Â鶹ÊÓƵ

Buenning, Nikolaus H ¹ ; Noone, David C ²

¹ Department of Atmospheric and Oceanic Sciences and Cooperative Institute for Research in Environmental Sciences, Unverisity of Colorado, Boulder
² Department of Atmospheric and Oceanic Sciences and Cooperative Institute for Research in Environmental Sciences, Unverisity of Colorado, Boulder

The isotopic composition of precipitation (δ) is widely used for both hydrology and climate variability studies. Mapping out the spatial distribution of δ values has been done by several studies using different methods. However, there has been little work on representing the temporal variations (both seasonal and interannual variations). In this study, both long-term mean and interannual datasets are constructed based on the Global Network for Isotopes in Precipitation (GNIP) observations. This was done by using a spatial regression against physical quantities to represent the annual mean, a series of Fourier coefficients to fit the seasonal cycle, and another regression to represent the interannual component. For the regression model, temperature, precipitation amount, elevation, and latitude are used as predictors. A bias term is also included in this model, to account for non-local processes that aren’t captured by the local regression. This regression and transform (RT) approach is compared to errors and biases associated with General Circulation Models.

Trends in δ values are calculated at each grid cell, and attributed to a non-local (advected) component or a local temperature and/or precipitation variation. Figure 1a dn b show trends %delta values for two periods: 1979-2001 and 1990-2001. Recent increasing trends in δ values indicate a regional warming over certain regions of the high northern latitudes, which is captured by the regression via temperature changes. The generated datasets also indicate an increase in precipitation over Southeast Asia, and a strengthening of the Walker Circulation (for the 1979-2001 period). Trends in δ values are consistent with a north-northeastward shift in the Aleutian Low and the North Pacific storm track; a feature that was not captured by the regression alone. This approach was able to identify regions where the hydrological cycle is largely influenced by local conditions and other regions where it’s largely affect by other factors such as changes in moisture advection.