Known as one of the largest semiarid to arid regions in the world, central Asia and its economy and ecosystem are highly sensitivity to the changes in precipitation. The observed precipitation and related hydrographic characteristics have exhibited robust decadal variations in the past decades, but the reason remains unknown. Using the pacemaker experiments of the Community Earth System Model (CESM1.2), we find that the tropical Pacific decadal variability (TPDV) and the Atlantic multidecadal variability (AMV) are the main drivers of the interdecadal variations in central Asian precipitation during 1955–2004. Both the decadal-scale warming of the tropical Pacific and North Atlantic are favorable for wetter conditions over central Asia. The positive TPDV is accompanied with high sea level pressure (SLP) over the Indo–western Pacific warm pool. Southwesterly winds along the northwestern flank of the high SLP can transport more moisture to southeastern central Asia. The warm AMV can excite a circumglobal teleconnection (CGT) pattern. A trough node of the CGT to the west of central Asia drives an anomalous ascending motion and increased precipitation over this region. The results based on the CESM model are further demonstrated by the pacemaker experiments of MRI-ESM2-0. Based on the observational TPDV and AMV indices, we reasonably reconstruct the historical precipitation over central Asia. Our results provide hints for the decadal prediction of precipitation over central Asia.
Keywords: Precipitation; Air-sea interaction; Interdecadal variability
Central Asia, which includes Kazakhstan, Uzbekistan, Turkmenistan, Kyrgyzstan, and Tajikistan, is a semiarid to arid region located in the midlatitude of Eurasia. Large variations in water cycle and the significant warming trend make central Asia one of the "hotspots" of climate change ([
While the physical processes that affect the interannual variations of central Asian precipitation have been well documented, including the role of the subtropical westerly jet (SWJ), El Niño–Southern Oscillation (ENSO), the North Atlantic Oscillation (NAO), the east Atlantic–western Russia pattern (EA/WR), and the polar–Eurasian pattern (PE) ([
Previous work has found that the interdecadal and multidecadal variability in the Pacific and Atlantic can effectively affect regional and even global climate, including southwest Asia, East Asia, and South Asia, the neighboring regions of central Asia (e.g., [
Precipitation records are more reliable after the 1950s over central Asia (Fig. 1). The different phase combinations of the interdecadal variability in the Pacific and Atlantic may obscure the SST-related signals in observation during the short period. The impacts of SST over single ocean basins on global or regional climate anomalies are investigated by using either AGCMs or ocean–atmosphere coupled general circulation models (CGCMs) (e.g., [
Graph: Fig. 1. (a) Total count of monthly station observations contributing to each grid cell over central Asia during 1891–2016 in the GPCC precipitation dataset. The average of monthly station observations contributing to each grid cell during (b) 1920–54 and (c) 1955–2004. The contours in (b) and (c) denote the geopotential height (GH = 0.5, 1.0, 2.5, 4.0 km).
Recently, pacemaker-like numerical experiments that restore the historical observational SST anomalies in specific basins based on CGCMs are an emerging useful method to understand the remote impact of SST anomalies over the Pacific or Atlantic (e.g., [
The remainder of the paper is organized as follows. We describe the observational data, model data, and methods in section 2. In section 3, we investigate the impact of TPDV and AMV on central Asian precipitation and related physical processes. Results derived from different models are also compared in section 3. Finally, a summary and concluding remarks are presented in section 4.
The monthly precipitation dataset constructed by the Global Precipitation Climatology Centre (GPCC; [
Graph: Fig. 2. (a) The time series of 9-yr running average of the area-averaged annual mean central Asian precipitation anomaly (black; unit: mm day−1), TPDV index (green; unit: 1), and AMV index (yellow; unit: 1) during 1955–2004. The solid (dashed) black line denotes the central Asian precipitation derived from GPCC (CRU). The solid (dashed) green and yellow lines donate the normalized TPDV and AMV indices derived from Kaplan SST (ERSST). The numbers denote the correlation coefficients between central Asian precipitation derived from GPCC and two indices derived from Kaplan SST and related p values. (b) The time series of 9-yr running average of the annual mean central Asian precipitation anomaly during 1955–2004 (unit: mm day−1) derived from GPCC (black), and the CAP index reconstructed by the observational normalized TPDV and AMV indices derived from Kaplan SST (red). The numbers denote the correlation coefficients between GPCC and restructured precipitation index and related p value.
In addition, the Hadley Centre Global Climate Extremes Index 3 (HadEX3) developed through the coordination of the joint World Meteorological Organization (WMO) Commission for Climatology/World Climate Research Programme/Joint Technical Commission for Oceanography and Marine Meteorology (CCI/WCRP/JCOMM) Expert Team on Climate Change Detection and Indices (ETCCDI) is also used. This dataset comprises gridded global land in situ-based dataset of precipitation with a spatial resolution of 1.25° × 1.25° for the period of 1920–2018. At least three stations with valid data are needed within the decorrelation length scale (DLS) for a value to be set for a grid box to ensure that regions with sparse station density will not influenced by inhomogeneities or other data issues ([
The tropical Pacific decadal variability (TPDV) index is defined as the leading empirical orthogonal function (EOF) of 9-yr low-pass SST anomalies in the equatorial Pacific (5°S–5°N) following [
Three sets of experiments were conducted by using the Community Earth System Model version 1.20 (CESM1.2) developed by NCAR ([
The first set of the experiments is historical simulations (hereinafter HIST), in which model was freely coupled and forced by historical radiative forcing based on phase 5 of the Coupled Model Intercomparison Project (CMIP5; [
Graph: Fig. 3. Climatological annual mean precipitation derived from (a) GPCC, and the historical simulations of (b) CESM1.2 and (c) MRI-ESM2-0 during the period of 1955–2004.
In addition, the outputs of the three-member MRI-ESM2-0 that participates the GMMIP are used to verify the results derived from CESM1.2. The MRI-ESM2-0 is developed by the Meteorological Research Institute (MRI) of the Japan Meteorological Agency. This model consists of four component models coupled by Scup, including the atmosphere–land component MRI-AGCM3.5 with a horizontal resolution of T
A 9-yr running average is applied to the observation and model outputs to remove interannual signals. One-dimensional linear regression analysis and multiple linear regression analysis are applied on the low-pass filtered time series of precipitation, AMV, and TPDV indices. A 1000-ensemble Monte Carlo nonparametric method (hereinafter MC test) is used to test the statistical significance at the 99%, 95%, and 90% confidence levels (p < 0.01, p < 0.05, and p < 0.1) of correlation coefficients and regression coefficients of the filtered time series.
To investigate the propagation of wave energy, the wave activity flux is diagnosed. Refer to [
(
where u = (u, υ) denotes the horizontal wind velocity, and ψ denotes the eddy streamfunctions. Overbars and primes denote the mean states and anomalies from the mean states, respectively. Subscript x and y denote zonal and meridional gradient, respectively.
A thorough analysis of the long-term and decadal variations of central Asian precipitation is hampered by the sparse observational precipitation records. Only few stations have long-term observational records more than 100 years, and those are mainly located on the foothills (Fig. 1). Before the 1950s, there were few monthly precipitation observations over central Asia, and the total count of monthly station observations contributing to each grid cell is less than 600 (Fig. 1a). The number of station observation in central Asia has decreased since the 1990s because of the breaking down of the Soviet Union in 1991 ([
Less precipitation is observed before 1980, following a wetter condition (Fig. 2a). The TPDV index is significantly correlated with the observed central Asian precipitation (r = 0.52) at the 10% level. However, [
To obtain the combined effects of TPDV and AMV on central Asian precipitation, we applied the multiple linear regression technique to central Asian precipitation derived from GPCC. The multiple linear regression model for central Asian precipitation is
(
where CAP denotes the reconstructed annual mean central Asian precipitation; TPDV_norm and AMV_norm are the normalized 9-yr running average of observational TPDV index and AMV index, respectively. A significant correlation is found between the CAP index and the observation (r = 0.57) at the 5% level during 1955–2004, which is higher than the correlation coefficients for the individual TPDV and AMV indices (Fig. 2b). The above results show that the interdecadal variations of observational central Asian precipitation are modulated by the combined effect of TPDV and AMV. The positive signs of the regression coefficients of these two indices indicate that during the positive phase of TPDV and the warm phase of AMV, central Asia receives above-normal precipitation. Thus, we use pacemaker experiments of CESM1.2, which restore observed SST anomalies over the tropical Pacific or North Atlantic, to identify and isolate the contributions of TPDV and AMV on central Asian precipitation in the subsequent analysis.
To separate the contributions of TPDV and AMV on central Asian precipitation, observed SST anomalies are restored in the tropical Pacific (20°S–20°N, 175°E–75°W) and North Atlantic (0°–70°N, 70° W–0°) in the historical simulation of the CESM1.2 model. These experiments are referred as HIST-TPAC and HIST-NATL, respectively (see section 2b). A significant correlation (0.68, p < 0.01) is found between the precipitation over central Asia derived from the ensemble mean of the HIST-TPAC and the observational TPDV index (Fig. 4a). The correlation coefficient is higher than that in the observation, partly because the AMV-related anomalies and other stochastic noises are greatly suppressed by the multimember ensemble average. For one standard deviation of the TPDV index, central Asian precipitation increases at a rate of 0.24 (0.18–0.30 for the 95% confidence range) standard deviations (Fig. 4b).
Graph: Fig. 4. (a) The time series of 9-yr running average of simulated area-averaged central Asian precipitation anomaly (black; unit: mm day−1) derived from the HIST-TPAC and the observational normalized TPDV index (blue; unit: 1) and (b) their relationship. (c) The time series of 9-yr running average of simulated area-averaged central Asian precipitation anomaly (black; unit: mm day−1) derived from the HIST-NATL and the observational normalized AMV index (blue; unit: 1) and (d) their relationship. The black lines in (a) and (c) denote the results of ensemble mean; the light shades are for plus and minus half of the standard deviation about the ensemble mean. The numbers denote the correlation coefficient between central Asian precipitation and the TPDV and AMV indices and related p values. The red fitting lines in (b) and (d) are obtained by the least squares method based on the normalized precipitation anomaly and normalized TPDV and AMV indices. Blue and gray dashed curves denote the 95% confidence range and prediction range of the linear regression, respectively. The numbers denote the regression coefficients and 95% confidence ranges.
To understand how SST anomalies over the tropical Pacific drive the interdecadal variations in central Asian precipitation, we regressed SST, SLP, precipitation, and wind anomalies derived from the HIST-TPAC onto area-averaged precipitation over central Asia (Figs. 5a,c). The increase of central Asian precipitation is accompanied with positive SST anomalies over the tropical eastern Pacific and negative SST anomalies over the subtropical western Pacific, which agrees with the positive phase of TPDV (Fig. 5c). Consistent with the distribution of SST anomalies, the SLP anomalies show a seesaw pattern with negative signals over the eastern Pacific and positive signals over the Indo–western Pacific warm pool (Fig. 5c). The southwesterly winds along the northwestern flank of the expansive high SLP over the Indo–western Pacific warm pool can transport more moisture to southeastern central Asia from the Arabian Sea and tropical Africa, leading to the increase of precipitation over this region (Fig. 5a).
Graph: Fig. 5. The annual mean precipitation (shading; unit: mm day−1) and vertical integrated moisture fluxes (vectors; unit: kg m s−1) regressed onto the area-averaged annual mean central Asian precipitation (unit: mm day−1) in (a) HIST-TPAC and (b) HIST-NATL. (c) The annual mean sea surface temperature (shading; unit: K), sea level pressure (contours; unit: Pa), and 850-hPa wind anomalies (vectors; unit: m s−1) regressed onto the area-averaged annual mean central Asian precipitation in the HIST-TPAC. (d) The annual mean sea surface temperature (shading; unit: K), 200-hPa geopotential height anomalies relative to the zonal mean (contours; unit: m), and 700-hPa wind anomalies (vectors; unit: m s−1) regressed onto the area-averaged annual mean central Asian precipitation in the HIST-NATL. The white hatched patterns indicate that the regression coefficients are significant at the 10% level using the MC test. The regression coefficients of moisture fluxes and wind anomalies shown are significant at the 10% level using the MC test. All variables are derived from the ensemble mean of HIST-TPAC or HIST-NATL of CESM1.2.
The high SLP anomalies over the Indo–western Pacific result from weakened zonal gradient of tropical SSTs in the positive TPDV phase, which is similar to that induced by El Niño on the interannual time scales ([
Graph: Fig. 6. The annual mean (a) precipitation (unit: mm day−1), (b) 200-hPa geopotential height anomalies relative to the zonal mean (unit: m) and 200-hPa wind (vectors; unit: m s−1), and (c) 850-hPa streamfunction (unit: 107 m2 s−1) relative to the zonal mean and 850-hPa wind (vectors) regressed onto the area-averaged annual mean central Asian precipitation in the HIST-TPAC. The white hatched patterns indicate the regression coefficients are significant at the 10% level using the MC test. All variables are derived from the ensemble mean of HIST-TPAC of CESM1.2.
We further examine the relationship between central Asian precipitation and the AMV using the HIST-AMV experiments. After removing the TPDV-related responses and statistic noises, a significant correlation (0.70, p < 0.01) between central Asian precipitation and the observational AMV index is obtained, indicating that the warm (cold) phase of AMV favors increased (decreased) precipitation over central Asia (Fig. 4c). One standard deviation of the AMV index corresponds to an increase of the central Asian precipitation at a rate of 0.24 (range of 0.18–0.29) standard deviations (Fig. 4d).
The AMV-related precipitation responses are different from those forced by the TPDV (Figs. 5a,b), implying that the SST anomalies over the North Atlantic influence central Asian precipitation in a different way. To investigate the related physical mechanisms, we regressed SST, precipitation, geopotential height, and wind anomalies derived from the HIST-NATL onto area-averaged precipitation over central Asia (Figs. 5b,d and 7). The positive precipitation anomalies over central Asia derived from the HIST-NATL accompany the warm phase of AMV, showing a basinwide warming in the North Atlantic (Fig. 5d). The positive SST anomalies over the North Atlantic can result in an eastward barotropic zonal wave train located along the Northern Hemisphere westerly jet with positive centers over western Europe and the North Pacific, and negative centers over the high-latitude North Atlantic and central Asia (Fig. 5d). This zonal wave train is evident in both upper troposphere and lower troposphere, and intensifies with height (Fig. 7).
Graph: Fig. 7. The annual mean (a) 200-, (b) 500-, and (c) 700-hPa geopotential height anomalies relative to the zonal mean (unit: m) regressed onto the area-averaged annual mean central Asian precipitation (unit: mm day−1); (a) also includes the corresponding wave-activity fluxes (vectors) in the HIST-NATL. The white hatched patterns indicate the regression coefficients are significant at the 10% level using the MC test. All variables are derived from the ensemble mean of HIST-NATL of CESM1.2.
The barotropic stationary wave pattern dynamically corresponds to an anomalous cyclone at low-level troposphere over northern central Asia (Fig. 5d). Moisture from the North Atlantic and northern Europe is transported into northern central Asia by strengthened westerlies along the southern flank of the cyclonic circulation, leading to an increase of precipitation over central Asia, especially the northern part (Fig. 5b). In addition, the wave train interacts with the mean mid- and upper-level tropospheric westerly jet flows, resulting in a strengthened ascending motion over northern central Asia. The southerlies along the eastern flank of the anomalous cyclonic circulation can result in warm advection, which favors an anomalous ascending motion and increased precipitation over northern central Asia (Fig. 5d). This teleconnection pattern is similar to the AMV-related interdecadal circumglobal teleconnection pattern over the range of North Atlantic to central Asia ([
Our results derived from CESM1.2 provide a comprehensive picture of how the TPDV and AMV impact central Asian precipitation. We further confirmed the results with MRI-ESM2-0 to avoid model dependence. In MRI-ESM2-0, significant correlation coefficients between TPDV/AMV indices and central Asian precipitation under related experiments (r = 0.84 and 0.66 for HIST-TPAC and HIST-NATL, respectively; p < 0.01 for both) strongly supports our results based on CESM1.2 (Fig. 8). Besides, the physical processes are also significant in MRI-ESM2-0 although there are some discrepancies (Fig. 9). For example, the precipitation signals are strong and significant over the whole of central Asia both for the TPDV and AMV (Figs. 9a,b). The precipitation-related negative SST anomalies are weaker and insignificant over the western Pacific but stronger over the North Atlantic in the HIST-TPAC (Fig. 9c). Besides, the AMV-induced wave train is different with CESM1.2 over East Asia (Fig. 9d). These discrepancies may due to fewer members of MRI-ESM2-0, which could make the TPDV (AMV)-related responses not be thoroughly removed from the ensemble mean of the HIST-NATL (HIST-TPAC) experiments. In addition, the uncertainties of the atmospheric responses to the AMV-related SST anomalies in different models may come from the atmospheric internal variability, which is related with the interactions between the transient eddy and large-scale flow ([
Graph: Fig. 8. As in Figs. 4a and 4c , but derived from MRI-ESM2-0.
Graph: Fig. 9. As in Fig. 5 , but derived from MRI-ESM2-0.
In this study, we investigate the influences of TPDV and AMV on the interdecadal variations of central Asian precipitation based on pacemaker experiments of the CESM1.2 and MRI-ESM2-0 models that restore the historical observational SST anomalies in the tropical Pacific and North Atlantic. We analyzed the physical processes related to the TPDV and AMV. The major results are summarized in Fig. 10 and given below.
DIAGRAM: Fig. 10. Schematic diagram of the effect of (a) TPDV and (b) AMV on the interdecadal variability in central Asian precipitation. The red and blue shading denotes the positive and negative SST anomalies, respectively; green shading denotes the positive precipitation anomalies over central Asia. (a) The red curve denotes the range of anomalous high sea level pressure, the blue arrow denotes the anomalous low-level anticyclonic circulation, and the black arrows denotes the wave responses. (b) The blue L and red H denote the anomalous upper-level low and high geopotential heights, respectively.
The TPDV plays a dominant role in driving the interdecadal variations in central Asian precipitation. The positive phase of TPDV can lead to an increase of precipitation over central Asia, especially the southeastern region. The negative SST anomalies in the western Pacific are associated with high SLP over the Indo–western Pacific warm pool; the southwesterly fluxes along the northwestern flank of the high SLP can bring more water vapor to southeastern central Asia and lead to the increase of precipitation (Fig. 10a). The positive SST anomalies in the tropical eastern Pacific can induce a weakened Walker circulation, which is associated with a decrease of precipitation over the western Pacific, an increase of precipitation over the central eastern Pacific, and related changes in latent heat release. Both the poleward-propagating equivalent barotropic Rossby wave trains emanating from the central equatorial Pacific and the westward-propagating baroclinic Rossby wave trains emanating from the western Pacific can influence precipitation over southeastern central Asia.
The AMV can also modulate the interdecadal variations in precipitation over central Asia. A warmer North Atlantic can lead to wetter conditions over central Asia, especially the northern part. The positive phase of AMV can induce an eastward circumglobal stationary baroclinic wave train located along the Northern Hemisphere westerly jet, with positive centers over western Europe and the North Pacific and negative centers over the North Atlantic and central Asia (Fig. 10b). The trough node to the west of central Asia favors the ascending motion over northern central Asia and an increase of precipitation over this region. Moisture from the North Atlantic and northern Europe is transported into northern central Asia by strengthened westerlies along the southern flank of the anomalous cyclonic circulation.
When SST anomalies over the tropical eastern Pacific and North Atlantic are in the same phase, the precipitation responses are enhanced compared with those forced by SST anomalies over the tropical Pacific or North Atlantic individually. The TPDV and AMV are in the same phase during 1962–77, but are out of phase during 1978–97, resulting in the low correlation between the observational central Asian precipitation and the TPDV and AMV indices individually since 1950s. Based on the observational TPDV and AMV indices, we successfully reconstruct the historical precipitation in central Asia, which is significantly correlated (r = 0.57, p < 0.05) with he observation. We also extend the study period using HadEX3, finding a significant correlation coefficient (r = 0.55, p < 0.01) between central Asian precipitation derived from HadEX3 and the CAP index calculated based on the observational TPDV and AMV indices during 1920–2018 (Fig. 11). The relationship between central Asian precipitation and both TPDV and AMV is robust at a longer time period.
Graph: Fig. 11. The time series of 9-yr running average of annual mean central Asian precipitation anomaly during 1920–2018 (unit: mm day−1) derived from HadEX3 (black) and reconstructed based on the observational normalized TPDV and AMV indices derived from Kaplan SST (red). The numbers denote the correlation coefficient between HadEX3 and the reconstructed precipitation index and related p value.
Our study shows how the TPDV and AMV influence central Asian precipitation (Fig. 10). The westward-propagating baroclinic Rossby wave train, the eastward-propagating barotropic Rossby wave train, and the anomalous low-level circulation associated with the anomalous SLP all favor more precipitation over central Asia in the positive phase of the TPDV. But which mechanism is the primary factor remains unknown. Further efforts (e.g., sensitivity experiments) should be devoted to understanding and distinguishing the relative contributions of different processes related to the TPDV-induced precipitation signals.
Given the evidence that the TPDV and AMV are the main drivers of the interdecadal variations of central Asian precipitation, this raises the possibility of predicting overall trends in central Asian precipitation in the next few decades. If both the TPDV and AMV shift from negative phase to positive phase, an increase of precipitation will be observed over central Asia. If the TPDV and AMV change out of phase (e.g., the TPDV shifts from negative phase to positive phase and the AMV shifts from positive to negative), central Asian precipitation will increase at a smaller rate. The TPDV has shifted into its cold phase since the 2000s and has turned into a warm phase in recent years ([
This work is supported by the National Natural Science Foundation of China under Grants 41988101 and 41775091 and the Chinese Academy of Sciences under Grant XDA20060102.
By Jie Jiang; Tianjun Zhou; Xiaolong Chen and Bo Wu
Reported by Author; Author; Author; Author