Future impacts of two types of El Niño on East Asian rainfall based on CMIP5 model projections

Abstract

In this study, future change of El Niño-related East Asian (EA) rainfall and the diversity of this relationship are investigated on the basis of the historical and representative concentration pathway 8.5 (RCP 8.5) simulations taken from the Coupled Model Intercomparison Project phase 5 (CMIP5). By evaluating the East Asian Summer Monsoon (EASM) climatology and interannual variations in simulations contributing to CMIP5, nine models are verified to be capable of reproducing El Niño diversity and EASM simultaneously. Six of these models are selected for projecting the multi-model ensemble (MME) mean of two types of El Niño-related EA/western North Pacific (WNP) rainfall patterns and low-level atmospheric circulations under global warming, considering the realism in their simulated El Niño and EASM phenomena. It was found that, under a warmer background climate, the general patterns of anomalous circulation and rainfall will persist, but with amplification of the rainfall intensity during mature boreal winter and decaying summer for both Eastern-Pacific (EP) and Central-Pacific (CP) El Niño. Amplification of CP type-related rainfall seems to be stronger than that for EP type El Niño. Further analyses show that a moister atmosphere tends to always strengthen the rainfall variations for both El Niño flavors, regardless of how the El Niño-related circulation amplitude is modulated in various seasons. However, in boreal summer during the El Niño decaying phase, strengthened anomalous circulation also enhances the rainfall variability, with an effect comparable to the background moisture increase. Some of these atmospheric circulation changes might be associated with modified sea surface temperature anomalies (SSTA) of El Niño and its diversity, under global warming. Our results indicate the importance of better preparedness and higher resilience in the EA region to enhanced El Niño-induced hydrological variations under a warming climate.

Appendix: Moisture budget decomposition

For historical and future runs in each model, original values of monthly mean wind velocity and specific humidity during El Niño can be expressed as $\overrightarrow{V}=\overrightarrow{{V}_{c}}+\overrightarrow{{V}^{{\prime }}}$ and $q={q}_{c}+{q}^{{\prime }}$, respectively, where $\overrightarrow{{V}_{c}}$ and ${q}_{c}$ denote their climatologies in each model, $\overrightarrow{{V}^{{\prime }}}$ and ${q}^{{\prime }}$ represent the respective deviations due to El Niño from the corresponding climatology of each model. Thus, the moisture flux convergence in Eq. (1) can be expanded as:

$$-\frac{1}{g{\rho }_{w}}\nabla \cdot {\int }_{{p}_{s}}^{{p}_{t}}\left(q\overrightarrow{V}\right)dp=-\frac{1}{g{\rho }_{w}}\nabla \cdot {\int }_{{p}_{s}}^{{p}_{t}}\left({q}_{c}+{q}^{{\prime }}\right)\left(\overrightarrow{{V}_{c}}+\overrightarrow{{V}^{{\prime }}}\right)dp=-\left(\frac{1}{g{\rho }_{w}}\nabla \cdot {\int }_{{p}_{s}}^{{p}_{t}}\left({q}_{c}\overrightarrow{{V}_{c}}\right)dp+\frac{1}{g{\rho }_{w}}\nabla \cdot {\int }_{{p}_{s}}^{{p}_{t}}\left({q}^{{\prime }}\overrightarrow{{V}_{c}}\right)dp+\frac{1}{g{\rho }_{w}}\nabla \cdot {\int }_{{p}_{s}}^{{p}_{t}}\left({q}_{c}\overrightarrow{{V}^{{\prime }}}\right)dp+\frac{1}{g{\rho }_{w}}\nabla \cdot {\int }_{{p}_{s}}^{{p}_{t}}\left({q}^{{\prime }}\overrightarrow{{V}^{{\prime }}}\right)dp\right).$$

(9)

The two terms, $-\frac{1}{g{\rho }_{w}}\nabla \cdot {\int }_{{p}_{s}}^{{p}_{t}}\left({q}_{c}\overrightarrow{{V}^{{\prime }}}\right)dp$ (related to dynamic effect) and $-\frac{1}{g{\rho }_{w}}\nabla \cdot {\int }_{{p}_{s}}^{{p}_{t}}\left({q}^{{\prime }}\overrightarrow{{V}_{c}}\right)dp$ (related to thermodynamic effect), are major contributors to the deviations of moisture flux convergence in the models, while the term by transient eddies $\frac{1}{g{\rho }_{w}}\nabla \cdot {\int }_{{p}_{s}}^{{p}_{t}}\left({q}^{{\prime }}\overrightarrow{{V}^{{\prime }}}\right)dp$ is neglectable since both ${q}^{{\prime }}$ and $\overrightarrow{{V}^{{\prime }}}$ are small deviations from their climatologies, and $\frac{1}{g{\rho }_{w}}\nabla \cdot {\int }_{{p}_{s}}^{{p}_{t}}\left({q}_{c}\overrightarrow{{V}_{c}}\right)dp$ is stationary within the simulation period chosen. Finally, the El Niño-related anomalous moisture flux convergence can be expressed as:

$$-\frac{1}{g{\rho }_{w}}{\int }_{{p}_{s}}^{{p}_{t}}\nabla \cdot {\left(q\overrightarrow{V}\right)}^{\prime}dp\approx -\left(\frac{1}{g{\rho }_{w}}{\int }_{{p}_{s}}^{{p}_{t}}\nabla \cdot \left({q}^{{\prime }}\overrightarrow{{V}_{c}}\right)dp+\frac{1}{g{\rho }_{w}}{\int }_{{p}_{s}}^{{p}_{t}}\nabla \cdot \left({q}_{c}\overrightarrow{{V}^{{\prime }}}\right)dp\right).$$

(10)

Thus, we can have the anomalous moisture flux convergence and its dynamical and thermodynamical component for both types of El Niño under present and future scenarios. According to Eq. (4), the future changes of anomalous moisture transport can be approximated as:

$$\frac{1}{g{\rho }_{w}}\delta \left(\nabla \cdot {\int }_{{p}_{s}}^{{p}_{t}}{\left(q\overrightarrow{V}\right)}^{\prime}dp\right)\approx \frac{1}{g{\rho }_{w}}\delta \left(\nabla \cdot {\int }_{{p}_{s}}^{{p}_{t}}\left({q}^{{\prime }}\overrightarrow{{V}_{c}}\right)dp\right)+\frac{1}{g{\rho }_{w}}\delta \left(\nabla \cdot {\int }_{{p}_{s}}^{{p}_{t}}\left({q}_{c}\overrightarrow{{V}^{{\prime }}}\right)dp\right).$$

(11)

To better understand how El Niño-related rainfall will be modified in a warmer climate, the perturbations of dynamical term (1st term on RHS) and thermodynamical term (2nd term on RHS) due to climate change can be further separated as follows:

$$\frac{1}{g{\rho }_{w}}\delta \left(\nabla \cdot {\int }_{{p}_{s}}^{{p}_{t}}\left({q}_{c}\overrightarrow{{V}^{{\prime }}}\right)dp\right)=\frac{1}{g{\rho }_{w}}\left(\nabla \cdot {\int }_{{p}_{s}}^{{p}_{t}}\left({q}_{cF}\overrightarrow{{V}_{F}^{\prime}}\right)dp\right)-\frac{1}{g{\rho }_{w}}\left(\nabla \cdot {\int }_{{p}_{s}}^{{p}_{t}}\left({q}_{cH}\overrightarrow{{V}_{H}^{\prime}}\right)dp\right)=\frac{1}{g{\rho }_{w}}\nabla \cdot {\int }_{{p}_{s}}^{{p}_{t}}{q}_{cH}\left(\overrightarrow{{V}_{F}^{\prime}}-\overrightarrow{{V}_{H}^{\prime}}\right)dp+\frac{1}{g{\rho }_{w}}\nabla \cdot {\int }_{{p}_{s}}^{{p}_{t}}\overrightarrow{{V}_{F}^{\prime}}\left({q}_{cF}-{q}_{cH}\right)dp=\frac{1}{g{\rho }_{w}}\nabla \cdot {\int }_{{p}_{s}}^{{p}_{t}}{q}_{cH}\delta \left(\overrightarrow{{V}^{{\prime }}}\right)dp+\frac{1}{g{\rho }_{w}}\nabla \cdot {\int }_{{p}_{s}}^{{p}_{t}}\overrightarrow{{V}_{F}^{{\prime }}}\delta \left({q}_{c}\right)dp \left(A4\right)$$

(12)

and

$$\frac{1}{g{\rho }_{w}}\delta \left(\nabla \cdot {\int }_{{p}_{s}}^{{p}_{t}}\left({q}^{{\prime }}\overrightarrow{{V}_{c}}\right)dp\right)=\frac{1}{g{\rho }_{w}}\left(\nabla \cdot {\int }_{{p}_{s}}^{{p}_{t}}\left({q}_{F}^{{\prime }}\overrightarrow{{V}_{cF}}\right)dp\right)-\frac{1}{g{\rho }_{w}}\left(\nabla \cdot {\int }_{{p}_{s}}^{{p}_{t}}\left({q}_{H}^{{\prime }}\overrightarrow{{V}_{cH}}\right)dp\right)=\frac{1}{g{\rho }_{w}}\nabla \cdot {\int }_{{p}_{s}}^{{p}_{t}}{q}_{F}^{{\prime }}\left(\overrightarrow{{V}_{cF}}-\overrightarrow{{V}_{cH}}\right)dp+\frac{1}{g{\rho }_{w}}\nabla \cdot {\int }_{{p}_{s}}^{{p}_{t}}\overrightarrow{{V}_{cH}}\left({q}_{F}^{{\prime }}-{q}_{H}^{{\prime }}\right)dp=\frac{1}{g{\rho }_{w}}\nabla \cdot {\int }_{{p}_{s}}^{{p}_{t}}\overrightarrow{{V}_{cH}}\delta \left({q}^{{\prime }}\right)dp+\frac{1}{g{\rho }_{w}}\nabla \cdot {\int }_{{p}_{s}}^{{p}_{t}}{q}_{F}^{{\prime }}\delta \left(\overrightarrow{{V}_{c}}\right)dp \left(A5\right).$$

(13)

Thus, Eq. (3) can be transformed as follows:

$$ \begin{gathered} \iint {\delta ({\text{P}}^{{\prime}} \hat{{\text{P}}}^{{\prime}} )dxdy = \iint { - \frac{1}{{g\rho _{w} }}\delta \left( {\upphi \left( {\left( {q\vec{V}} \right)^{\prime } } \right)\hat{{\text{P}} }^{{\prime}} } \right)dxdy + }\iint {\delta \left( {E^{\prime}\hat{{\text{P}} }^{{\prime}} } \right)dxdy = }} \hfill \\ \iint {\left( { - \frac{1}{{g\rho _{w} }}\left( {\delta \left( {\upphi \left( {q^{\prime}\overrightarrow {{V_{c} }} } \right)\hat{{\text{P}} }^{{\prime}} } \right) - \frac{1}{{g\rho _{w} }}\delta \left( {\upphi \left( {q_{c} \overrightarrow {{V^{\prime}}} } \right)\hat{{\text{P}} }^{{\prime}} } \right)} \right)} \right)dxdy + \iint {\delta \left( {E^{\prime}\hat{{\text{P}} }^{{\prime}} } \right)dxdy = }} \hfill \\ \iint {\left( { - \frac{1}{{g\rho _{w} }}\upphi \left( {q_{{cH}} \delta \left( {\overrightarrow {{V^{\prime}}} } \right)} \right)} \right)\hat{P}^{\prime}_{H} dxdy + }\iint {\left( { - \frac{1}{{g\rho _{w} }}\upphi \left( {\overrightarrow {{V_{F} ^{\prime } }} \delta \left( {q_{c} } \right)} \right)} \right)\hat{P}^{\prime}_{H} dxdy + } \hfill \\ \iint {\left( { - \frac{1}{{g\rho _{w} }}\upphi \left( {\overrightarrow {{V_{{cH}} }} \delta \left( {q^{\prime}} \right)} \right)} \right)\hat{P}^{\prime}_{F} dxdy + }\iint {\left( { - \frac{1}{{g\rho _{w} }}\upphi \left( {q^{\prime}_{F} \delta \left( {\overrightarrow {{V_{c} }} } \right)} \right)} \right)\hat{P}^{\prime}_{F} dxdy + } \hfill \\ \iint {\delta \left( {E^{\prime}\hat{{\text{P}} }^{{\prime}} } \right)dxdy,} \hfill \\ \end{gathered} $$

(14)

where ${\upvarphi }\left(\cdot \right)=\nabla \cdot {\int }_{{\text{P}}_{\text{s}}}^{{\text{P}}_{\text{t}}}\left(\cdot \right)\text{d}\text{p}$ is defined for simplicity and assume $\hat{P}_{H}^{{\prime}}$ and $\hat{P}_{F}^{{\prime}}$ are identical with each other. Thus, the relative contributions of the different moisture transport processes in determining the future changes of El Niño-related rainfall anomalies can be quantified and compared.