Multi-century (635-year) spring season precipitation reconstruction from northern Pakistan revealed increasing extremes

Abstract

The Hindu Kush Himalaya region is experiencing rapid climate change with adverse impacts in multiple sectors. To put recent climatic changes into a long-term context, here we reconstructed the region's climate history using tree-ring width chronologies of climate-sensitive Cedrus deodara and Pinus gerardiana. Growth-climate analysis reveals that the species tree-growth is primarily limited by moisture stress during or preceding the growing season, as indicated by a positive relationship between the chronology and precipitation and scPDSI, and a negative one with temperature. We have reconstructed 635 years (1384-2018 CE) of February-June precipitation using a robust climate reconstruction model that explains about 53% variance of the measured precipitation data. Our reconstruction shows several dry and wet episodes over the reconstruction period along with an increase in extreme precipitation events during recent centuries or years. Long, very wet periods were observed during the following years: 1392-1393, 1430-1433, 1456-1461, 1523-1526, 1685-1690, 1715-1719, 1744-1748, 1763-1767, 1803-1806, 1843-1846, 1850-1855, 1874-1876, 1885-1887, 1907-1909, 1921-1925, 1939-1944, and 1990-1992, while long dry periods were observed during the following years: 1398-1399, 1464-1472, 1480-1484, 1645-1649, 1724-1727, 1782-1786, 1810-1814, 1831-1835, 1879-1881, 1912-1918, 1981-1986, 1998-2003, and 2016-2018 CE. We found predominantly short-term periodicity cycles of 2.0, 2.2, 2.3, 2.4, 2.6-2.7, 2.9, 3.3, 4.8, 8.1-8.3, and 9.4-9.6 years in our reconstruction. Spatial correlation analyses reveal that our reconstruction is an effective representation of the precipitation variability in the westerly climate-dominated areas of Pakistan and adjacent regions. In addition to the influence of regional circulation systems like western disturbances, we found possible teleconnections between the precipitation variability in northern Pakistan and broader-scale climate modes or phases like AMO and ENSO. The study also highlights the prospects of tree-ring application to explore linkages between western disturbance, increasing intensity and frequency of extreme climate events, and analysis of long-term atmospheric circulation over the western Himalayan region.

Figure 1

Location of the sampling sites, species sampled and climate station used in this study.

Figure 2

Climograph showing monthly climate at the study area (a) with temporal trend in the annual temperature (b) and precipitation.

Figure 3

Regional composite tree-ring width residual chronology (blue) from Northern Pakistan along with their 10-year smoothing spline (thick red line) and no of cores (blue dotted line) (a). The running EPS (red line) with commonly used threshold value of 0.85 (pink dotted horizontal line) and running R-bar (blue line) are presented in the lower panel (b).

Figure 4

Relationship (correlation coefficient) between the tree-ring composite residual chronology and monthly and seasonal climatic data. The horizontal dashed line represents the significance of correlation coefficients at 95% level.

Figure 5

Seasonal correlation (seascorr) between the composite chronology and Drosh station precipitation and temperature data. Precipitation is primary and temperature is secondary variable. The analysis is done taking one, three, five, and twelve-month periods. The shaded bar represents significant correlation.

Figure 6

Heat map showing moving correlation between the regional composite tree-ring width residual chronology and monthly and seasonal precipitation of Drosh station from1965 to 2016. The DJF, MAM/MAMJ/FMAMJ, and JJAS represent winter, spring, and summer seasons, respectively. The * symbol in the map indicates significant correlation (analysis was done using the 'treeclim' package in R).

Figure 7

Comparisons between the observed (black line) and reconstructed (red line) spring season (Feb–June) precipitation from Northern Pakistan during the calibration period of 1965–2016. Spring season (Feb–June) precipitation reconstruction from Northern Pakistan (green line) with 10-year low passes (red line). The horizontal dashed lines indicate the precipitation beyond the one standard deviation (σ) from the long-term mean (black horizontal line).

Figure 8

Spatial correlation between observed (a) and reconstructed (b) spring season precipitation with the CRU gridded precipitation and sea surface temperature (c, d) for the common period to observed data (1965–2018 CE). The map was prepared online using KNMI climate explorer (https://climexp.knmi.nl/).

Figure 9

Power spectral (a) and Morlet wavelet (b) analysis of the reconstructed spring precipitation data from western Nepal Himalaya.

Figure 10

Comparison between February-June reconstructed precipitation (standardized value) from northern Pakistan in our study and other independent reconstructions from the Hindu Kush-Karakorum Himalaya (HKKH) and adjacent region. The compared other reconstructions include March–August PDSI from northern Pakistan, March–June precipitation from Kumaun region in western India Himalaya, March–July precipitation from Himachal region in western India Himalaya and March-June precipitation from western Nepal Himalaya.