• Albergel, C., and Coauthors, 2013: Skill and global trend analysis of soil moisture from reanalyses and microwave remote sensing. J. Hydrometeor., 14, 12591277.

    • Search Google Scholar
    • Export Citation
  • Blunden, J., , and D. S. Arndt, 2012: State of the climate in 2011. Bull. Amer. Meteor. Soc., 93, S1S282.

  • Chiew, F., , T. Piechota, , J. Dracup, , and T. McMahon, 1998: El Nino/Southern Oscillation and Australian rainfall, streamflow and drought: Links and potential for forecasting. J. Hydrol., 204, 138149, doi:10.1016/S0022-1694(97)00121-2.

    • Search Google Scholar
    • Export Citation
  • Dai, A., 2013: Increasing drought under global warming in observations and models. Nat. Climate Change, 3, 5258, doi:10.1038/nclimate1633.

    • Search Google Scholar
    • Export Citation
  • de Jeu, R. A. M., , W. Wagner, , T. R. H. Holmes, , J. Dolman, , N. C. Giesen, , and J. Friesen, 2008: Global soil moisture patterns observed by space borne microwave radiometers and scatterometers. Surv. Geophys., 29, 399420, doi:10.1007/s10712-008-9044-0.

    • Search Google Scholar
    • Export Citation
  • de Viron, O., , J. O. Dickey, , and M. Ghil, 2013: Global modes of climate variability. Geophys. Res. Lett., 40, 18321837, doi:10.1002/grl.50386.

    • Search Google Scholar
    • Export Citation
  • Dorigo, W. A., , R. A. M. de Jeu, , D. Chung, , R. Parinussa, , Y. Y. Liu, , W. Wagner, , and D. Fernndez-Prieto, 2012: Evaluating global trends (1988–2010) in harmonized multi-satellite surface soil moisture. Geophys. Res. Lett.,39, L18405, doi:10.1029/2012GL052988.

  • Esquivel, P., , and A. R. Messina, 2008: Complex empirical orthogonal function analysis of wide-area system dynamics. Proc. 2008 IEEE Power and Energy Society General Meeting—Conversion and Delivery of Electrical Energy in the 21st Century, Pittsburgh, PA, IEEE, 1–7.

  • Field, C. B., , V. Barros, , T. F. Stocker, , and Q. Dahe, Eds., 2012: Managing the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation. Cambridge University Press, 582 pp.

  • Forootan, E., , J. Awange, , J. Kusche, , B. Heck, , and A. Eicker, 2012: Independent patterns of water mass anomalies over Australia from satellite data and models. Remote Sens. Environ., 124, 427443, doi:10.1016/j.rse.2012.05.023.

    • Search Google Scholar
    • Export Citation
  • García-García, D., , C. C. Ummenhofer, , and V. Zlotnicki, 2011: Australian water mass variations from GRACE data linked to Indo-Pacific climate variability. Remote Sens. Environ., 115, 21752183, doi:10.1016/j.rse.2011.04.007.

    • Search Google Scholar
    • Export Citation
  • Gerten, D., , S. Schaphoff, , and W. Lucht, 2007: Potential future changes in water limitations of the terrestrial biosphere. Climatic Change, 80, 277299.

    • Search Google Scholar
    • Export Citation
  • Gong, D., , and S. Wang, 1999: Definition of Antarctic Oscillation index. Geophys. Res. Lett., 26 (4), 459462.

  • Hannachi, A., , I. T. Jolliffe, , and D. B. Stephenson, 2007: Empirical orthogonal functions and related techniques in atmospheric science: A review. Int. J. Climatol., 1152, 11191152, doi:10.1002/joc.1499.

    • Search Google Scholar
    • Export Citation
  • Hendon, H. H., , D. W. J. Thompson, , and M. C. Wheeler, 2007: Australian rainfall and surface temperature variations associated with the Southern Hemisphere annular mode. J. Climate, 20, 24522467.

    • Search Google Scholar
    • Export Citation
  • Ho, M., , A. S. Kiem, , and D. C. Verdon-Kidd, 2012: The southern annular mode: A comparison of indices. Hydrol. Earth Syst. Sci., 16, 967982, doi:10.5194/hess-16-967-2012.

    • Search Google Scholar
    • Export Citation
  • Hudson, D., , O. Alves, , H. H. Hendon, , and A. G. Marshall, 2011: Bridging the gap between weather and seasonal forecasting: Intraseasonal forecasting for Australia. Quart. J. Roy. Meteor. Soc., 137, 673689, doi:10.1002/qj.769.

    • Search Google Scholar
    • Export Citation
  • Huntington, T. G., 2006: Evidence for intensification of the global water cycle: Review and synthesis. J. Hydrol., 319, 8395, doi:10.1016/j.jhydrol.2005.07.003.

    • Search Google Scholar
    • Export Citation
  • Kajikawa, Y., , B. Wang, , and J. Yang, 2009: A multi-time scale Australian monsoon index. Int. J. Climatol., 30, 11141120, doi:10.1002/joc.1955.

    • Search Google Scholar
    • Export Citation
  • Koster, R. D., and Coauthors, 2004: Regions of strong coupling between soil moisture and precipitation. Science, 305, 11381140, doi:10.1126/science.1100217.

    • Search Google Scholar
    • Export Citation
  • Legates, D. R., , R. Mahmood, , D. F. Levia, , T. L. DeLiberty, , S. M. Quiring, , C. Houser, , and F. E. Nelson, 2010: Soil moisture: A central and unifying theme in physical geography. Prog. Phys. Geogr., 35, 6586, doi:10.1177/0309133310386514.

    • Search Google Scholar
    • Export Citation
  • Liu, Y. Y., , R. A. M. de Jeu, , A. I. J. M. van Dijk, , and M. Owe, 2007: TRMM-TMI satellite observed soil moisture and vegetation density (1998–2005) show strong connection with El Niño in eastern Australia. Geophys. Res. Lett., 34, L15401, doi:10.1029/2007GL030311.

    • Search Google Scholar
    • Export Citation
  • Liu, Y. Y., , A. I. J. M. van Dijk, , R. A. M. de Jeu, , and T. R. H. Holmes, 2009: An analysis of spatiotemporal variations of soil and vegetation moisture from a 29-year satellite-derived data set over mainland Australia. Water Resour. Res., 45, W07405, doi:10.1029/2008WR007187.

    • Search Google Scholar
    • Export Citation
  • Liu, Y. Y., , R. M. Parinussa, , W. A. Dorigo, , R. A. M. de Jeu, , W. Wagner, , A. I. J. M. van Dijk, , M. F. McCabe, , and J. P. Evans, 2011: Developing an improved soil moisture dataset by blending passive and active microwave satellite-based retrievals. Hydrol. Earth Syst. Sci., 15, 425436, doi:10.5194/hess-15-425-2011.

    • Search Google Scholar
    • Export Citation
  • Liu, Y. Y., , W. Dorigo, , R. Parinussa, , R. A. M. de Jeu, , W. Wagner, , M. McCabe, , J. Evans, , and A. I. J. M. van Dijk, 2012: Trend-preserving blending of passive and active microwave soil moisture retrievals. Remote Sens. Environ., 123, 280297, doi:10.1016/j.rse.2012.03.014.

    • Search Google Scholar
    • Export Citation
  • Loew, A., , T. Stacke, , W. A. Dorigo, , R. A. M. de Jeu, , and S. Hagemann, 2013: Potential and limitations of multidecadal satellite soil moisture observations for climate model evaluation studies. Hydrol. Earth Syst. Sci. Discuss., 10, 35413594, doi:10.5194/hessd-10-3541-2013.

    • Search Google Scholar
    • Export Citation
  • Luo, J.-J., , S. Behera, , Y. Masumoto, , H. Sakuma, , and T. Yamagata, 2008: Successful prediction of the consecutive IOD in 2006 and 2007. Geophys. Res. Lett., 35, L14S02, doi:10.1029/2007GL032793.

    • Search Google Scholar
    • Export Citation
  • Meneghini, B., , I. Simmonds, , and I. N. Smith, 2007: Association between Australian rainfall and the southern annular mode. Int. J. Climatol., 27, 109121, doi:10.1002/joc.1370.

    • Search Google Scholar
    • Export Citation
  • Naeimi, V., , Z. Bartalis, , and W. Wagner, 2009: ASCAT soil moisture: An assessment of the data quality and consistency with the ERS scatterometer heritage. J. Hydrometeor., 10, 555563.

    • Search Google Scholar
    • Export Citation
  • Nan, S., , and J. Li, 2003: The relationship between the summer precipitation in the Yangtze River valley and the boreal spring Southern Hemisphere annular mode. Geophys. Res. Lett.,30, 2266, doi:10.1029/2003GL018381.

  • Nicholls, N., 1989: Sea surface temperatures and Australian winter rainfall. J. Climate, 2, 965973.

  • Nicholls, N., 2009: Local and remote causes of the southern Australian autumn-winter rainfall decline, 1958–2007. Climate Dyn., 34, 835845, doi:10.1007/s00382-009-0527-6.

    • Search Google Scholar
    • Export Citation
  • North, G. R., , T. L. Bell, , R. F. Cahalan, , and F. J. Moeng, 1982: Sampling errors in the estimation of empirical orthogonal functions. Mon. Wea. Rev., 110, 699706.

    • Search Google Scholar
    • Export Citation
  • Overland, J. E., , and R. W. Preisendorfer, 1982: A significance test for principal components applied to a cyclone climatology. Mon. Wea. Rev., 110, 14.

    • Search Google Scholar
    • Export Citation
  • Owe, M., , R. A. M. de Jeu, , and T. Holmes, 2008: Multisensor historical climatology of satellite-derived global land surface moisture. J. Geophys. Res., 113, F01002, doi:10.1029/2007JF000769.

    • Search Google Scholar
    • Export Citation
  • Rasmusson, E. M., , P. A. Arkin, , W.-Y. Chen, , and J. B. Jalickee, 1981: Biennial variations in surface temperature over the United States as revealed by singular decomposition. Mon. Wea. Rev., 109, 587598.

    • Search Google Scholar
    • Export Citation
  • Rebel, K. T., , R. A. M. de Jeu, , P. Ciais, , N. Viovy, , S. L. Piao, , G. Kiely, , and A. J. Dolman, 2012: A global analysis of soil moisture derived from satellite observations and a land surface model. Hydrol. Earth Syst. Sci., 16, 833847, doi:10.5194/hess-16-833-2012.

    • Search Google Scholar
    • Export Citation
  • Risbey, J. S., , M. J. Pook, , P. C. McIntosh, , M. C. Wheeler, , and H. H. Hendon, 2009: On the remote drivers of rainfall variability in Australia. Mon. Wea. Rev., 137, 32333253.

    • Search Google Scholar
    • Export Citation
  • Saji, N. H., , and T. Yamagata, 2003: Possible impacts of Indian Ocean dipole mode events on global climate. Climate Res., 25, 151169, doi:10.3354/cr025151.

    • Search Google Scholar
    • Export Citation
  • Saji, N. H., , B. N. Goswami, , P. N. Vinayachandran, , and T. Yamagata, 1999: A dipole mode in the tropical Indian Ocean. Nature, 401, 360363, doi:10.1038/43854.

    • Search Google Scholar
    • Export Citation
  • Simmonds, I., 1990: A modelling study of winter circulation and precipitation anomalies associated with Australian region ocean temperatures. Aust. Meteor. Mag., 38, 151161.

    • Search Google Scholar
    • Export Citation
  • Simmonds, I., , and A. Rocha, 1991: The association of Australian winter climate with ocean temperatures to the west. J. Climate, 4, 11471161.

    • Search Google Scholar
    • Export Citation
  • Simmonds, I., , and A. H. Lynch, 1992: The influence of pre-existing soil moisture content on Australian winter climate. Int. J. Climatol., 12, 3354, doi:10.1002/joc.3370120105.

    • Search Google Scholar
    • Export Citation
  • Simmonds, I., , and P. Hope, 1998: Seasonal and regional responses to changes in Australian soil moisture conditions. Int. J. Climatol., 18, 11051139, doi:10.1002/(SICI)1097-0088(199808)18:10<1105::AID-JOC308>3.0.CO;2-G.

    • Search Google Scholar
    • Export Citation
  • Solomon, S., , D. Qin, , M. Manning, , M. Marquis, , K. Averyt, , M. B. Tignor, , H. L. Miller Jr., , and Z. Chen, Eds., 2007: Climate Change 2007: The Physical Science Basis. Cambridge University Press, 996 pp.

  • Taylor, C. M., , R. A. M. de Jeu, , F. Guichard, , P. P. Harris, , and W. A. Dorigo, 2012: Afternoon rain more likely over drier soils. Nature, 489, 423426, doi:10.1038/nature11377.

    • Search Google Scholar
    • Export Citation
  • Timbal, B., , and H. Hendon, 2011: The role of tropical modes of variability in recent rainfall deficits across the Murray-Darling Basin. Water Resour. Res., 47, W00G09, doi:10.1029/2010WR009834.

    • Search Google Scholar
    • Export Citation
  • Trenberth, K., 2011: Changes in precipitation with climate change. Climate Res., 47, 123138, doi:10.3354/cr00953.

  • Troup, A. J., 1965: The ‘Southern Oscillation.' Quart. J. Roy. Meteor. Soc., 91, 490506, doi:10.1002/qj.49709139009.

  • Ummenhofer, C. C., , M. H. England, , P. C. McIntosh, , G. A. Meyers, , M. J. Pook, , J. S. Risbey, , A. Sen Gupta, , and A. S. Taschetto, 2009: What causes southeast Australia's worst droughts? Geophys. Res. Lett., 36, L04706, doi:10.1029/2008GL036801.

    • Search Google Scholar
    • Export Citation
  • van Dijk, A. I. J. M., , H. E. Beck, , R. S. Crosbie, , R. A. M. de Jeu, , Y. Y. Liu, , G. M. Podger, , B. Timbal, , and N. R. Viney, 2013: The Millennium Drought in southeast Australia (2001–2009): Natural and human causes and implications for water resources, ecosystems, economy, and society. Water Resour. Res., 49, 10401057, doi:10.1002/wrcr.20123.

    • Search Google Scholar
    • Export Citation
  • Verdon-Kidd, D. C., , and A. S. Kiem, 2009: On the relationship between large-scale climate modes and regional synoptic patterns that drive Victorian rainfall. Hydrol. Earth Syst. Sci., 13, 467479, doi:10.5194/hess-13-467-2009.

    • Search Google Scholar
    • Export Citation
  • von Storch, H., , and F. Zwiers, 2002: Statistical Analysis in Climate Research. Cambridge University Press, 496 pp.

  • Wagner, W., , J. Noll, , M. Borgeaud, , and H. Rott, 1999: Monitoring soil moisture over the Canadian Prairies with the ERS scatterometer. IEEE Trans. Geosci. Remote Sens., 37, 206216, doi:10.1109/36.739155.

    • Search Google Scholar
    • Export Citation
  • Wagner, W., , W. A. Dorigo, , R. A. M. de Jeu, , D. Fernandez, , J. Benveniste, , E. Haas, , and M. Ertl, 2012: Fusion of active and passive microwave observations to create an essential climate variable data record on soil moisture. ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol. I-7, 315–321.

  • Wilks, D., 2011: Statistical Methods in the Atmospheric Sciences. 3rd ed. Academic Press, 704 pp.

  • Yuan, Y., , and C. Li, 2008: Decadal variability of the IOD-ENSO relationship. Chin. Sci. Bull., 53, 17451752, doi:10.1007/s11434-008-0196-6.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    First CEOF mode that explains 46.3% of the total variance in the SSM field in Australia. (a) The spatial amplitude and (b) temporal amplitude. The mode's (c) spatial phase and (d) temporal phase. Dots indicate phase values at DJF for better readability.

  • View in gallery

    As in Fig. 1, but for the second CEOF mode that explains 14.5% of the total variance.

  • View in gallery

    As in Fig. 1, but for the third CEOF mode that explains 6.3% of the total variance.

  • View in gallery

    As in Fig. 1, but for the fourth CEOF mode that explains 4.2% of the total variance.

  • View in gallery

    Spearman correlation coefficient of SOI with seasonal SSM means from (a) the original data time series and from (b) the synthesized time series from the first four CEOF modes.

  • View in gallery

    Mean difference of spatial amplitude values of CEOF modes 1–4 to reference modes for all subsamples, indicated by the starting year. The scale is relative to the maximum amplitude of the four reference modes.

  • View in gallery

    Plots of imaginary parts of CPC (a) 1, (b) 3, and (c) 4, together with relevant climate indices. In (c), the Indian Ocean DMI is inverted, accounting for its negative correlation with SSM.

  • View in gallery

    Spearman correlation coefficient of (top)–(bottom) climate indices with SSM time series synthesized from the first three seasonal CEOF modes. White areas are masked out due to poor significance (p < 0.05).

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 87 87 14
PDF Downloads 54 54 8

How Oceanic Oscillation Drives Soil Moisture Variations over Mainland Australia: An Analysis of 32 Years of Satellite Observations

View More View Less
  • 1 Vienna University of Technology, Vienna, Austria
  • | 2 Fenner School of Environment and Society, Australian National University, Canberra, Australia
© Get Permissions
Full access

Abstract

Australia is frequently subject to droughts and floods. Its hydrology is strongly connected to oceanic and atmospheric oscillations (climate modes) such as the El Niño–Southern Oscillation (ENSO), Indian Ocean dipole (IOD), and southern annular mode (SAM). A global 32-yr dataset of remotely sensed surface soil moisture (SSM) was used to examine hydrological variations in mainland Australia for the period 1978–2010. Complex empirical orthogonal function (CEOF) analysis was applied to extract independent signals and to investigate their relationships to climate modes. The annual cycle signal represented 46.3% of the total variance and a low but highly significant connection with SAM was found. Two multiannual signals with a lesser share in total variance (6.3% and 4.2%) were identified. The first one had an unstable period of 2–5 yr and reflected an east–west pattern that can be associated with ENSO and SAM but not with IOD. The second one, a 1- to 5-yr oscillation, formed a dipole pattern between the west and north and can be linked to ENSO and IOD. As expected, relationships with ENSO were found throughout the year and are especially strong during southern spring and summer in the east and north. Somewhat unexpectedly, SAM impacts strongest in the north and east during summer and is proposed as the key driver of the annual SSM signal. The IOD explains SSM variations in the north, east, and southeast during spring and also in the west during winter.

Supplemental information related to this paper is available at the Journals Online website: http://dx.doi.org/10.1175/JCLI-D-13-00149.1.s1.

Corresponding author address: Bernhard Bauer-Marschallinger, Department of Geodesy and Geoinformation, Remote Sensing Research Group, Vienna University of Technology, Gusshausstrasse 27-29, 1040 Vienna, Austria. E-mail: bernhard.bauer-marschallinger@geo.tuwien.ac.at

Abstract

Australia is frequently subject to droughts and floods. Its hydrology is strongly connected to oceanic and atmospheric oscillations (climate modes) such as the El Niño–Southern Oscillation (ENSO), Indian Ocean dipole (IOD), and southern annular mode (SAM). A global 32-yr dataset of remotely sensed surface soil moisture (SSM) was used to examine hydrological variations in mainland Australia for the period 1978–2010. Complex empirical orthogonal function (CEOF) analysis was applied to extract independent signals and to investigate their relationships to climate modes. The annual cycle signal represented 46.3% of the total variance and a low but highly significant connection with SAM was found. Two multiannual signals with a lesser share in total variance (6.3% and 4.2%) were identified. The first one had an unstable period of 2–5 yr and reflected an east–west pattern that can be associated with ENSO and SAM but not with IOD. The second one, a 1- to 5-yr oscillation, formed a dipole pattern between the west and north and can be linked to ENSO and IOD. As expected, relationships with ENSO were found throughout the year and are especially strong during southern spring and summer in the east and north. Somewhat unexpectedly, SAM impacts strongest in the north and east during summer and is proposed as the key driver of the annual SSM signal. The IOD explains SSM variations in the north, east, and southeast during spring and also in the west during winter.

Supplemental information related to this paper is available at the Journals Online website: http://dx.doi.org/10.1175/JCLI-D-13-00149.1.s1.

Corresponding author address: Bernhard Bauer-Marschallinger, Department of Geodesy and Geoinformation, Remote Sensing Research Group, Vienna University of Technology, Gusshausstrasse 27-29, 1040 Vienna, Austria. E-mail: bernhard.bauer-marschallinger@geo.tuwien.ac.at

1. Introduction

Climate change is expected to accelerate the global hydrological cycle, with implications for ecosystems and feedback to regional and global climate (Huntington 2006). Precipitation is expected to be less reliable, with intensification of both dry and wet conditions and more extreme rainfall events (and floods; Trenberth 2011) and an increased risk of aridity and droughts (Dai 2013). Globally, soil moisture (a good indicator of water availability for plant growth) is expected to decrease over large parts of Earth (Gerten et al. 2007). Surface soil moisture (SSM), defined as the water in the top few centimeters soil, is very sensitive to changes in precipitation, temperature, solar irradiation, humidity, and wind. It is an important integrator of climatic conditions and can be a driver of local climate (Legates et al. 2010) and weather (Taylor et al. 2012). Variations in heat fluxes are affected by spatial and temporal dynamics in soil moisture, which alter near-surface air temperature and humidity. Consequently, soil moisture estimation can be a critical skill in temperature, humidity, and precipitation forecasts (Koster et al. 2004). Earlier research illuminated the feedbacks of large-scale soil moisture anomalies on atmospheric conditions for the particular case of Australia (Simmonds and Lynch 1992; Simmonds and Hope 1998). Soil moisture, being a state variable, has certain benefits for monitoring hydrological conditions when compared to flux variables such as precipitation or evapotranspiration, whose observation is generally more demanding. Since 2008, soil moisture is listed as an essential climate variable (ECV) within the framework of the Global Climate Observing System (GCOS; Blunden and Arndt 2012).

The hydrology of Australia is highly variable and the continent is frequently subject to droughts and floods. Recently, there have been several extreme dry and wet events, including a protracted drought from 1995–2001 to 2009 known as the “Big Dry” or “Millennium Drought” (Ummenhofer et al. 2009; van Dijk et al. 2013), continually dry conditions in southwest Western Australia since the 1970s, and extensive floods from 2009 to 2012 (e.g., Channel Country 2009/10 and Queensland 2010/11). Surrounded by oceans and located in the tropics and subtropics, Australia's hydrology responds strongly to changes in global ocean circulation patterns. Its mainly semiarid-to-arid characteristics and the lack of significant mountain ranges make it sensitive to climate modes.

Climate modes are large-scale oceanic or atmospheric oscillations that can have a global as well as a specific regional impact. Different climate modes interact directly or via teleconnections across the globe. Within a region, oceanic oscillations are coupled to atmospheric effects that appear as variations in the direction and strength of major wind patterns, giving rise to precipitation anomalies (Field et al. 2012). At least three major climate modes that are relevant for Australia's hydrology have been identified (Nicholls 2009; Risbey et al. 2009; Verdon-Kidd and Kiem 2009; an overview is given by van Dijk et al. 2013). 1) The El Niño–Southern Oscillation (ENSO) refers to climate phenomena resulting from interactions between coupled large-scale oceanic and atmospheric circulation processes in the Pacific region (Solomon et al. 2007). It is driven by cyclic cooling and warming of the equatorial Pacific (termed El Niño and La Niña phases as opposite) coupled with fluctuations of an atmospheric global-scale tropical and subtropical surface pressure pattern. El Niño and La Niña events cause dry and wet anomalies, respectively, in Australia and the entire western Pacific area (Chiew et al. 1998). 2) The Indian Ocean dipole (IOD) is an irregular cycle of warming and cooling of ocean waters in the equatorial Indian Ocean and was first described by Saji et al. (1999). The phases of the IOD oscillation are termed positive, negative, and neutral. During a positive event, evaporation, convective activities, and precipitation are reduced in Indonesia and Australia and augmented in eastern Africa (Saji and Yamagata 2003). These effects are inverted during a negative event when northwesterly winds are generated that pick up moisture from the ocean and carry it deep into Australia's southeastern parts, as described earlier by Nicholls (1989) and examined by Simmonds (1990) and Simmonds and Rocha (1991). Yuan and Li (2008) described in detail the nature of the ENSO–IOD interactions, while Ummenhofer et al. (2009) and van Dijk et al. (2013) examined the relationship of southeast Australian droughts with Indo-Pacific interactions. 3) The southern annular mode [SAM; also known as the Antarctic Oscillation (AAO) as defined by Gong and Wang (1999)] describes an oscillating pattern of weaker and stronger westerly winds between mid- and high latitudes in the Southern Hemisphere. The SAM positive phase is linked to negative sea level pressure anomalies over the polar regions and intensified westerly winds (Field et al. 2012). During a negative SAM event, an equatorward expansion of the jet stream is experienced, and the belt of (weaker) westerly winds expands toward the equator. Depending on the season of the year, this latitudinal shift of the westerly winds (and hence low-pressure systems) can alter precipitation patterns in the Southern Hemisphere.

Present studies on the impact of climate modes on the hydrology of Australia show some agreement but also contradictions: Chiew et al. (1998) linked ENSO to long-term data on rainfall, droughts, and streamflow in Australia, associating dry conditions during the latter part of the calendar year with El Niño. Liu et al. (2007, 2009) found evidence of strong connections of ENSO to remotely sensed SSM variation in the spring and winter season and to a lesser extent in autumn and summer. However, Verdon-Kidd and Kiem (2009) argued that ENSO alone cannot explain major droughts in southeast Australia (SEA). Ummenhofer et al. (2009) investigated observation and reanalysis datasets over SEA and found a clear impact of the IOD that exceeded the contribution of ENSO. García-García et al. (2011) and Forootan et al. (2012) examined water mass variations derived from the Gravity Recovery and Climate Experiment (GRACE) data. The latter found that Indo-Pacific influences on Australia's hydrology are regionally different: ENSO predominantly affects northern and eastern Australia, while IOD impacts the south and SEA. This appears to contradict the findings of Timbal and Hendon (2011), who asserted that ENSO and IOD did not contribute to rainfall deficits during the Millennium Drought as they affect spring rainfall only, whereas the main rainfall deficits were in fall and winter. This is supported by the work of Nicholls (2009), who concluded that southern Australia's rainfall variability during fall and winter cannot be explained by changes in tropical phenomena like ENSO and IOD but instead are driven by the extratropical SAM. Meneghini et al. (2007) found a positive relation between rainfall variability in northern Australia with SAM and negative relation in southern Australia but in the latter case for winter only.

For this study, we considered soil moisture as an integrator of near-surface hydrological processes over Australia, recognizing that previous studies examined the hydrology in different regions and periods and with diverse data. The relationship with precipitation is not necessarily a direct one, as soil moisture storage is the combined result of precipitation, evapotranspiration, and soil moisture redistribution. Our study aimed to clarify the contribution of climate modes on the variability of soil hydrology over all of mainland Australia by investigating the multidecadal remotely sensed SSM dataset. Key questions were 1) Does the remotely sensed SSM variation consist of separable signals? 2) If so, what are the spatial and temporal patterns of these separate signals? 3) Can these separate signals be linked to the climate modes ENSO, IOD, and SAM?

Remote sensing has demonstrated capability in observing SSM in a globally consistent and comprehensive way (de Jeu et al. 2008). Liu et al. (2012) and Wagner et al. (2012) presented a global 32-yr SSM dataset, covering the period from October 1978 to December 2010. This is the very first observation-based global dataset describing SSM over such a long period. Its data were collected from four passive and two active satellite microwave sensors. Our study was motivated by the very first possibility of analyzing remotely sensed SSM variations over a continuous period of 32 yr.

In this study, the SSM history of Australia was first derived from the multidecadal dataset and then decomposed into seasonal means. Complex empirical orthogonal function (CEOF) analysis was then applied to the SSM data to extract the principal signals in Australia's soil hydrology in the form of paired spatial and temporal functions. The CEOF technique was chosen because of its proven capability to reveal propagating patterns in geophysical data fields. Connections between the principal SSM CEOF functions and candidate climate indices were examined with Spearman rank correlation analysis, to identify the spatial distribution and the strength of their relationship.

2. Data and methods

a. Climate mode indices

Climatic oscillation patterns such as ENSO, IOD, and SAM are described by normalized indices. The index's magnitude quantifies the strength, and the sign indicates the phase of the oscillations.

The Southern Oscillation index (SOI) defined by Troup (1965) is a measure of ENSO and is determined by the differences of atmospheric pressure at Tahiti in the equatorial Pacific and Darwin in northern Australia. SOI was preferred to other ENSO indices since it is based on conditions in the western Pacific. Strong positive values are associated with La Niña episodes and strong negative values with El Niño, respectively. The monthly SOI data used in this study were available from the Australian Bureau of Meteorology (BOM; http://www.bom.gov.au/climate/enso/).

The IOD is characterized by the dipole mode index (DMI), which is the normalized difference between sea surface temperatures (SSTs) in the western and the southeastern tropical Indian Ocean (Saji et al. 1999). The phases of the IOD are named positive and negative according to the sign of the DMI. (The monthly data used here were obtained from http://www.jamstec.go.jp/frcgc/research/d1/iod/e/iod/dipole_mode_index.html.)

Finally, the southern annular mode index (SAMI) is the difference between the normalized monthly zonal mean sea level pressure (MSLP) in the Southern Ocean at 40° and 70°S, respectively. Proposed by Nan and Li (2003), SAMI is a modification of the original definition of the Antarctic Oscillation index (AOI) by Gong and Wang (1999). We chose SAMI because of the suitable time coverage and high correlation with other SAM indices as shown by Ho et al. (2012). (The SAMI data were available from http://ljp.lasg.ac.cn/dct/page/65572.)

The monthly climate indices were resampled to seasonal means to derive values for (southern) summer [December–February (DJF)], fall [March–May (MAM)], winter [June–August (JJA)], and spring [September–November (SON)] for the period from December 1978 to November 2010.

b. ECV SM multidecadal surface soil moisture

The remotely sensed 32-yr SSM dataset used here was constructed and distributed as part of the European Space Agency's (ESA) Water Cycle Multimission Observation Strategy (WACMOS) and Soil Moisture Climate Change Initiative (Soil Moisture CCI) projects (Liu et al. 2012). The dataset [named Essential Climate Variable Soil Moisture (ECV SM)] combines two different types of microwave Earth observation systems and six separate missions. It comprises data from four passive systems, including the Scanning Multichannel Microwave Radiometer (SMMR) onboard Nimbus-7 (1978–87), the Special Sensor Microwave Imager (SSM/I) of the Defense Meteorological Satellite Program (1987–2007), the Tropical Rainfall Measuring Mission Microwave Imager (TMI; since 1997), and the Advanced Microwave Scanning Radiometer for Earth Observing System (AMSR-E) onboard the Aqua satellite (since 2002). Data from active microwave sensors are from the scatterometers (SCAT) onboard the European Remote Sensing satellites (ERS-1 and ERS-2) since July 1991 and the Advanced Scatterometer (ASCAT) onboard the Meteorological Operation-A (MetOp-A) satellite that was launched in October 2006.

The SSM estimates from passive data were obtained with the Vrije Universiteit Amsterdam–National Aeronautics and Space Administration (VUA–NASA) algorithm, which is based on the Land Parameter Retrieval Model (LPRM; Owe et al. 2008). For the active sensors, SSM estimates were provided by the Vienna University of Technology (TU-Wien) change detection algorithm originally developed for SCAT onboard ERS (Wagner et al. 1999) and applied with minor adaptations to ASCAT (Naeimi et al. 2009). The combination of passive and active microwave satellite-based retrievals in the ECV SM product offers enhanced data coverage and density for SSM at the global scale (Liu et al. 2011, 2012). A global trend analysis was performed by Dorigo et al. (2012), showing consistency with trends in vegetation and precipitation data as well as with land surface models. (The ECV SM data are available from http://www.esa-soilmoisture-cci.org/.) Detailed overviews of the quality and error characteristics of the merged ECV SM product are given in Albergel et al. (2013), Loew et al. (2013), and Dorigo et al. (2013, manuscript submitted to Remote Sens. Environ.).

The ECV SM dataset consists of time series of daily data at a 0.25° grid over global landmasses, excluding grid cells containing sea, lakes, permanent snow–ice cover, and dense vegetation. For Australia, this includes about 11 000 grid cells, each spanning a daily time series of 32 yr. For this study, the SSM daily time series were resampled to seasonal means, as done for the climate indices.

The dataset suffers from some data gaps (Loew et al. 2013; Dorigo et al. 2013, manuscript submitted to Remote Sens. Environ.). There are gaps over vegetated areas in the north and east of Australia, especially for the period 1987–91. This is due to the short wavelength used by the SSM/I sensors, which limits the retrieval of soil moisture to areas with no or sparse vegetation. Power supply degradation affected the observations of SMMR in 1986 and caused further discontinuity in the data. A calibration issue disturbed observations of ERS-1 SCAT in 2001 and caused anomalies in the data as well. To avoid artifacts from anomalies or partial data, we disregarded all data for MAM–JJA 1986, DJF–SON 1991, and DJF 2001–MAM 2003 and obtain a sample size for each season of 28–30, totaling 116 with all seasons combined.

c. Complex empirical orthogonal function analysis

A CEOF analysis was applied to the ECV SM data in order to derive the dominant signals and patterns of its variability (called modes). With this method, the ECV SM data, considered as a statistical field, were decomposed into uncorrelated complex-valued vectors in time [complex principal components (CPC)] and space (CEOF). Specifically, we used the Hilbert empirical orthogonal functions analysis, introduced by Rasmusson et al. (1981), based on the relation of the data samples with their (shifted) Hilbert transforms. The methodology followed the generic recipe by Hannachi et al. (2007), with computational improvements by Wilks (2011). To deal with data gaps, the methods by von Storch and Zwiers (2002) were applied.

CEOF analysis was favored over noncomplex EOF analysis due to its ability to detect the propagation of spatial patterns, enabling the analysis of moving modes. The main advantage for our study is that it reveals (in addition to the variance characteristics) the change of the geophysical processes (von Storch and Zwiers 2002) and so provides quantitative understanding of the underlying oscillating mechanisms (Esquivel and Messina 2008). Comparison with the results of noncomplex EOF analyses suggested that the imaginary part of the CEOFs reflects the static nonshifted patterns and therefore can be used for correlation analyses concerning climate modes.

Only grid points with a time series sample size greater than an empirically determined threshold of 86 out of 116 samples were considered valid. This was done to achieve a workable compromise between spatial and temporal integrity.

First, the analysis was carried out for ECV SM data from all the seasons combined (but partially truncated for 1986, 1991, and 2001–03). Hendon et al. (2007), Nicholls (2009), Liu et al. (2009), and others found season-dependent impacts of climate oscillations on Australia's hydrology, and therefore a separate analysis was also performed, examining time series for individual seasons only. This procedure reduced the maximum sample size at each location from 116 to 28, 29, and 30 for MAM, DJF and JJA, and SON, respectively. Again, periods with insufficient data are disregarded and the minimum sample size per location was set at 20. To improve readability, the resulting (arbitrary scaled) values of the spatial CEOF amplitudes were normalized by their individual maximum and the temporal CPC amplitudes were normalized by their individual standard deviation.

d. Correlation analysis

To quantify the strength of the relationship between the SSM modes and climate modes, the Spearman rank correlation coefficients between the conjugate imaginary parts of the all-season-combined and individual season CPC time series and the climate indices (SOI, DMI, and SAMI) were calculated. We chose the nonparametric Spearman correlation because it does not require assumptions about the probability distributions of the samples. Rebel et al. (2012) found that the temporal autocorrelation of SSM, estimated from remote sensing, in situ, and modeling, is certainly not more than three months. However, we applied formulas from Wilks (2011) to account for any autocorrelation and to determine the effective sample size of CPC and the climate index time series.

3. Results and discussion

a. Spatiotemporal CEOF analysis of ECV SM data

The first mode of the CEOF decomposition using data from all seasons represents 46.3% of the total variance in SSM from 1979 to 2010 (Fig. 1). The spatial amplitude (CEOF 1 amplitude; Fig. 1a) shows the highest magnitude in the north and on the southeastern and southwestern edges of the continent. These locations coincide with Australia's humid regions, excluding the east coast.

Fig. 1.
Fig. 1.

First CEOF mode that explains 46.3% of the total variance in the SSM field in Australia. (a) The spatial amplitude and (b) temporal amplitude. The mode's (c) spatial phase and (d) temporal phase. Dots indicate phase values at DJF for better readability.

Citation: Journal of Climate 26, 24; 10.1175/JCLI-D-13-00149.1

The spatial phase (CEOF 1 phase; Fig. 1c) together with the temporal phase (CPC 1 phase; Fig. 1d) illustrate the propagating nature of the pattern: it is a clear annual signal characterized by one phase revolution per year (Fig. 1c, with 0° = 360°). This is related to the spatial phase, which shows a propagation from north to south during one cycle (365 days ≈ 360°): at the beginning of the year (15 January) the phase is zero degrees, hence maximum amplitude is reached in the north around January–February (i.e., 30°) and in the south around July–August (i.e., 225°). The signal remains stable throughout all 32 yr. The irregular peaks in the phase graph with 360° stem from small shifts in the annual cycle (i.e., earlier onsets of wet or dry periods). It also becomes apparent that the data truncation in 1986, 1991, and 2001–03 does not degrade the analysis, even though particular seasons are excluded.

The findings for mode 1 reflect the general seasonal rainfall pattern. First, the north receives predominately monsoonal precipitation during the southern summer season (December–March). Second, the south receives the most rainfall during the southern winter season (June–October). Third, low amplitudes on the actual humid central east coast reflect the lack of a distinct rainfall seasonality, and a seasonal SSM signal is not necessarily expected here. Finally, the continent's interior is arid without clear seasonality and consequently features no relevant amplitude at all. The first principal component, representing the temporal amplitude of the first CEOF mode (CPC 1 amplitude; Fig. 1b), reflects the magnitude of the mode's SSM signal over time. It peaks regularly during summer, which coincides with the monsoon season that predominantly impacts the northern regions (CEOF 1 has the highest amplitudes there), where the spatial phase already highlighted the maximum amplitude during January–February. The varying maximum value can mainly be linked to the variation in monsoon strength from year to year. For example, strong monsoons were recorded in (January) 1991, 1997, 1999–2002, 2004, 2008, and 2009 and weak monsoons were recorded in 1983, 1988, 1990, 1998, 2005, and 2007 as reported in Kajikawa et al. (2009). The comparison of the amplitudes of summer maxima in CPC 1 with these strong and weak monsoons, respectively, shows a high level of agreement, especially in the period from 1998 to 2005. This does not apply for the 1983 and 2010 maxima, but the latter can be explained by an exceptional La Niña event. Minima are experienced in the middle of the year and reflect the overall low average precipitation across Australia during winter. The relatively low values in 2003, 2005, 2007, and 2008 agree with the worst years of the Millennium Drought as reported by van Dijk et al. (2013). Overall, the patterns of mode 1 agree very well with the results of García-García et al. (2011), who applied a CEOF decomposition to water mass variations of Australia derived from GRACE gravity anomaly data. However, our study investigated a much longer period and thus extends this previous research, demonstrating the usefulness of the ECV SM data.

The second mode (CEOF 2; Fig. 2) represents 14.5% of the total variance and has a rather uniform continentwide pattern (Fig. 2a). The CEOF 2 phase (Fig. 2c) features only a small range between 315° and 30° and thus suggests that mode 2 is a rather static and nonoscillatory SSM variation component. The CPC 2 amplitude (Fig. 2b) and phase (Fig. 2d) are not easy to interpret with ambiguous cycle lengths, giving this mode a random character. Obviously, the CEOF method fails to properly resolve the temporal amplitude and phase of nonoscillatory patterns, as it induces arbitrary Hilbert transforms. Nonetheless, the variance represented by this mode could be considered the base signal of Australian inland SSM. Its varying amplitude may be linked with the inland penetration of the northern monsoon and tropical cyclone storms, accounting for the maxima of CPC 2 appearing often (but irregularly) early in the year.

Fig. 2.
Fig. 2.

As in Fig. 1, but for the second CEOF mode that explains 14.5% of the total variance.

Citation: Journal of Climate 26, 24; 10.1175/JCLI-D-13-00149.1

By contrast, CEOF mode 3, containing 6.3% of variance, reveals a clear multiannual east–west contrasting pattern (Fig. 3c). The CEOF 3 amplitude (Fig. 3a) highlights two regions, one in the northwest toward the Indian Ocean and the other in the northeast toward the Pacific Ocean. The CEOF 3 phase plot (Fig. 3c) indicates that the occurrence of the CEOF 3's maxima alternate as a dipole oscillation from west to east. From the CPC 3 phase plot (Fig. 3d), an irregular period length of 2–5 yr can be inferred. This evidence suggests a connection with ENSO and IOD: the east–west character meets typical ENSO patterns, and the CPC extremes coincide with ENSO extremes (with minima in 1987 and 1997 coinciding with El Niños and maxima in 1998, 2006, 2008, and 2010 with La Niñas; Fig. 3b). The detected multiyear cycle is also in accordance with the nature of ENSO, for which de Viron et al. (2013) found quasi-biennial and quasi-quadrennial periods. Abundant rainfall during southern summer 2009/10 and 2010/11 has been related to La Niña (Blunden and Arndt 2012) and caused record-breaking floods in Queensland (QLD) and New South Wales (NSW). Mode 3 accounts for these wet events: CEOF 3 features high amplitude in the inland areas of NSW and at the greater coastal area of QLD, while CPC 3 shows maximum values in summer 2009/10 and 2010/11.

Fig. 3.
Fig. 3.

As in Fig. 1, but for the third CEOF mode that explains 6.3% of the total variance.

Citation: Journal of Climate 26, 24; 10.1175/JCLI-D-13-00149.1

The fourth mode, explaining 4.2% of variance, shows similar CPC graphs in amplitude (Fig. 4b) and in phase (Fig. 4d) as mode 3. However, its spatial pattern (Fig. 4a) is strongest in the north and in the northwest. From the spatial phase map (Fig. 4c) it appears that both these focus areas reach peak values consecutively with an about one-half cycle difference (φ ≈ 90° in the west and φ ≈ 225° in the north). As a note, phase maps are not meaningful in areas where the associated amplitude is close to zero. The period varies from 1 to 5 yr and does not show a stable behavior (Fig. 4b). Noticeable is the high number of peaks in the CPC 4 amplitude in the middle of the year that cannot be associated with the monsoonal rainfall regime.

Fig. 4.
Fig. 4.

As in Fig. 1, but for the fourth CEOF mode that explains 4.2% of the total variance.

Citation: Journal of Climate 26, 24; 10.1175/JCLI-D-13-00149.1

1) Data synthesis from principal CEOFs

Since a CEOF analysis is a decomposition of the original data field, the latter can be reconstructed by summing up the vector products of CEOF and CPC of the respective principal modes. The first four modes comprise 71.3% variance. A synthesis using these four modes gives a rebuild of the initial dataset such that, as an example, the correlation maps of SOI with the fields (initial and CEOF synthesized) match satisfactorily (Fig. 5). The initial dataset is built from 11 000 times 128 samples, while the synthesized dataset is built from 4 times 116 samples. This demonstrates the efficiency of the CEOF method and its skill to successfully fill data gaps. A compression of the huge original data field into a smart set of CEOF and CPC functions preserves most of the total variance by taking advantage of the redundancy from spatial and temporal dependencies.

Fig. 5.
Fig. 5.

Spearman correlation coefficient of SOI with seasonal SSM means from (a) the original data time series and from (b) the synthesized time series from the first four CEOF modes.

Citation: Journal of Climate 26, 24; 10.1175/JCLI-D-13-00149.1

The remaining differences between Figs. 5a and 5b cannot be used to attribute the remaining variance (28.7%). Any EOF or CEOF analysis cannot resolve meaningful patterns for modes of lower variability (Wilks 2011). This limitation is a consequence of the method's mathematical design, which demands mutual orthogonality between EOFs. Hence, the distinction between measurement noise, mathematical artifacts, and minor signals is not possible for lower-ranked modes. Nonetheless, the results obtained with the synthesized data provide more confidence in the empirically found threshold for the number of values required; here, 86 out of 116 were sufficient. The comparison with results from analyses using a higher or lower threshold showed either less coverage or more deviations but no improvements when reconstructing the original data.

2) CEOF robustness

The CEOF synthesis approach can also be used to test the robustness of the calculated CEOF and CPC patterns. In this application, the ECV SM dataset is repeatedly synthesized through the summation of all modes but each time skipping one leading mode more. The datasets produced this way are then decomposed again into new CEOFs, and the results are compared to the initial CEOFs. Modes that are not reproduced by this procedure are considered degenerated, since their patterns appear to be dependent on the method instead on the data. For the above modes 2, 3, and 4 (mode 1, being the leading mode, obviously cannot be compared), all CEOF spatial and temporal patterns are also obtained through this procedure, showing only small deviations. Of course, the new modes' share in total variation increases proportionally to the sum of previously skipped variance. Moreover, a masking over 51 grid points, visible as distinctive spots in the CEOF 1 phase map (Fig. 1c), was necessary since these outliers disturb this procedure.

The above assessment receives support from conventional tests for EOF robustness, which provide criteria for the reliability of the obtained functions. The Kaiser's rule, as in Wilks (2011), suggests retaining the first 25 modes. Rule N, following Overland and Preisendorfer (1982)'s work in contrasting the modes' eigenvalues with eigenvalues from randomized (noiselike) data, accepts the above modes as well. The rule of thumb by North et al. (1982) suggests retaining the first five modes but classifies lower-ranked modes as degenerate.

Furthermore, the stability of CEOFs as a function of the analysis period was examined. To this end, several CEOF analyses of individual ECV SM subsamples were carried out and evaluated against the reference CEOFs of the complete 1979–2010 time series. The 30 subsamples were created, each starting 1 yr later from 1980 on and all ending in 2010, accounting for the gradually increasing quality of the remotely sensed observations over the years. For every subsample, the CEOF spatial amplitude value was compared to the reference at each grid point. Figure 6 shows the mean of these differences over all grid points per CEOF modes 1–4. It becomes apparent that subsamples containing time series spanning more than 16 yr (starting before 1994) feature only little deviation in the mean values below 10% for each mode, relative to the maximum amplitude of the four individual reference CEOFs. Further reduction of the subsample period introduces considerable deviations to all modes, with less impact on the seasonal signal (CEOF 1). Deviations of CEOFs 3 and 4 increase considerably after 1993, which is plausible since they represent multiannual signals and thus demand a longer observation period. Therefore, for analyses such as those carried out here, a lower limit for the investigated time series length of about 10–15 yr would seem appropriate. Interestingly, CEOF 2 has the second lowest mean difference for the longest subsamples, but when reducing sample sizes, it soon exceeds the lower-ranked CEOFs 3 and 4. This confirms that CEOF 2 is not captured well by complex EOF analysis because it reflects a quasi-static pattern.

Fig. 6.
Fig. 6.

Mean difference of spatial amplitude values of CEOF modes 1–4 to reference modes for all subsamples, indicated by the starting year. The scale is relative to the maximum amplitude of the four reference modes.

Citation: Journal of Climate 26, 24; 10.1175/JCLI-D-13-00149.1

b. Impact of climate modes on all-season ECV SM modes

A correlation analysis was performed to assess the overall impact of the ocean oscillations on Australia's SSM signals. Spearman correlation coefficients were calculated for the four obtained CEOFs as well as noncomplex EOFs, on the one hand, and the ENSO, IOD, and SAM indices, on the other hand (Table 1). The table lists the correlation of climate indices with the imaginary part of the complex conjugate CPCs for the CEOF case and with the PCs in the EOF case. It is apparent that the CEOF modes better capture the impact of climate modes than the EOF modes; the results from the CEOF analysis are better for both the measured correlation coefficient and correlation significance. The propagating nature of CEOF modes 1, 3, and 4 supports this conclusion.

Table 1.

Spearman correlation coefficients between climate indices SOI, SAMI, and DMI and PC time series of CEOF and EOF analysis. For the CEOF, the imaginary part of the (complex conjugate) CPCs is used. Double asterisks indicate correlation at high significance (p < 0.01), single asterisk indicate significance (p < 0.05). No asterisk means significance 0.05 < p < 0.10. Blanks indicate no or insignificant correlation. The mode's share in total SSM variance is also displayed.

Table 1.

For CEOF mode 1, which comprises almost half of the total variance and reflects the annual north–south shift of Australia's rainfall regime, a weak but highly significant connection (p < 0.01) to SAM was found. This suggests that the variation of the seasonal SSM signal is influenced by the state of SAM. This can be explained by the relation between SAM and the latitudinal position of the westerly wind belt that drives location and the strength of cold fronts and midlatitude storm systems (Ho et al. 2012). Indeed, the normalized imaginary and real part amplitudes of CEOF 1 depict a clear gradient from −1 in the north to +1 in the south (see the supplemental information). This supports the results of Meneghini et al. (2007), who proposed that SAM is directly related to rainfall variance in northern Australia and the inverse in southern Australia. Hendon et al. (2007) found a contrast in SSM influence between seasons: in fall and winter, positive SAM is associated with decreased rainfall in southeast and southwest of Australia. During the cool season, the low-pressure belt is usually located south of Australia and when it contracts toward the South Pole, rainfall tends to be reduced. By contrast, during spring/summer, positive SAM conditions are associated with increased rainfall in these regions. To eliminate the possibility of spurious correlation due to similar seasonal patterns in SAMI and SSM, the seasonality in CPC 1 was removed by subtracting the seasonal averages (the climatologies). The correlation between the CPC 1 anomalies with SAMI is then even stronger (ρ = −0.33; see plots in Fig. 7). Thus, SAM appears to be a driver of the variation of the seasonal SSM signal in Australia.

Fig. 7.
Fig. 7.

Plots of imaginary parts of CPC (a) 1, (b) 3, and (c) 4, together with relevant climate indices. In (c), the Indian Ocean DMI is inverted, accounting for its negative correlation with SSM.

Citation: Journal of Climate 26, 24; 10.1175/JCLI-D-13-00149.1

CEOF mode 2 (14.5% variance) appears to contain nonseasonal and rather static variations lacking a systematic pattern (based on Fig. 2). Sure enough, no significant connections between complex mode 2 and the climate modes were found. Comparison with the noncomplex results provokes the assumption that some SAM-related variance captured by CEOF 1 is incorporated in EOF 2 (which features a seasonal-like signal; not shown). That said, nonperiodic signals may not be disentangled properly from periodic signals by the regular EOF analysis. Indeed, EOF data synthesis excluding mode 2 and consecutive CEOF analysis produces a less smooth gradient for mode 1 in the spatial phase and a lower correlation with SAMI [reduced from ρ = −0.27 (p < 0.01) to −0.23 (p < 0.05)].

CEOF mode 3 captures a multiyear east–west signal and 6.3% of the total variance and correlates like mode 1 with SAMI. Thus, SAM seems to also contain a longitudinal element. High correlation was also found with SOI, confirming the findings of Chiew et al. (1998) and Liu et al. (2007) and further emphasizing the connection between ENSO and SSM in northeastern Australia (Figs. 3a,b and 5). The absence of any correlation with DMI is curious, as the spatial and temporal pattern of mode 3 would be expected to be at least partially IOD induced.

However, for CEOF mode 4 (4.2% of the total variance), a correlation with DMI of ρ = −0.30 is found. The maxima of CPC 4 appear midyear or later in the year (Fig. 4), suggesting this mode is a cool season component of the SSM signal. The imaginary part of CEOF 4 features strong negative amplitudes in the west of the continent, toward the Indian Ocean and positive amplitudes north of the center (not shown). ENSO is positively correlated with this mode and thus can be the reason for these positive amplitudes.

c. Impact of climate modes on seasonal ECV SM modes

The ECV SM dataset was also decomposed into CEOFs for each season (DJF, MAM, JJA, and SON) separately. The robustness of these seasonal CEOFs was determined using the conventional rules mentioned before. This suggested the following modes to be degenerated (only North's rule of thumb was critical): DJF modes 3 and 4, MAM modes 2 and 3, JJA mode 4, and SON mode 4 (failed the rule N test as well). However, the following modes were recovered by the synthesis reanalysis approach: all DJF modes except 4, all MAM modes, all JJA modes, and all SON modes except 4. Overall, DJF and SON mode 4 may be considered degenerated, and therefore all modes of rank lower than 3 were disregarded in further analyses.

Analogous to the analysis of all-season-combined ECV SM data, the correlations coefficient between imaginary parts of CPCs from seasonal data and seasonal climate indices were calculated. The results for the leading three CEOF modes are listed in Table 2. The ECV SM data were synthesized from these first three CEOF modes for each season, and the Spearman correlation with climate modes is mapped in Fig. 8. Even though they comprise only about 60% of the total variance each, the correlation plots in Fig. 8 agree qualitatively with those of the initial ECV SM data (not shown) satisfactorily, underlining the efficiency of the CEOF analysis.

Table 2.

Spearman correlation coefficients between climate indices SOI, SAMI, and DMI and imaginary part of (complex conjugate) CPC time series of seasonal CEOF analysis. Each season was decomposed separately. Notation as in Table 1.

Table 2.
Fig. 8.
Fig. 8.

Spearman correlation coefficient of (top)–(bottom) climate indices with SSM time series synthesized from the first three seasonal CEOF modes. White areas are masked out due to poor significance (p < 0.05).

Citation: Journal of Climate 26, 24; 10.1175/JCLI-D-13-00149.1

The connection between SSM and climate indices is season dependent, with correlation coefficients that vary strongly during the year (Table 2). It is noted that the sign of the coefficients also depends on the polarity of the imaginary part of the CEOF functions. Their (linked) spatial and temporal amplitudes are either positive or negative, and these signs are configured by higher-ranked CEOFs. In Fig. 8, the season dependency is even more apparent and the changing locations of influence of the climate modes illustrate the dynamic character of the interactions. Also evident from Fig. 8 is that SOI and SAMI have similar behavior during summer and fall but not during winter and spring. The Spearman correlation between the indices confirms this [DJF ρ = +0.50 (p < 1%); MAM ρ = +0.45 (p < 1%); JJA ρ = −0.07 (not significant); SON ρ = 0.21 (not significant)]. The IOD does not seem to share impact patterns with ENSO or SAM.

For ENSO, strong connections were found in summer [DJF mode 1 (31.7%; ρ = +0.36)] and spring [SON mode 2 (18.7%; ρ = +0.37) and mode 3 (10.2%; ρ = −0.63)]. To a lesser extent, ENSO impact was found for fall [MAM mode 2 (11.1%; ρ = +0.44)] and winter [JJA mode 3 (10.8%; ρ = +0.52)].

SAM could be identified as a notable SSM driver in summer [DJF mode 1 (31.7%; ρ = +0.57)]. The positive SAM impact in large parts in northern Australia during this period can be seen in Fig. 8. A minor contribution to SSM variance was found for fall [MAM mode 2 (11.1%; ρ = +0.42)] in the north and off-coast SEA. No significant contribution was found for the other seasons. This agrees with the all-season CPC 1 (Fig. 1), which has maximum amplitude at the end and start of the year and was already found to be related to SAM.

For IOD, a strong connection to SSM was found for spring [SON mode 1 (33.8%; ρ = −0.36)]. A smaller share of total SSM variation is explained by IOD in winter [JJA mode 2 (18.0%; ρ = −0.41)]. This agrees with the interpretation of the all-season CEOF mode 4, which was related to IOD and peaks during the cool season. Looking at patterns in Fig. 8, IOD appears to drive SSM variances in the southeast and south-central parts of Australia during spring and in the west during winter.

Our study confirms the findings of Liu et al. (2007), who related ENSO and IOD to remotely sensed Australian SSM for 1998–2005. However, we also found strong correlation between SSM and SOI for summer and IOD in winter. The much longer observation period can explain this difference. Liu et al. (2009) analyzed data for the period 1979–2006 and (contrary to our findings) they localized ENSO impact during summer mainly in the west. Nicholls (2009) asserted that the tropical climate modes ENSO and IOD, contrary to the extratropical SAM, did not explain southern Australian rainfall deficits during fall and winter in the period of 1958–2007. However, our results do not show a relation of SAM to Australian SSM during winter and only some relation during fall. By contrast, ENSO shows a winter influence in SEA and IOD in the southwest. Interestingly, CEOF summer mode 1 and fall mode 2 show similar patterns and feature high positive amplitude in northern Australia as well as in SEA (recognizable also in the impact maps in Fig. 8). Both modes correlate significantly with SOI and SAMI, suggesting that ENSO and SAM are connected to SSM in northern and southeastern Australia during summer and fall. The fall influence of ENSO is not expected according to Timbal and Hendon (2011), who noted that the Millennium Drought was mainly a consequence of SEA's rainfall deficits during fall, driven by extratropical rather than by tropical processes. This may be due to antecedent rainfall deficits that integrate to dryer fall SSM conditions, but nevertheless ENSO is diagnosed to have a stronger impact than SAM in that area. The results of Ummenhofer et al. (2009), who propose IOD as the main driver of the Big Dry, are also not supported by the results of our study, as IOD appears to impact SEA in spring only. It should be noted, however, that results from hydrological studies on anomalies or extreme events may not necessarily be comparable with those of studies on long-term variations. For example, van Dijk et al. (2013) found differences between the long-term rainfall variance explained by climate modes and their predicted contributions to rainfall anomalies during the Millennium Drought.

4. Conclusions

A new dataset of 32 yr of remotely sensed surface soil moisture (SSM) was examined for mainland Australia and related to oceanic climate oscillation indices. We extended previous research on Australia's hydrology to a longer observation period of over three decades and also addressed earlier research on the interaction of climate modes with hydrological conditions, represented by SSM observations. Our results demonstrated the merit of a multidecadal dataset such as the ECV SM.

The CEOF analysis shed light on the nature of SSM conditions over Australia and our findings are in strong accordance with climatic understanding and previous studies (e.g., García-García et al. 2011). The CEOF technique's ability to extract moving oscillatory signals was demonstrated and thus made it possible to determine not only where but also when variations occur. Furthermore, the capability of the CEOF technique to compress multidimensional geophysical data in a meaningful and efficient manner is highlighted. Only four leading modes comprised almost the complete variation with physical relevance and allowed proper spatiotemporal analysis. The CEOF analysis enabled the extraction of both seasonal and multiannual signals. The seasonal signal (CEOF mode 1) explained about half of all SSM variation, peaking in February–March in the north and in August–September in the south. A low but highly significant influence of SAM was found in the seasonal signal. Two multiannual signals explaining a lesser share of the total variance (6.3% and 4.2%) were identified: one (CEOF mode 3) had an unstable period of 2–5 yr and an east–west pattern that can be associated with ENSO and SAM but not with IOD. The second one (CEOF mode 4) had a 1- to 5-yr oscillation, formed a dipole pattern between the west and north, and was linked to ENSO and IOD. In addition, a similar analysis was performed for the individual seasons. Strong relationships were found between ENSO and SSM in northern and eastern Australia for southern spring and summer and to a lesser extent for fall. During winter, 10% of the total SSM variation in SEA can be linked to ENSO. SAM shows no clear impact during winter or spring but does during summer and fall in northern and eastern Australia, in similar ways as ENSO. In spring, IOD is clearly impacting SSM variations in the entire east, southeast, and center. After spring, IOD disappears as a SSM driver, recurring in winter in the west and south.

The detected influence of the tropical modes ENSO and IOD on SSM variation in the north and east during spring was already clear from previous research. However, except for spring, no connection of IOD to the hydrological signal over SEA was found. Instead, IOD only seems to also drive a small share of the variance in the north and west, with varying intensity, following a multiannual cycle. By contrast, ENSO contributes to the northern and eastern SSM signal throughout the year (but in winter only in SEA), also following a multiannual cycle. This persistence has not been found so clearly in the existing literature. SAM, however, shows a relation to the seasonal SSM signal of all of Australia, which comprises half of the total variation. Thus, SAM can be suggested as a key driver of the continent's seasonal soil moisture variation. Furthermore, its influence in the north is found to be higher than expected.

When applied in hydroclimatological studies, CEOF analysis appears to be a powerful technique to identify climate mode–induced signals. This could be useful for long-term trend analyses of, for example, regional or global precipitation, evaporation, or SSM (as in Dorigo et al. 2012), as multiyear phenomena like ENSO, IOD, or SAM can significantly change the magnitude and even the sign of simple trends in the data and therefore may need to be removed and analyzed separately.

Climate indices such as SOI, DMI, or SAMI are simplified but representative measures of oceanic circulations. Predicting oceanic oscillation patterns may be possible one to two seasons in advance: there are already operational forecasts for ENSO (e.g., by BOM) and successful studies for IOD (e.g., Luo et al. 2008). With further improvement of such climate mode prediction capabilities, the seasonal forecast of SSM and the hydrology of Australia in general should be improved. Hudson et al. (2011) already highlighted that predicting ENSO, IOD, and SAM enhances intraseasonal forecast skills for winter and spring rainfall in Australia. Exploiting the knowledge of correlations between climate modes and hydrological patterns provides a promising potential to increase seasonal hydrological forecasts, especially in Australia, which is so strongly affected by ocean circulations.

Acknowledgments

This study was performed in the framework of the STSE Water Cycle Multimission Observation Strategy (WACMOS) project funded by the European Space Agency (Contract 22086/08/1-EC) and ESA's Climate Change Initiative (CCI) for Soil Moisture (Contract 4000104814/11/I-NB).

REFERENCES

  • Albergel, C., and Coauthors, 2013: Skill and global trend analysis of soil moisture from reanalyses and microwave remote sensing. J. Hydrometeor., 14, 12591277.

    • Search Google Scholar
    • Export Citation
  • Blunden, J., , and D. S. Arndt, 2012: State of the climate in 2011. Bull. Amer. Meteor. Soc., 93, S1S282.

  • Chiew, F., , T. Piechota, , J. Dracup, , and T. McMahon, 1998: El Nino/Southern Oscillation and Australian rainfall, streamflow and drought: Links and potential for forecasting. J. Hydrol., 204, 138149, doi:10.1016/S0022-1694(97)00121-2.

    • Search Google Scholar
    • Export Citation
  • Dai, A., 2013: Increasing drought under global warming in observations and models. Nat. Climate Change, 3, 5258, doi:10.1038/nclimate1633.

    • Search Google Scholar
    • Export Citation
  • de Jeu, R. A. M., , W. Wagner, , T. R. H. Holmes, , J. Dolman, , N. C. Giesen, , and J. Friesen, 2008: Global soil moisture patterns observed by space borne microwave radiometers and scatterometers. Surv. Geophys., 29, 399420, doi:10.1007/s10712-008-9044-0.

    • Search Google Scholar
    • Export Citation
  • de Viron, O., , J. O. Dickey, , and M. Ghil, 2013: Global modes of climate variability. Geophys. Res. Lett., 40, 18321837, doi:10.1002/grl.50386.

    • Search Google Scholar
    • Export Citation
  • Dorigo, W. A., , R. A. M. de Jeu, , D. Chung, , R. Parinussa, , Y. Y. Liu, , W. Wagner, , and D. Fernndez-Prieto, 2012: Evaluating global trends (1988–2010) in harmonized multi-satellite surface soil moisture. Geophys. Res. Lett.,39, L18405, doi:10.1029/2012GL052988.

  • Esquivel, P., , and A. R. Messina, 2008: Complex empirical orthogonal function analysis of wide-area system dynamics. Proc. 2008 IEEE Power and Energy Society General Meeting—Conversion and Delivery of Electrical Energy in the 21st Century, Pittsburgh, PA, IEEE, 1–7.

  • Field, C. B., , V. Barros, , T. F. Stocker, , and Q. Dahe, Eds., 2012: Managing the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation. Cambridge University Press, 582 pp.

  • Forootan, E., , J. Awange, , J. Kusche, , B. Heck, , and A. Eicker, 2012: Independent patterns of water mass anomalies over Australia from satellite data and models. Remote Sens. Environ., 124, 427443, doi:10.1016/j.rse.2012.05.023.

    • Search Google Scholar
    • Export Citation
  • García-García, D., , C. C. Ummenhofer, , and V. Zlotnicki, 2011: Australian water mass variations from GRACE data linked to Indo-Pacific climate variability. Remote Sens. Environ., 115, 21752183, doi:10.1016/j.rse.2011.04.007.

    • Search Google Scholar
    • Export Citation
  • Gerten, D., , S. Schaphoff, , and W. Lucht, 2007: Potential future changes in water limitations of the terrestrial biosphere. Climatic Change, 80, 277299.

    • Search Google Scholar
    • Export Citation
  • Gong, D., , and S. Wang, 1999: Definition of Antarctic Oscillation index. Geophys. Res. Lett., 26 (4), 459462.

  • Hannachi, A., , I. T. Jolliffe, , and D. B. Stephenson, 2007: Empirical orthogonal functions and related techniques in atmospheric science: A review. Int. J. Climatol., 1152, 11191152, doi:10.1002/joc.1499.

    • Search Google Scholar
    • Export Citation
  • Hendon, H. H., , D. W. J. Thompson, , and M. C. Wheeler, 2007: Australian rainfall and surface temperature variations associated with the Southern Hemisphere annular mode. J. Climate, 20, 24522467.

    • Search Google Scholar
    • Export Citation
  • Ho, M., , A. S. Kiem, , and D. C. Verdon-Kidd, 2012: The southern annular mode: A comparison of indices. Hydrol. Earth Syst. Sci., 16, 967982, doi:10.5194/hess-16-967-2012.

    • Search Google Scholar
    • Export Citation
  • Hudson, D., , O. Alves, , H. H. Hendon, , and A. G. Marshall, 2011: Bridging the gap between weather and seasonal forecasting: Intraseasonal forecasting for Australia. Quart. J. Roy. Meteor. Soc., 137, 673689, doi:10.1002/qj.769.

    • Search Google Scholar
    • Export Citation
  • Huntington, T. G., 2006: Evidence for intensification of the global water cycle: Review and synthesis. J. Hydrol., 319, 8395, doi:10.1016/j.jhydrol.2005.07.003.

    • Search Google Scholar
    • Export Citation
  • Kajikawa, Y., , B. Wang, , and J. Yang, 2009: A multi-time scale Australian monsoon index. Int. J. Climatol., 30, 11141120, doi:10.1002/joc.1955.

    • Search Google Scholar
    • Export Citation
  • Koster, R. D., and Coauthors, 2004: Regions of strong coupling between soil moisture and precipitation. Science, 305, 11381140, doi:10.1126/science.1100217.

    • Search Google Scholar
    • Export Citation
  • Legates, D. R., , R. Mahmood, , D. F. Levia, , T. L. DeLiberty, , S. M. Quiring, , C. Houser, , and F. E. Nelson, 2010: Soil moisture: A central and unifying theme in physical geography. Prog. Phys. Geogr., 35, 6586, doi:10.1177/0309133310386514.

    • Search Google Scholar
    • Export Citation
  • Liu, Y. Y., , R. A. M. de Jeu, , A. I. J. M. van Dijk, , and M. Owe, 2007: TRMM-TMI satellite observed soil moisture and vegetation density (1998–2005) show strong connection with El Niño in eastern Australia. Geophys. Res. Lett., 34, L15401, doi:10.1029/2007GL030311.

    • Search Google Scholar
    • Export Citation
  • Liu, Y. Y., , A. I. J. M. van Dijk, , R. A. M. de Jeu, , and T. R. H. Holmes, 2009: An analysis of spatiotemporal variations of soil and vegetation moisture from a 29-year satellite-derived data set over mainland Australia. Water Resour. Res., 45, W07405, doi:10.1029/2008WR007187.

    • Search Google Scholar
    • Export Citation
  • Liu, Y. Y., , R. M. Parinussa, , W. A. Dorigo, , R. A. M. de Jeu, , W. Wagner, , A. I. J. M. van Dijk, , M. F. McCabe, , and J. P. Evans, 2011: Developing an improved soil moisture dataset by blending passive and active microwave satellite-based retrievals. Hydrol. Earth Syst. Sci., 15, 425436, doi:10.5194/hess-15-425-2011.

    • Search Google Scholar
    • Export Citation
  • Liu, Y. Y., , W. Dorigo, , R. Parinussa, , R. A. M. de Jeu, , W. Wagner, , M. McCabe, , J. Evans, , and A. I. J. M. van Dijk, 2012: Trend-preserving blending of passive and active microwave soil moisture retrievals. Remote Sens. Environ., 123, 280297, doi:10.1016/j.rse.2012.03.014.

    • Search Google Scholar
    • Export Citation
  • Loew, A., , T. Stacke, , W. A. Dorigo, , R. A. M. de Jeu, , and S. Hagemann, 2013: Potential and limitations of multidecadal satellite soil moisture observations for climate model evaluation studies. Hydrol. Earth Syst. Sci. Discuss., 10, 35413594, doi:10.5194/hessd-10-3541-2013.

    • Search Google Scholar
    • Export Citation
  • Luo, J.-J., , S. Behera, , Y. Masumoto, , H. Sakuma, , and T. Yamagata, 2008: Successful prediction of the consecutive IOD in 2006 and 2007. Geophys. Res. Lett., 35, L14S02, doi:10.1029/2007GL032793.

    • Search Google Scholar
    • Export Citation
  • Meneghini, B., , I. Simmonds, , and I. N. Smith, 2007: Association between Australian rainfall and the southern annular mode. Int. J. Climatol., 27, 109121, doi:10.1002/joc.1370.

    • Search Google Scholar
    • Export Citation
  • Naeimi, V., , Z. Bartalis, , and W. Wagner, 2009: ASCAT soil moisture: An assessment of the data quality and consistency with the ERS scatterometer heritage. J. Hydrometeor., 10, 555563.

    • Search Google Scholar
    • Export Citation
  • Nan, S., , and J. Li, 2003: The relationship between the summer precipitation in the Yangtze River valley and the boreal spring Southern Hemisphere annular mode. Geophys. Res. Lett.,30, 2266, doi:10.1029/2003GL018381.

  • Nicholls, N., 1989: Sea surface temperatures and Australian winter rainfall. J. Climate, 2, 965973.

  • Nicholls, N., 2009: Local and remote causes of the southern Australian autumn-winter rainfall decline, 1958–2007. Climate Dyn., 34, 835845, doi:10.1007/s00382-009-0527-6.

    • Search Google Scholar
    • Export Citation
  • North, G. R., , T. L. Bell, , R. F. Cahalan, , and F. J. Moeng, 1982: Sampling errors in the estimation of empirical orthogonal functions. Mon. Wea. Rev., 110, 699706.

    • Search Google Scholar
    • Export Citation
  • Overland, J. E., , and R. W. Preisendorfer, 1982: A significance test for principal components applied to a cyclone climatology. Mon. Wea. Rev., 110, 14.

    • Search Google Scholar
    • Export Citation
  • Owe, M., , R. A. M. de Jeu, , and T. Holmes, 2008: Multisensor historical climatology of satellite-derived global land surface moisture. J. Geophys. Res., 113, F01002, doi:10.1029/2007JF000769.

    • Search Google Scholar
    • Export Citation
  • Rasmusson, E. M., , P. A. Arkin, , W.-Y. Chen, , and J. B. Jalickee, 1981: Biennial variations in surface temperature over the United States as revealed by singular decomposition. Mon. Wea. Rev., 109, 587598.

    • Search Google Scholar
    • Export Citation
  • Rebel, K. T., , R. A. M. de Jeu, , P. Ciais, , N. Viovy, , S. L. Piao, , G. Kiely, , and A. J. Dolman, 2012: A global analysis of soil moisture derived from satellite observations and a land surface model. Hydrol. Earth Syst. Sci., 16, 833847, doi:10.5194/hess-16-833-2012.

    • Search Google Scholar
    • Export Citation
  • Risbey, J. S., , M. J. Pook, , P. C. McIntosh, , M. C. Wheeler, , and H. H. Hendon, 2009: On the remote drivers of rainfall variability in Australia. Mon. Wea. Rev., 137, 32333253.

    • Search Google Scholar
    • Export Citation
  • Saji, N. H., , and T. Yamagata, 2003: Possible impacts of Indian Ocean dipole mode events on global climate. Climate Res., 25, 151169, doi:10.3354/cr025151.

    • Search Google Scholar
    • Export Citation
  • Saji, N. H., , B. N. Goswami, , P. N. Vinayachandran, , and T. Yamagata, 1999: A dipole mode in the tropical Indian Ocean. Nature, 401, 360363, doi:10.1038/43854.

    • Search Google Scholar
    • Export Citation
  • Simmonds, I., 1990: A modelling study of winter circulation and precipitation anomalies associated with Australian region ocean temperatures. Aust. Meteor. Mag., 38, 151161.

    • Search Google Scholar
    • Export Citation
  • Simmonds, I., , and A. Rocha, 1991: The association of Australian winter climate with ocean temperatures to the west. J. Climate, 4, 11471161.

    • Search Google Scholar
    • Export Citation
  • Simmonds, I., , and A. H. Lynch, 1992: The influence of pre-existing soil moisture content on Australian winter climate. Int. J. Climatol., 12, 3354, doi:10.1002/joc.3370120105.

    • Search Google Scholar
    • Export Citation
  • Simmonds, I., , and P. Hope, 1998: Seasonal and regional responses to changes in Australian soil moisture conditions. Int. J. Climatol., 18, 11051139, doi:10.1002/(SICI)1097-0088(199808)18:10<1105::AID-JOC308>3.0.CO;2-G.

    • Search Google Scholar
    • Export Citation
  • Solomon, S., , D. Qin, , M. Manning, , M. Marquis, , K. Averyt, , M. B. Tignor, , H. L. Miller Jr., , and Z. Chen, Eds., 2007: Climate Change 2007: The Physical Science Basis. Cambridge University Press, 996 pp.

  • Taylor, C. M., , R. A. M. de Jeu, , F. Guichard, , P. P. Harris, , and W. A. Dorigo, 2012: Afternoon rain more likely over drier soils. Nature, 489, 423426, doi:10.1038/nature11377.

    • Search Google Scholar
    • Export Citation
  • Timbal, B., , and H. Hendon, 2011: The role of tropical modes of variability in recent rainfall deficits across the Murray-Darling Basin. Water Resour. Res., 47, W00G09, doi:10.1029/2010WR009834.

    • Search Google Scholar
    • Export Citation
  • Trenberth, K., 2011: Changes in precipitation with climate change. Climate Res., 47, 123138, doi:10.3354/cr00953.

  • Troup, A. J., 1965: The ‘Southern Oscillation.' Quart. J. Roy. Meteor. Soc., 91, 490506, doi:10.1002/qj.49709139009.

  • Ummenhofer, C. C., , M. H. England, , P. C. McIntosh, , G. A. Meyers, , M. J. Pook, , J. S. Risbey, , A. Sen Gupta, , and A. S. Taschetto, 2009: What causes southeast Australia's worst droughts? Geophys. Res. Lett., 36, L04706, doi:10.1029/2008GL036801.

    • Search Google Scholar
    • Export Citation
  • van Dijk, A. I. J. M., , H. E. Beck, , R. S. Crosbie, , R. A. M. de Jeu, , Y. Y. Liu, , G. M. Podger, , B. Timbal, , and N. R. Viney, 2013: The Millennium Drought in southeast Australia (2001–2009): Natural and human causes and implications for water resources, ecosystems, economy, and society. Water Resour. Res., 49, 10401057, doi:10.1002/wrcr.20123.

    • Search Google Scholar
    • Export Citation
  • Verdon-Kidd, D. C., , and A. S. Kiem, 2009: On the relationship between large-scale climate modes and regional synoptic patterns that drive Victorian rainfall. Hydrol. Earth Syst. Sci., 13, 467479, doi:10.5194/hess-13-467-2009.

    • Search Google Scholar
    • Export Citation
  • von Storch, H., , and F. Zwiers, 2002: Statistical Analysis in Climate Research. Cambridge University Press, 496 pp.

  • Wagner, W., , J. Noll, , M. Borgeaud, , and H. Rott, 1999: Monitoring soil moisture over the Canadian Prairies with the ERS scatterometer. IEEE Trans. Geosci. Remote Sens., 37, 206216, doi:10.1109/36.739155.

    • Search Google Scholar
    • Export Citation
  • Wagner, W., , W. A. Dorigo, , R. A. M. de Jeu, , D. Fernandez, , J. Benveniste, , E. Haas, , and M. Ertl, 2012: Fusion of active and passive microwave observations to create an essential climate variable data record on soil moisture. ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol. I-7, 315–321.

  • Wilks, D., 2011: Statistical Methods in the Atmospheric Sciences. 3rd ed. Academic Press, 704 pp.

  • Yuan, Y., , and C. Li, 2008: Decadal variability of the IOD-ENSO relationship. Chin. Sci. Bull., 53, 17451752, doi:10.1007/s11434-008-0196-6.

    • Search Google Scholar
    • Export Citation

Supplementary Materials

Save