Research article
An entropy-based analysis method of precipitation isotopes revealing main moisture transport corridors globally

https://doi.org/10.1016/j.gloplacha.2020.103134Get rights and content

Highlights

  • The entropy concept is introduced to reanalyze the global precipitation isotope data.

  • Spatial distributions of Eδ18O and EδD reveal oceanic sources of water vapor globally.

  • 18O and EδD patterns reveal global moisture corridors from oceans to continents.

  • 18O and EδD patterns reflect seasonal variations of global moisture corridors.

Abstract

The hydrogen (δD) and oxygen (δ18O) isotopic compositions in the water molecule have been widely used as tracers for studying the global water cycle. In 1961, the Global Network of Isotopes in Precipitation (GNIP) was established to measure D and 18O isotopes contents in precipitation around the world. However, on the spatial scale, the long-term arithmetic and/or long-term precipitation weighted mean δD and δ18O have been most commonly used to interpret the GNIP isotope data for over sixty years. The spatial distributions of mean δD and δ18O depict well the regional moisture transport, but they vary predominantly with latitude on the global scale, especially over continental areas, obscuring the continental and circulation effects. We developed a new method of using the entropy concept to reanalyze precipitation isotopic compositions data from GNIP. Calculated entropies of isotopic compositions in precipitation at GNIP stations around the world strongly correlate in a linear fashion with a slope coefficient close to unity. The spatial distributions of both isotopic compositions entropies generally reveal oceanic sources of water vapor and main moisture transport pathways from oceans to continents globally, with different patterns between summer and winter seasons. Although these results have mostly been reported in previous studies, they provide the verification of this new analysis method. The entropy method proposed here is expected to provide a new tool for data interpretation of water isotopic compositions, with implications for tracing global hydrological processes.

Introduction

The hydrogen (δD) and oxygen (δ18O) isotopic compositions in the water molecule are very sensitive to the hydrology processes. Therefore, they have been widely used as tracers for studying the global water cycle (Dansgaard, 1964; Galewsky et al., 2016; Winnick et al., 2014; Worden et al., 2007). In 1961, the International Atomic Energy Agency (IAEA), in cooperation with the World Meteorological Organization (WMO), established the Global Network of Isotopes in Precipitation (GNIP) to measure D and 18O isotopes contents in precipitation around the world (IAEA/WMO). As described by the Rayleigh model (Dansgaard, 1964), precipitation isotopic compositions depend on three climatic parameters, i.e., air temperature, precipitation amount, and air mass history, which is termed as temperature effect, amount effect, and circulation effect, respectively (Le Duy et al., 2018). The circulation effect of precipitation stable isotopes, including the origin, condensation history and mixing of the water vapor with continental air masses caused by evaporation and/or evapotranspiration, and the free troposphere vapor along transport path (Cole et al., 1999).

On the temporal scale, a number of isotope studies paid attention to monthly, seasonal, annual, and inter-annual variability using the GNIP database (Cai and Tian, 2016; Moerman et al., 2013; Vachon et al., 2010). On the spatial scale, the long-term arithmetic and long-term weighted mean δD and δ18O (weighing by the amount of precipitation) have been most commonly used to interpret the GNIP isotope data for over sixty years (Araguás-Araguás et al., 1998; Bowen and Wilkinson, 2002; Craig, 1961a, Craig, 1961b; Dansgaard, 1964; Galewsky et al., 2016; IAEA/WMO, 2019; Rozanski et al., 1993; Winnick et al., 2014; Worden et al., 2007). A few other methods have been adopted to extract information from isotope data, such as spatial precipitation isotope gradients (Liu et al., 2010). These long-term means show spatial distributions that vary predominantly with latitude, reflecting the temperature effect of isotope fractionation during vapor condensation (Dansgaard, 1964), and depicts well the moisture transport regionally (Araguás-Araguás et al., 1998; Liu et al., 2010; Rozanski et al., 1993). Unfortunately, on the global scale, the ocean-to-continent paths of moisture movement are not clear, as well as the circulation effect of precipitation stable isotopes.

Precipitation on the continents is closely linked to atmospheric fronts, monsoons, and cyclones (Catto et al., 2012; Ding et al., 2015; Wang et al., 2012; Wang et al., 2014, Wang et al., 2017a, Wang et al., 2017b). For example, the water vapor of precipitation on the Asia continent is dominated by five moist air masses, originating from the northern Pacific, the western equatorial Pacific, the Indian Ocean, the Arctic and local continental evapotranspiration, with the first three driven by the East Asian monsoon, South Asian monsoon, and Indian monsoon, respectively, and the last two together driven by cold fronts (Catto et al., 2012; Wang et al., 2016). Although these important ocean-land interaction aspects of the moisture transportation are shown in the current analysis of the GNIP isotope data based on the long-term arithmetic and/or long-term weighted mean δD and δ18O (Araguás-Araguás et al., 1998; Liu et al., 2010; Rozanski et al., 1993), they are limited to regional and continental scales. On the global scale, can the precipitation isotope data be analyzed further to reveal, directly, the origin and movement of water vapor in connection with evapotranspiration, atmospheric circulation, and condensation?

The GNIP data, based on monthly composite samples of precipitation at each station, provides episodic information about isotopic composition of atmospheric water vapor resulting from successive precipitation events, each typically with variable isotope characteristics. In addition to the long-term mean as a statistical measure of the random isotopic compositions data, we explore here how information entropy in connection to the probability distribution of measured isotopic compositions may be applied as a new proxy for further analyzing the GNIP data. Similar to the original entropy concept for a thermodynamic system, information entropy or Shannon entropy was introduced to quantify the amount of uncertainty associated with a probability distribution (e.g., shown in a time series of a random variable) and serve as a measure of the level of mixing and chaos within a system (Shannon, 1948; Singh, 1997). It has been widely applied in hydrology study (Singh, 1997; Singh et al., 2017), such as analysis of precipitation variability (Mishra et al., 2009; de Rodrigues da Silva et al., 2016), flow evaluation and forecast (Cui and Singh, 2016; Hao and Singh, 2012), characterizing drought and flood (Rajsekhar et al., 2015; Rajsekhar et al., 2013), estimation of evaporation (Wang and Bras, 2011; Xu et al., 2019), hydrometric network evaluation and design (Li et al., 2012; Wang et al., 2019), etc. Here, we applied the information entropy concept to analyze the GNIP isotope data with the aim to examine possible links between the spatial variations of isotope entropy and moisture movement on the global scale.

Section snippets

Materials and methods

On the pathway of water vapor moving from the ocean origins to continents, mixing of different air masses increases the variability of isotopes in water vapor (Fig. 1). This in turn raises the variability of isotopes in precipitation, which is also affected by fluctuations of local temperature and precipitation amount, due to the effects of both parameters on fractionation during vapor condensation. If the raw data of precipitation isotopes are filtered to have the local temperature and

δD and δ18O entropy calculation

At each selected GNIP station, monthly averaged values of δD, δ18O, near-surface air temperature, and precipitation rate were taken as the raw data from the GNIP Database. Power spectral densities (PSD) calculated from the time series of these data show similar, predominantly seasonal (over a yearly cycle) periodicity of the δD and δ18O time series to that of temperature and precipitation rate (Fig. 2, Fig. 3). The results shown in these two figures are based on the Ottawa and Hong Kong

Uncertainties

The two important effects of local temperature and precipitation amount were filtered before entropy calculation. To do so, we assumed a simple linear relation of isotopic compositions values with temperature and precipitation rate, which may cause some uncertainties. It was suggested that there existed a piecewise relationship between isotopic compositions values and temperature (Dansgaard, 1964), and a power relationship between isotopic compositions values and precipitation amount (Fischer

Concluding remarks

The global water cycle has been extensively studied based on δD and δ18O in the water molecule. However, the interpretation of GNIP isotope data on the spatial scale has been so far limited in the long-term arithmetic and/or long-term weighted mean δD and δ18O (weighing by the amount of precipitation). The spatial distributions of mean δD and δ18O show the regional moisture transport well, but on the global scale, they obscure the global ocean-to-continent paths of moisture movement. We

Declaration of Competing Interest

The authors declare that they have no conflict of interest.

Acknowledgements

The authors would like to acknowledge the support of the Fundamental Research Funds for the Central Universities (No. 2018B48814 and No. 2019B11314), the International Postdoctoral Exchange Fellowship Program from Chinese Postdoctoral Council (No. 20180071), and the National Natural Science Foundation of China (No. 41803001). All data are available from the GNIP Database and accessible at: https://nucleus.iaea.org/wiser. The data reported in this paper are tabulated in the supplementary

References (51)

  • G.J. Bowen et al.

    Spatial distribution of 18O in meteoric precipitation

    Geology

    (2002)
  • Z. Cai et al.

    Atmospheric controls on seasonal and interannual variations in the precipitation isotope in the East Asian Monsoon region

    J. Clim.

    (2016)
  • J.L. Catto et al.

    Relating global precipitation to atmospheric fronts

    Geophys. Res. Lett.

    (2012)
  • J.E. Cole et al.

    Climatic controls on interannual variability of precipitation δ18O: simulated influence of temperature, precipitation amount, and vapor source region

    J. Geophys. Res.

    (1999)
  • H. Craig

    Isotopic variations in meteoric waters

    Science (80-. )

    (1961)
  • H. Craig

    Standard for reporting concentrations of deuterium and oxygen-18 in natural waters

    Science (80-. ).

    (1961)
  • H. Cui et al.

    Minimum relative entropy theory for streamflow forecasting with frequency as a random variable

    Stoch. Env. Res. Risk A.

    (2016)
  • W. Dansgaard

    Stable isotopes in precipitation

    Tellus

    (1964)
  • V.P. de Rodrigues da Silva et al.

    Shannon information entropy for assessing space-time variability of rainfall and streamflow in semiarid region

    Sci. Total Environ.

    (2016)
  • Y. Ding et al.

    From MONEX to the global monsoon: a review of monsoon system research

    Adv. Atmos. Sci.

    (2015)
  • M.J. Fischer et al.

    A climate-isotope regression model with seasonally-varying and time-integrated relationships

    Clim. Dyn.

    (2011)
  • J. Galewsky et al.

    Stable isotopes in atmospheric water vapor and applications to the hydrologic cycle

    Rev. Geophys.

    (2016)
  • L. Gimeno et al.

    On the origin of continental precipitation

    Geophys. Res. Lett.

    (2010)
  • L. Gimeno et al.

    Influence of the intensification of the major oceanic moisture sources on continental precipitation

    Geophys. Res. Lett.

    (2013)
  • Z. Hao et al.

    Entropy-copula method for single-site monthly streamflow simulation

    Water Resour. Res.

    (2012)
  • Cited by (10)

    • A global perspective on propagation from meteorological drought to hydrological drought during 1902–2014

      2022, Atmospheric Research
      Citation Excerpt :

      The water vapor transport capacity can affect the hydrological cycle, and the hydrological cycle in the humid climate region has greater impacts on drought propagation than other factors (such as land cover and human activities) (Forootan et al., 2019; Ding et al., 2021a). Wang et al. (2020) showed that the water vapor transport in the humid climate region was stronger than that in the arid climate region, and there was more water vapor in the humid climate region than in the arid climate region. Therefore, compared with the arid climate regions, the drought response relationship of the humid climate regions was greater.

    • Propagation of meteorological to hydrological drought for different climate regions in China

      2021, Journal of Environmental Management
      Citation Excerpt :

      The different drought propagation levels among wet and dry climates (Forootan et al., 2019) clearly indicate that moisture transport impacts the hydrological cycle. The wetter climate areas have more moisture than drier climate zones (Wang et al., 2020b); therefore, a kind of hydrological cycle is produced by moisture transport in wet climate areas. The kind of hydrological cycle might have a larger impact on drought propagation than other factors (such as land cover, landform and human activity) in wet climate areas such as Brazil, South Asia and Central Africa.

    • Atmospheric and Land Drivers of Streamflow Flash Droughts in India

      2024, Journal of Geophysical Research: Atmospheres
    View all citing articles on Scopus
    View full text