1.1 The cloud cover trend: external forcing and internal variability
Clouds can influence top of atmosphere (TOA) energy budget through their radiative effects and hence impact the climate system. The disagreement of the future projections between global climate models (GCMs) is mostly due to the diversity of clouds and the accompanied radiative effects (Andrews et al. 2012; Bellomo et al. 2014). Bony and Dufresne (2005) proposed that it is the marine boundary layer clouds that contribute to most of the inter-model spread of cloud radiative effects. Given the large uncertainty and strong impacts of clouds, understanding the response of clouds to global warming is critical for future projections.
Since the beginning of the satellite record in 1979, the reliability of observational clouds data had improved significantly (Schiffer and Rossow 1983). The observation, aside from climate models, provides another way to estimate the cloud responses to
anthropogenic forcing. However, in addition to external forcing, the observed cloud cover trend may also be influenced by low-frequency internal variability which has comparable timescale with the satellite records. Based on the observed sea surface temperature (SST) in the 20th century, the top three dominant modes of low-frequency variations include global warming, Inter-decadal Pacific Oscillation (IPO) and Atlantic Multi-decadal Oscillation (AMO), explaining about 56%, 10% and 8% of low-frequency SST variation, respectively (Parker et al. 2007). During the satellite era, in addition to the raising of global mean
2
in Pacific (Trenberth et al. 2014), suggesting that changes of climate system during satellite era may be also largely influenced by IPO. As external forcing and internal variability both contribute to the observed cloud cover trend, a better understanding of cloud changes associated with IPO, which represents the most important internal variability over the Pacific basin, may be helpful for understanding the observed trend of cloud cover.
In previous literature, the relationship between IPO and clouds had not yet been explored thoroughly, due to limited time and spatial converge of cloud observations. In this study, we investigate the cloud responses to IPO and global warming by GCM simulations, and the cloud responses to IPO will be investigated in detail. We will further estimate that, given the phase switch of IPO and the anthropogenic warming trend observed during satellite era, how will IPO and global warming contribute to the observed cloud cover trend.
1.2 IPO and clouds
The IPO is the dominant decadal time scale SST variation over the Pacific basin. It is characterized by an ENSO-like SST pattern, but with the magnitude of SST anomaly in the tropics comparable with which in mid-latitudes (Dong and Dai 2015; Mantua et al.
1997; Newman et al. 2015). The IPO can influence climate in nearby area (Dai 2013;
Newman et al. 2015), and have global impacts on the climate system (Dong and Dai 2015).
Throughout the past 100 years, IPO accounts for most of the observed discrepancy in global mean temperature from the anthropogenically forced climate response (Kosaka and Xie 2013). The global mean surface temperature has much smaller trend after 1998,
when the phase change of IPO occurs. Moreover, the SST difference before and after 1998 is very IPO-like, and the geopotential height at 300 hPa also exhibits wave-like pattern triggered in the tropical central Pacific and propagating poleward, which coincident with the character of IPO, suggesting an IPO impact on climate system (Trenberth et al. 2014).
England et al. (2014) shows that the strengthening in Pacific trade winds accompanied by the phase shift of IPO increases the ocean heat uptake, and result in the global warming hiatus in past two decades.
Despite limited time and spatial converge, there are a few observational evidence demonstrating the influence of IPO on clouds. Using Synoptic Cloud Reports Archive from ships, Norris (2000) shows that, in the northern Pacific, the decadal variation of storm tracks, nimbostratus frequency and marine stratiform cloudiness is significantly correlated with the decadal variation of SST. In Clement et al. (2009), cloud fraction observed by satellite (International Satellite Cloud Climatology Project, ISCCP) is regressed on smoothed area mean SST in subtropical Northeast Pacific, which is correlated with IPO.
The result suggests, during positive phase of IPO, cloud fraction decreases in Maritime Continent and stratocumulus zone in the eastern Pacific, and increases in the tropical western Pacific and the tropical central Pacific.
However, it is challenging to obtain an accurate global view for the cloud changes associated with IPO using the limited length of satellite observation. The GCMs, on the other hand, are complimentary tools for investigating the problem, as they can provide pure natural variability signal by experimental design, and are able to generate unlimited length
4
1.3 Global warming and clouds
The global warming is the increase in Earth’s average surface temperature due to anthropogenic emission of green-house gases (Eickemeier et al. 2014; Lashof and Ahuja 1990). The global warming can largely influence the general circulation and hydrological cycle, and has great impact on global climate (Bengtsson, Hodges, and Roeckner 2006;
Held and Soden 2006; Lau and Kim 2015; Marvel et al. 2015).
In previous studies, some robust characters about how climate system responds to global warming are investigated. Chang et al. (2012) showed the storm track will shift poleward under global warming. Lau and Kim (2015) demonstrated that, the global warming is accompanied by the strengthening and narrowing of deep-tropics, the
weakening subsidence in the subtropics, and the expansion of Hadley cell in mid-latitude.
Marvel et al. (2015) investigated the latitude where the maximum and minimum of zonal mean total cloud fraction locates, and found that under global warming, the latitude of maximum and minimum zonal mean total cloud fraction will shift poleward, the height of high cloud in these latitude will rise, and the cloud fraction increases (decreases) in the latitude of maximum (minimum) zonal mean total cloud fraction.
Despite these robust features under global warming, the global mean of net cloud radiative effect (CRE) is very different across models, contributes to most of the
uncertainty in climate sensitivity (Andrews et al. 2012). The inter-model spread of net CRE is mostly contributed by the marine boundary layer clouds (Bony and Dufresne 2005;
Zelinka et al. 2013). Bellomo et al. (2014) further investigated several regions with larger
cloud amount feedback, which represents the changes of CRE due to cloud fraction
changes, in the period of 1954~2005 in observation and CMIP5 historical simulations, and found large disagreement between models. The CREs behavior can be different between models, even when the cloud cover responses are consistent.
1.4 The scope of the study
The goal of this study is to investigate the relationships between clouds and IPO, and estimate how IPO contributes to the observed cloud cover trend. To reach this goal, we first analyze the 500-year pre-industrial GCM simulations to construct a thorough
understanding of the large-scale fields and the cloud changes associated with IPO. The relationships between clouds and large-scale fields are examined (Chapter 3.1). The global warming impacts on clouds are investigated to compare with those of IPO (Chapter 3.2).
We then estimate the contributions of IPO-related and global warming related cloud changes in satellite observation, using reanalysis and observational data (Chapter 4).
We attempt to address the following questions: How do clouds respond to IPO?
How important is the IPO-related cloud cover trend in observation? What is the relative contribution of IPO and global warming to the cloud cover trend observed by satellite?
How do we make use of observational data to evaluate the reliability of GCMs?
6