Northern Hemisphere, demonstrating a well-defined Little Ice Age signal in the South Asian monsoon region.

Steps to reproduce

In the laboratory, the cores were air-dried, mounted, and the surfaces sanded using 400 & 600 grit sandpaper. High-resolution image scanning was applied to measure the tree-ring widths. The Epson Expression 12000XL was used to scan all the sample cores, and the CooRecorder 9.4 tree-ring width measurement software (www.cybis.se) was used to measure the tree-ring widths in the scanned images (the measurement accuracy was within 0.001 mm); the COFECHA program was used to check the data quality.Friedman super smoother was applied to remove the age dynamics for each series using the ARSTAN program. Pearson correlation analysis showed that the five tree-ring width chronologies were significantly positively correlated with each other. Therefore, we combined all the tree-ring width series to develop a regional chronology (RC) . The original width measurements for the five sampling sites were combined and detrended using RCSigfree software, which is often used to develop regional tree-ring chronologies. First, we used data-adaptive power transformation based on the local mean and standard deviation to reduce the potential heteroscedasticity of the original ring width measurements. Second, we used spline curve detrending with the Friedman variable span smoother to detrend all series, and a data-adaptive smoothing technique was designed to retain the low-frequency variance (alpha value set to 7). Third, we calculated the bi-weight robust mean to average all the series into a chronology and then used the techniques outlined by Osborn et al., (1997) to stabilize the variance of the chronology to reduce the impact of differences in sample size. Using a 51-year moving window with a step size of 1 year and an overlap of 50 years, the expressed population signal (EPS) and mean inter-series correlation (Rbar) were calculated to confirm the reconstructable time frame. The regional chronology (RC) used in the mean temperature reconstructions was clipped before 1420 CE according to the reconstruction standard: EPS ≥ 0.85 and the minimum sample depth ≥ 6 cores (three trees) .

Institutions

Categories

Funding

Licence