QIAN Chengcheng, and CHEN Ge

Department of Marine Technology, College of Information Science and Engineering, Ocean University of China, Qingdao 266100, P. R. China

Warmer-Get-Wetter or Wet-Get-Wetter? A Criterion to Classify Oceanic Precipitation

QIAN Chengcheng, and CHEN Ge*

Department of Marine Technology, College of Information Science and Engineering, Ocean University of China, Qingdao 266100, P. R. China

In this study, the temporal and spatial variations of observed global oceanic precipitation during 1979–2010 are investigated. It is found that the global trend in precipitation during this period varies at a rate of 1.5%/K of surface warming while the rate is 6.6%/K during 2006–2010. The precipitation is highly correlated with Sea Surface Temperature (SST) in both the temporal and the spatial patterns since the strong 1997–98 El Ni?o event. Considering the distributions of precipitation and SST, seven oceanic regions are classified and presented using the observed Global Precipitation Climatology Project (GPCP) data and Extended Reconstructed Sea Surface Temperatures, version 3 (ERSST.v3) data. Further examining the mechanisms of the classified oceanic precipitation regions is conducted using the Tropical Rainfall Measuring Mission (TRMM) satellite, GFDL-ESM-2G model precipitation and SST data and Hadley Center sea ice and SST version 1 (HadISST1) data. More than 85% of global oceanic precipitations are controlled by either one or both of the warmer-get-wetter mechanism and wet-get-wetter mechanism. It is estimated that a 0.5 SST signal-tonoise ratio, representing the trend of SST time series to the standard deviation, is a criterion to distinguish the mechanism of a region. When the SST ratio is larger than 0.5, the precipitation of this region is controlled by the warmer-get-wetter mechanism. SST, rather than the humidity, is the pivotal factor. On the other hand, when the SST ratio is less than 0.5, the precipitation is controlled by the wet-get-wetter mechanism. The SST variability is a significant factor contributing to the precipitation variation.

oceanic precipitation; criterion; global warming; SST

1 Introduction

It is more difficult to forecast precipitation than any other parameters because of the intermittent, localized and rapid variability of the process. As one of the most important parameters of the hydrological cycle and being crucial to understand the variability of weather and climate, precipitation influences the spatial distribution of local water resources, the formation and evolution of ecological environment. The precipitation has decreased by 6% to 7% within the area from the equator to 30?N since the 1930s and precipitation extremes will continue to increase for the foreseeable future. The changes in precipitation can have subversive influences on the human society and the environment (Minet al., 2011; Petersonet al., 2008; Parryet al., 2007). Under global warming, understanding these changes is critical for reliable prediction of future changes.

With global warming, climate models and satellite observations both reveal a distinct link between precipitation and temperature, with heavy precipitation events increasing during warm periods and decreasing during cold periods (Allan and Soden, 2008). For global averages, the total amount of water in the atmosphere will increase at a rate of 7%/K of surface warming (Mitchellet al., 1987; Allen and Ingram, 2002; Held and Soden, 2006; Wentz and Schabel, 2000; Trenberthet al., 2005). However, climate models also predict that global precipitation will increase at a rate of 1 to 3%/K, several times lower than that of the total water. Water vapor variability relates strongly to changes in SST, in terms of both spatial structure of trends and temporal variability (Trenberthet al., 2005).

To precisely forecast future precipitation changes, it is important to gain an insight into the mechanisms of precipitation changes (Muller and O’Gorman, 2011). Through examining precipitation under global warming using modeling tools and observations, several mechanisms were discovered. Many of these mechanisms have been discussed to understand the change in global average precipitation by considering the thermodynamic component and dynamic feedback. On the one hand, the rich-get- richer mechanism, or called the wet-get-wetter mechanism, was examined in different approximations (Chouet al., 2009). Climate models reflect the tendency of precipitation to increase in convergence zones with large climatologicalprecipitation and to decrease in subsidence regions (Chou and Neelin, 2004; Held and Soden, 2000; Meehlet al., 2007). On the other hand, the warmer-get-wetter mechanism interpreted the cause of precipitation pattern. The precipitation changes in tropical region are positively correlated with the spatial deviations of SST warming relative to the tropical mean because the global moist instability is determined by relative SST changes (Xieet al., 2010). Huanget al.(2013) found that the seasonal mean rainfall combines the wet-get-wetter and warmer-get- wetter trends. Using an intermediate climate model, Neelinet al.(2003) identified a mechanism, the upped-ante mechanism, which causes the regional reduction in precipitation at the margins of convection zones during warming. Another mechanism named as the deepening of convection shows that the effect of deepened convection increases the gross moist stability and, therefore, stabilizes the atmosphere (Chou and Chen, 2010).

Previous studies commonly used climate models to investigate changes in precipitation and provide possible mechanisms, especially for those in tropical area. The start of the TRMM in 1997 ushered in a new era of satellite precipitation observations. A semi-quantitative retrieval method of precipitation was developed by using the TOPEX dataset in the 1990s and improved recently (Chenet al., 1997, 2003; Quartly, 2010). In this study, the global scale changes in precipitation under global warming are analyzed using the observed data. The study domain is divided into regions according to the causes of precipitation change, and physical mechanisms are discussed for each region. Data and method used here are described in Section 2. Temporal and spatial changes in precipitation are shown in Section 3. The criterion to distinguish mechanisms for changes in precipitation is described in Section 4, which is followed by the conclusions.

2 Data and Method

The main dataset used in this study is the monthly precipitation data from GPCP during 1979 to 2010. On a global 2.5?×2.5? gird, the data are combined with various other data: microwave-based data from the Special Sensor Microwave Imager (SSM/I), the infrared (IR) rainfall data from the geostationary and polar-orbiting satellites, and the surface rain gauges from the Global Historical Climatology Network (GHCN), the Climate Anomaly Monitoring System (CAMS) and the Global Precipitation Climatology Centre (GPCC). The precipitation data from the TRMM satellite are also used in this study. The TRMM launched in 1997 provides a uniquely accurate rainfall data of tropical region. For this study, the TRMM data and other sources of rainfall product or 3B43 are used. The 3B43 data is a time-series of monthly average at 0.25?×0.25? grid that covers the globe from 50?N to 50?S. In addition, the model-simulated precipitation results (GFDL-ESM-2G data), produced by Geophysical Fluid Dynamics Laboratory (GFDL), are used in the study, which covers the globe from 1861 to 2005. The monthly SST data, the HadISST1 data and model SST data are also used in the study. The monthly SST data are obtained from the observed dataset ERSST supported by the program for Comprehensive Ocean-Atmosphere Data Set (COADS) from 1854 to the present. The HadISST1 data, a blend of historical SST and current SST observations, are partly from historical ship and air-borne measurements and partly from satellite data, extending from 1870 to the present. The model SST data is from the GFDL-ESM-2G model, which covers the globe from 60?N to 60?S. This study focuses on the period of 1979 to 2010. The above datasets have been reconstructed on a 1?×1? grid by applying the optimal reanalysis method.

Global precipitation, however, is not regulated by the availability of moisture but by the atmospheric energy balance (Allen and Ingram, 2002; Johnet al., 2009). For this analysis, the monthly series of SST are chosen to compare with oceanic precipitation. Because the TRMM and GFDL-ESM-2G precipitation datasets do not cover the entire study period from 1979 to 2010, the TRMM dataset from 1998 to 2010 and the GFDL-ESM-2G dataset from 1979 to 2005 are used. The precipitation and SST data are then interpolated on a 1?×1? grid using an inverse distance weighting scheme with the same land-sea mask. To reveal the characteristics of climatology variations, the Z-score method is used for standardization. The monthly Z-score formula is:

wheredTis the trend of SST time series andσis the standard deviation of the detrended yearly SST.

3 Characteristics of Precipitation Variation

The monthly time series of oceanic precipitation and SST index are averaged over globe and plotted in Fig.1. The Clapeyron-Clausius (C-C) equation predicts that precipitation over the ocean will increase by 6.5%/K of surface warming. The calculations of this study show that the oceanic precipitation varies at a rate of 1.5%/K of surface warming for the 30-year period and 6.6%/K for the period 2006–2010, and the latter rate is close to the C-C results. Previous climate model simulations revealed that water vapor increases at a rate of 7.5%/K with air surface temperature warming (Allen and Ingram, 2002), while Wentzet al.(2007), using SSM/I data, estimated the precipitation and found that total atmospheric waterincreased at about the same rate of 7%/K from 1987 to 2006, which is also close to the results of this research for the period 2006–2010. The differences between these calculations might result from diverse data sources and different periods used to calculate the rates of global precipitation (Liepert and Previdi, 2009) because the shortterm variability and long-term trend in the datasets may not be well distinguished in a short period of time. The general impression from Fig.1 is that there is no obvious correlation between the time series of precipitation and SST, while, after a strong El Ni?o in 1997–98, the correlation becomes clear. The calculated correlation coefficient for the entire period is 0.21 but it is 0.72 for the period after the 1997–98 El Ni?o event.

As a relative measure of the interannual variability of global oceanic precipitation, the spatial differences of each interannual period are shown in Fig.2. Based on the time that data were collected, the datasets are divided into six parts: 1981–1985, 1986–1990, 1991–1995, 1996–2000, 2001–2005, and 2006–2010. It is well understood that on global average, the greenhouse warming is associated with increased SST to which precipitation increases in response (Xieet al., 2010). But as Fig.2 shows, the SST warming and precipitation increase are only tightly linked in the last 4 spatial patterns. The oceanic precipitation is controlled by the cooling of atmosphere and the increase in SST, which is clearly validated by the results after the quick SST warming in 1997 (Fig.2). In the tropical area, precipitation is decreasing in the western Pacific of the first three patterns (Figs.2(a)–(c)) while increasing in the eastern Pacific, and the trend is reversed in the last three patterns (Figs.2(d)–(f)). One of the key points to identify the regional climate change pattern in the tropical Pacific is to pin down the Walker circulation change (Tokinagaet al., 2012). The first three patterns shown in Fig.2 are associated with the atmospheric response to the weaken-ing of the Walker circulation while the last three patterns are in response to the strengthening of the Walker circulation. These patterns also reveal some evidence of the Walker circulation change.

Fig.1 The monthly time series of precipitation (GPCP data, blue line) and SST (ERSST data, red line). The data were smoothed using a 5-month moving average.

Fig.2 The spatial patterns of the differences between the 5-year and 30-year mean values. The orange and the green colors indicate the areas where the ERSST increases by more than 0.2? and decreases by more than 0.2?, respectively. The red and blue contour lines correspond to 0.3 mm d-1and -0.3 mm d-1of GPCP precipitation data, respectively.

4 Criterion of Precipitation Variation Mechanisms

Spatial variations in precipitation changes over the globe are non-uniform. In order to study the main mechanisms of precipitation change, the thermodynamic and dynamic processes are taken into account, and the study area is divided into seven regions based on the GPCP precipitation and ERSST data (Table 1). The wetter zone is positive precipitation anomalies region and represented by Y sign while the others are represented by N sign. The warmer zone is where the SST warming exceeds the global mean at any given time and represented by Y sign while the others are represented by N sign. The wet zone, mainly referring to the convergence zone, is defined asand represented by Y sign, while the others, mainly referring to the non-convergence zones, are defined asand represented by N sign, whereis the mean precipitation value at each grid over the entire study period,is the global mean precipitation value over the entire study period. As shown in Fig.3 and Table 1, region 1 has positive precipitation anomalies and larger SST warming while region 3 has smaller SST warming in the convergence zones. Region 2 has negative precipitation anomalies and larger SST warming while region 4, which could not be identified, has smaller SST warming in the convergence zones. Over the non-convergence zones, region 5 has positive precipitation anomalies associated with positive SST warming while region 7 has positive precipitation anomalies associated with negative SST warming. Regions 6 and 8 have negative precipitation anomalies in non-convergence zones. Region 6 corresponds to positive while region 8 negative SST warming.

Table 1 Classification of precipitation areas

Fig.3 Classification of precipitation area (Table 1). The global ocean is divided into seven regions using the GPCP and ERSST datasets from 1979 to 2010, and region 4 does not exist.

To understand the precipitation changes in these seven precipitation regions, the wet-get-wetter and warmer-getwetter mechanisms are considered. According to the definition of the mechanisms, the regions 3 and 6 are controlled by the wet-get-wetter, region 5 by the warmer-getwetter mechanism, and regions 1 and 8 by both mechanisms. An SST ratio is introduced in the study to understand the relationship between the two mechanisms and the precipitation classification. Fig.4 shows the ratio of the trend to the standard deviation of the ERSST dataset for 1979–2010. The classified regions overlay the ratio contours in Fig.5(a). Averaging the ratios along the boundaries between the regions, it can be seen that the ratios in regions 3, 6 and 8 are smaller than 0.5 and these regions are all controlled by the wet-get-wetter mechanism while the ratios in regions 1 and 5 are larger than 0.5 with both regions being controlled by the warmer-getwetter mechanism (Figs.4, 5(a), and Table 2). To validate the above results, further analysis was conducted using the TRMM and GFDL-ESM-2G precipitation datasets and the HadISST1 dataset and GFDL-ESM-2G SST datasets. The corresponding results are shown in Figs.5(b), 5(c), and Table 2, and in Fig.6 and Table 3, and in Fig.7 and Table 4 respectively.

Fig.4 The SST ratio (the trend to the standard deviation of the ERSST observations for the period 1979–2010).

Fig.5 The classified areas overlay the ERSST ratio contours. (a) the GPCP precipitation dataset from 1979 to 2010, (b) the TRMM precipitation dataset from 1998 to 2010, and (c) the GFDL-ESM-2G precipitation dataset from 1979 to 2005.

An intercomparison among the GPCP, TRMM and GFDL-ESM-2G precipitation datasets is conducted. The classified regions overlaid on the ratio contours of the ERSST, HadISST1 and GFDL-ESM-2G SST data areshown in Figs.5, 6 and 7, respectively. Because the GPCP is a combination of spaceborne and ground-based precipitation observations, and the TRMM and GFDL-ESM-2G datasets are satellite observations and model results respectively, the three datasets are considered independent. Fig.5 shows that the major zonal features of these datasets are rather similar while the discrepancies with the TRMM results can be seen around the southeastern subtropical Indian Ocean, and the GFDL-ESM-2G precipitation appears to be more complicated than the two others in the tropical Atlantic Ocean. Quantitatively, the area percentage that each region contributes for the three precipitation regimes varies from 0.4 (region 2 of GPCP and TRMM) to 11.5 (region 3 of GPCP and TRMM) (Table 5). The correlation coefficient is 0.7639 between the percentages of the GPCP and TRMM, 0.9218 between the GPCP and GFDL-ESM-2G, and 0.7291 between the TRMM and GFDL-ESM-2G regions (Table 6). Even different, all three datasets show that the ratio of 0.5 can be an approximate critical value to distinguish the two mechanisms. Considering the classification standard, regions 1, 3, 6 and 8 are controlled by the wet-get-wetter mechanism while regions 1, 5 and 8 are controlled by the warmer-get-wetter mechanism. When the SST ratio is larger than 0.5, SST becomes a more pivotal factor controlling the precipitation than humidity, such as being the cases for regions 1 and 5. When the SST ratio is less than 0.5, humidity becomes more important for controllingprecipitation such as are the cases for regions 3, 6 and 8. In addition, regions 1 and 8 both correspond to the two mechanisms. Region 1 contains a warm pool (SST>28℃) and SST is a deciding factor for precipitation in this area, while region 8 contains marine deserts (Chenet al., 2003) and humidity becomes a significant factor. Therefore, the SST ratio is larger than 0.5 in region 1. The warmer-getwetter mechanism has a greater impact on precipitation and seems to be the leading mechanism. In region 8, the SST ratio is less than 0.5 and the wet-get-wetter mechanism becomes a dominant mechanism. Region 2 has negative precipitation anomalies due substantially to the upped-ante mechanism, in which the inflows from lessmoistened descent regions shorten the time that the convective threshold is met, and tend to shift the margin of the convergence zone (Chouet al., 2009). The uppedante mechanism thus induces the upper layer moisture transport due to the horizontal moisture gradient, which results from a warm troposphere and the boundary layer moisture required to meet the increasing ‘a(chǎn)nte’ for convection. The upped-ante mechanism is associated with a dry advection from subsidence regions to convective regions and usually occurs over the margins of convective regions (Chou and Chen, 2010). The local wind may also affect region 7 and more future study is required on this topic.

Fig.6 The classified areas overlay the HadISST1 ratio contours. (a) the GPCP precipitation dataset from 1979 to 2010, (b) the TRMM precipitation dataset from 1998 to 2010, and (c) the GFDL-ESM-2G precipitation dataset from 1979 to 2005.

Fig.7 The classified areas overlay the GFDL-ESM-2G SST ratio contours. (a) the GPCP precipitation dataset from 1979 to 2010, (b) the TRMM precipitation dataset from 1998 to 2010, and (c) the GFDL-ESM-2G precipitation dataset from 1979 to 2005.

Table 2 The average ratio on the boundaries of each classified area from ERSST

Table 3 The same as Table 2, but from HadISST1

Table 4 The same as Table 2, but from GFDL-ESM-2G SST

Table 5 The percentages contributed by each classified area, from ERSST

Table 6 The correlation coefficients of percentages of classified areas between different precipitation datasets from ERSST

5 Conclusions

Based on the calculated and observed datasets over 30 years the findings of this study are as follows: 1) Global oceanic precipitation datasets indicate a variation in globalmean precipitation at a rate of 1.5%/K of surface warming from 1979 to 2010 while the rate is 6.6%/K for the last five years. 2) After an SST warming in 1997, the precipitation shows a higher correlation with SST. 3) The oceanic precipitation variation in tropical area is related to the Walker circulation. 4) An SST ratio of 0.5 is the criterion to distinguish the wet-get-wetter and warmerget-wetter mechanisms that control the classification of precipitation regions. If the SST ratio is larger than 0.5, SST is a controlling factor for precipitation. On the other hand, a SST ratio of less than 0.5 indicates that the amount of precipitation is controlled by humidity. Therefore, the SST variability is one of the most crucial factors for the precipitation variation.

Acknowledgements

This research was jointly supported by the National Basic Research Program of China (2012CB955603), the Natural Science Foundation of China (41076115) and Basic Scientific Research Operating Expenses of Ocean University of China. Comments from the two anonymous reviewers are very helpful in improving the quality of this paper. The authors also acknowledge the Data Center for producing and making available the output (listed in Data and method of this paper).

Allan, R. P., and Soden, B. J., 2008. Atmospheric warming and the amplification of precipitation extremes. Science, 321 (5895): 1481-1484.

Allen, M. R., and Ingram, W. J., 2002. Constraints on future changes in climate and the hydrologic cycle. Nature, 419 (12): 224-232.

Chen, G., Chapron, B., Tournadre, J., Katsaros, K., and Vandemark, D., 1997. Global oceanic precipitation: A joint view by TOPEX and the TOPEX microwave radiometer. Journal of Geophysical Research, 102: 457-471.

Chen, G., Ma, J., Fang, C., and Han, Y., 2003. Global oceanic precipitation derived from TOPEX and TMR: Climatology and variability. Journal of Climate, 16: 3888-3904.

Chou, C., and Chen, C. A., 2010. Depth of convection and the weakening of tropical circulation in global warming. Journal of Climate, 23 (11): 3019-3030.

Chou, C., and Neelin, J. D., 2004. Mechanisms of global warming impacts on reginoal tropical precipitation. Journal of Climate, 17: 2688-2614.

Chou, C., Neelin, J. D., Chen, C. A., and Tu, J. Y., 2009. Evaluating the ‘rich-get-richer’ mechanism in tropical precipitation change under global warming. Journal of Climate, 22 (8): 1982-2005.

Du, Y., and Xie, S. P., 2008. Role of atmospheric adjustments in the tropical Indian Ocean warming during the 20th century in climate models. Geophysical Research Letters, 35 (8), DOI: 10.1029/2008GL033631.

Held, I. M., and Soden, B. J., 2000. Water vapor feedback and global warming. Annual Review of Environment and Resources, 25: 441-475.

Held, I. M., and Soden, B. J., 2006. Robust responses of the hydrological cycle to global warming. Journal of Climate, 19: 5686-5616.

Huang, P., Xie, S.-P., Hu, K., Huang, G., and Huang, R., 2013. Patterns of the seasonal response of tropical rainfall to global warming. Nature Geoscience, DOI: 10.1038/NGEO1792.

John, V. O., Allan, R. P., and Soden, B. J., 2009. How robust are observed and simulated precipitation responses to tropical ocean warming? Geophysical Research Letters, 36, DOI: 10. 1029/2009GL038276.

Liepert, B. G., and Previdi, M., 2009. Do models and observations disagree on the rainfall response to global warming? Journal of Climate, 22 (11): 3156-3166.

Meehl, G., Stocker, T., and Collins, W., 2007. Global climate projections. In: Climate Change 2007: The Physical Science Basis. Solomon et al., eds., Cambridge University Press, New York, 747-845.

Min, S. K., Zhang, X., Zwiers, F. W., and Hegerl, G. C., 2011.

Human contribution to more-intense precipitation extremes. Nature, 470 (7334): 378-381.

Mitchell, J. F. B., Wilson, C. A., and Cunnington, W. M., 1987. On CO2climate sensitivity and model dependence of results. Quarterly Journal of the Royal Meteorological Society, 113: 293-322.

Muller, C. J., and O’Gorman, P. A., 2011. An energetic perspective on the regional response of precipitation to climate change. Nature Cliamte Change, 1: 266-271.

Neelin, J. D., Chou, C., and Su, H., 2003. Tropical drought regions in global warming and El Ni?o teleconnections. Geophysical Research Letters, 30 (24): 2275.

Parry, M. L., Canziani, O. F., Palutikof, J. P., Linden, P. J., and Hanson, C. E., 2007. Climate Change 2007: Impacts, Adaptation and Vulnerability. Cambridge University Press, Cambridge, UK, 811-842.

Peterson, T. C., Zhang, X., and Brunet, I. M., 2008. Changes in North American extremes derived from daily weather data. Journal of Geophysical Research, 113, D07113.

Quartly, G. D., 2010. Improving the altimetric rain record from Jason-1 and Jason-2. Journal of Geophysical Research, 115 (C3), DOI: 10.1029/2009JC005670.

Trenberth, K. E., Fasullo, J., and Smith, L., 2005. Trends and variability in column-integrated atmospheric water vapor. Climate Dynamics, 24 (7-8): 741-758.

Tokinaga, H., Xie, S. P., Timmermann, A., McGregor, S., Ogata, T., Kubota, H., and Okumura, Y. M., 2012. Reginal patterns of tropical Indo-Pacific climate change: Evidence of the Walker Circulation weakening. Journal of Climate, 25: 1689-1710.

Wentz, F. J., and Schabel, M., 2000. Precise climate monitoring using complementary satellite data sets. Nature, 403: 414-416.

Wentz, F. J., Ricciardulli, L., Hilburn, K., and Mears, C., 2007. How much more rain will global warming bring? Science, 317 (5835): 233-235.

Xie, S. P., Deser, C., Vecchi, G. A., Ma, J., Teng, H., and Wittenberg, A. T., 2010. Global warming pattern formation: Sea surface temperature and rainfall. Journal of Climate, 23 (4): 966-986.

(Edited by Xie Jun)

(Received October 8, 2012; revised February 4, 2013; accepted April 16, 2014)

? Ocean University of China, Science Press and Springer-Verlag Berlin Heidelberg 2014

* Corresponding author. Tel: 0086-532-66781265

E-mail: gechen@ouc.edu.cn