Depends on location. d18O is a good proxy for global temperatures for polar ice cores. d180 in sediments is more complex (same for 13C/12C).
Not even even in ice cores is the relationship by any means a "good" proxy for anything other than a final byproduct of the history of cloud formation. It has the same problems no matter the matrix in which it finally is deposited.
What is clear is d18O can only represent conditions (to such degree as it might) of the atmosphere through which it has passed. The altitude & place the water vapor condensed to form droplets and ice particles forming clouds and is very dependant upon the precise history of the formation of the precipitation and number of times the ice/rain has gone through the vapor/liquid phase to its final state and the route it takes to get to its final resting place in the glacier. The correspondence to any specific temperature and place is very uncertain being a process primarily of the high atmosphere in which the precipitation formed which, depending on meteorological factors, could be anywhere at the altitudes over the oceans where water vapor precipitates in its evolution in the formation of clouds.
The Summit Ice Cores (GISP2 and GRIP
http://www.agu.org/revgeophys/mayews01/node2.html#SECTION00020000000000000000The similarity (discussed below) of the GISP2 and GRIP records is compelling evidence that the stratigraphy of the ice is reliable and unaffected by extensive folding, intrusion, or hiatuses from the surface to 2790 m (~110,000 years ago). This agreement (between the two cores separated by 30 km, ~10 ice thicknesses) provides strong support of climatic origin for even the minor features of the records and implies that investigations of subtle environmental signals (e.g., rapid climate change events with 1-2 year onset and termination) can be rigorously pursued.
The d18O of ice has classically provided the basic stratigraphy and paleoclimatology of ice cores. Ice originates by evaporation of sea water. As an air mass travels away from the site of evaporation towards higher latitudes, it cools and is able to hold less water. Water is lost from the air mass as precipitation (rain or snow). When liquid and gaseous H2O interact, the heavy isotope of O (d18O, 0.2% natural abundance) is depleted in the gas phase relative to the light, major isotope (d18O, 99.8% natural abundance). Along the flow path of the air mass, residual gas becomes progressively more depleted in d18O as temperature becomes colder and more and more of the original H2O content is lost as precipitation. The d18O value is determined from:
where subscript s refers to sample, std refers to standard mean ocean water (SMOW), and
is parts per thousand.
Independent calibrations of the oxygen isotope-temperature relationship have been developed through the analysis of GISP2 borehole temperature, allowing conversion of isotope-derived surface-temperature histories to temperature-depth profiles [ Cuffey et al., 1992]. Thus it follows that variations with depth in the d18O of ice in a core reflect past variations of temperature with time at a study site. Changes in moisture sources feeding central Greenland may provide additional complications in the interpretation of the record [ Charles et al., 1994]. Grootes et al. [1993] measured the d18O of ice in the GISP2 core and compared their record with the previously published record for GRIP [ Dansgaard et al., 1993]. Down to a depth of 2790 m in GISP2 (corresponding to an age of about 110 kyr BP (kyr before present, where present is AD1950)), the GISP2 and GRIP records are nearly identical in general shape and in most of the details.
