If we are to use ¹⁸O and ²H to trace groundwater recharge, then it is necessary that their concentrations in precipitation provide a characteristic input signal that varies regionally and over time. The rainout process in clouds is driven by decreasing temperature, a parameter with both regional and temporal variability. It is temperature that controls the partitioning of isotopes in precipitation, and provides the variable input function used to trace groundwater recharge.

Characterizing the stable isotope distributions in meteoric waters is essential toin determining this input function. The local meteoric water line provides a baseline for groundwaters. The position of meteoric waters on this line is controlled by a series of temperature-based mechanisms that drive the rainout process. These include vapour mass trajectories over continents, rising over topographic features, moving to high latitudes, and seasonal effects. Each has a characteristic effect on the stable isotopic composition of precipitation.

From calculations in Chapter 2, we see that as decreasing temperature drives the rainout process, the precipitation becomes increasingly depleted in ¹⁸O and ²H. Weather is of course not so simple, and this evolution is complicated by re-evaporation and atmospheric mixing. Most weather systems acquire new sources of vapour along their paths that can mask general evolutionary trends in a evolving vapour mass. Nonetheless, a strong correlation exists between temperature and isotopes in precipitation. Accordingly, where temperature gradients exist, gradients in d¹⁸O and d²H should be observed.

In the following discussion, the temperature data used in the T–d¹⁸O correlations are surface measured temperatures, and not in-cloud temperatures. It is the in-cloud temperatures that control condensation and isotope fractionation. For obvious reasons, these cannot be routinely measured, and so correlation with surface air temperatures or mean annual air temperatures (MAAT) are made.

d ¹⁸O on the global scale

Dansgaard, in 1964, established a linear relationship between surface air temperatures and d¹⁸O for mean annual precipitation on a global basis (Fig. 3-1):

If monthly average temperatures are used, the global relationship for d¹⁸O becomes:

d¹⁸O = (0.338 ± 0.028) T_monthly – 11.99 ‰ VSMOW Yutsever and Gat (1981)

On average, a 1‰ decrease in average annual d¹⁸O corresponds to a decrease of about 1.1 to 1.7°C in the average annual temperature. Corresponding variations occur for deuterium, and this covariance is the principal reason for the linear relationship or GMWL defined by Craig.

The global map of d¹⁸O values for precipitation makes a nice illustration of the partitioning of isotopes between cold and warm regions. Fig. 3-2 was created from mean annual precipitation data collected within the IAEA-World Meterological Organization survey of precipitation, using a geographical information system (GIS) for contouring. On this global scale, the partitioning of ¹⁸O into warmer, low-latitude precipitation is clear.

However, the global T–d¹⁸O relationship is only an approximation. O, and on a regional basis it is far from linear. The extensive data base collected from IAEA stations over the past 30 years has been rigorously evaluated by Rozanski et al. (1993). These data are available at the IAEA website <www.iaea.or.at:80/programs/ri/gnip/gnipmain.htm>.

The extensive monitoring network of the IAEA (right Fig. 3-2diagram in Fig. 3-1) shows that the T–d¹⁸O relationship for worldwide precipitation comprises different curves for specific geographic regions. The distinctions between marine and continental stations in this figure show the importance of geographic effects. Marine stations correlate poorly with global data due to the damping of seasonal variations in temperature and precipitation. The Canadian interior stations of Fort Smith to Gimli depart from the global relationship due to continental effects and the seasonality of precipitation.

Fig. 3-1 The mean annual d¹⁸O-values for precipitation as a function of theglobal T–d¹⁸O relationship for precipitation, modified from Dansgaard, 1964 (left). Temperature is mean annual air temperature (MAAT) of the sampling station (modified from Dansgaard, 1964at the station. Data from the extensive IAEA Global Network for Isotopes in Precipitation (GNIP) shows this relationship to be a combination of regional T–d¹⁸O lines, with strong differences between marine, continental and interior stations, from Rozanski et al., 1993 (diagram on right).

Departures from the global T–d¹⁸O relationship occur at the regional to local scale, due to physiographic variations. Departures also occur when monitoring data are examined for only short time periods. The correlation of d¹⁸O with temperature at the event-scale is very poor, and demonstrates that individual weather patterns, storm tracks and air mass mixing are far too chaotic to develop a clear T–d¹⁸O relationship at the local or event scale. The stochastic nature of weather essentially precludes the use of d¹⁸O as a proxy for temperature at anything less than seasonal to multi-annual scale. Global climate data sets are available at the NOAA website found at <http://ncdc.noaa.gov>.

Fig. 3-2 Mean d¹⁸O distribution in precipitation on a global basis, for stations with at least 24 months of records. (based on IAEA World Meteorological Precipitation monitoring data summarized by Rozanski et al., 1992).

 
Local effects on T–d^18O

Most studies of groundwater recharge rely on local rather than continental scale variations in the isotopic composition of precipitation. Variations in T and d¹⁸O imparted by the local physiographic setting, i.e. local topography, proximity to surface water bodies, seasonal changes, etc. can provide characteristics that are preserved in the groundwater and provide insights into recharge.

Altitude effect

In any region with even minor relief, orographic precipitation will occur as a vapour mass rises over the landscape. Rainout proceeds as the air mass cools, imparting a depletion on precipitation. Thus, at higher elevation and cools adiabatically (by expansion), thus driving rainout. At higher altitudes where the average temperatures are lower, precipitation will be isotopically depleted. For ¹⁸O, the depletion varies between about –0.15 and –0.5‰ per 100-m rise in elevation, altitude, with a corresponding decrease of about –1 to –4‰ for ²H. This altitude effect (also called the alpine or elevation effect) is useful in hydrogeological studies, as it distinguishes groundwaters recharged at high altitudes from those recharged at low altitude. The effect is observed even in watersheds with elevation contrasts of less than a few hundred meters, provided that sufficient data are collected to resolve seasonal effects.

One of the nicest examples is presented by Bortolami et al. (1978) for a catchment in the maritime piedmont of the Italian Alps (Fig. 3-6). Here, two distinct altitude-d correlations were observed, each with almost identical gradients (~ –0.31‰ per 100 m rise), but differing slightly in their intercepts. The differences are seasonal: fall precipitation in this region originates over the Atlantic, whereas spring weather comes from the Mediterranean Sea. The meteoric water lines they calculate also reflect these seasonal patterns: for the October line, d²H = 8 d¹⁸O + 12 (close to the GMWL) and April, d²H = 7.9 d¹⁸O + 13.4 (closer to the EMWL for arid Mediterranean climates).

In a study of recharge to a geothermal system at Mount Meager, a Quaternary volcano in the Coast Range of western British Columbia, precipitation collected from 11 sites between 250 m and 3250 m altitude shows an altitude effect of –0.25‰ per 100-m rise (Clark et al., 1982) which provided evidence for the recharge environment of the thermal groundwaters (discussed in Chapter 9). In the Jura Mountains of northern Switzerland, Siegenthaler et al. (1983) calculate a gradient of –0.2‰ d¹⁸O per 100-m rise. Table 3-1 gives the altitude gradient found in a variety of locations studies.

Fig. 3-6 The relationship between altitude and d ¹⁸O in precipitation in Val Corsaglia, maritime piedmont of the Italian Alps (Bortolami, 1978). Samples were collected in October 1974 and April 1976, representing months of the fall and spring seasons with similar mean monthly temperatures. The mean gradient for these data is –0.31 ‰ d¹⁸O per 100-m rise.

Table 3-1 Range of values for the d¹⁸O-elevationaltitude gradient in different studies

A very detailed analysis of altitude effects was undertaken on Mount Cameroun on the Atlantic coast of Western Equatorial Africa. There, J.-Ch. Fontes and co-workers monitored precipitation during a 4-year period at 20 stations between sea level and 4095 m (Fontes and Olivry, 1977). A rather low gradient of –0.155 ± 0.005 d¹⁸O‰ per 100-m rise was obtained, since the temperature gradient is not very steep. The isotopic evolution of the vapour reservoir and the resulting precipitation can be described by a modified Rayleigh process where only partial removal of the liquid phase from the vapour reservoir occurs. This allows in-cloud re-equilibration between the liquid and vapour. Gonfinatini (1996) showed that with increasing altitude an increasing amount of liquid is retained in the cloud, and by 4000 m a liquid:total-water ratio of 0.45 was reached. Such may also be the case in other situations and can at least partially explain deviations from a simple Rayleigh distillation.

Kinetic effects of secondary evaporation

Most meteoric and subsurface processes shift the d¹⁸O–d²H signature of waters to a position below the local meteoric water line. It is rare to find precipitation or groundwater that plots above the line, i.e. showing a deuterium excess or ¹⁸O depletion. However, in low-humidity regions, re-evaporation of precipitation from local surface waters creates vapour masses with isotopic contents that plot above the local meteoric water line. If such vapour is re-condensed in any significant quantity before mixing with the larger tropospheric reservoir, the resulting water will also plot above the LMWL, along a condensation line with slope 8.

There are only a few meteorological systems that cause such shifts. Ingraham and Matthews show the effect for mountain fog in northern Kenya (Fig. 3-12A). Here, vapour evaporated from the hydrologically closed Chalbi desert basin rises into surrounding mountains where it condenses on local vegetation. In an Arctic environment, Lauriol and Clark (1993) show nonequilibrium evaporation from local surface waters as the vapour source for annual ice formations in Arctic caves (Fig. 3-12B). Condensation of this kinetically depleted vapour on the cold cave walls forms water and ice that also plot above the LMWL.

Fig. 3-12 Secondary evaporation effects causing meteoric waters to plot above the LMWL. A — Re-evaporation of local groundwaters in Kenya. (modified from Ingraham and Matthews, 1988). B — Evaporation of local surface waters in the northern Yukon condensing as cave ice in fossil karst terrain. In both cases, the condensed phase is in equilibrium with the vapour (equilibrium fractionation during condensation) and so plots on a line with slope ~ 8.

Problems and Solutions

1. On the global d¹⁸O map in Fig. 3-3, account for the equatorial belt of very flat gradients. Identify regions where steeper gradients provide examples of latitude, continental, and altitude effects.