![Hier klicken, um den Treffer aus der Auswahl zu entfernen](images/unchecked.gif) |
Titel |
Time-series Oxygen-18 Precipitation Isoscapes for Canada and the Northern United States |
VerfasserIn |
Carly J. Delavau, Kwok P. Chun, Tricia A. Stadnyk, S. Jean Birks, Jeffrey M. Welker |
Konferenz |
EGU General Assembly 2014
|
Medientyp |
Artikel
|
Sprache |
Englisch
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 16 (2014) |
Datensatznummer |
250088807
|
Publikation (Nr.) |
EGU/EGU2014-2963.pdf |
|
|
|
Zusammenfassung |
The present and past hydrological cycle from the watershed to regional scale can be greatly
enhanced using water isotopes (δ18O and δ2H), displayed today as isoscapes. The
development of water isoscapes has both hydrological and ecological applications, such as
ground water recharge and food web ecology, and can provide critical information when
observations are not available due to spatial and temporal gaps in sampling and data
networks. This study focuses on the creation of δ18O precipitation (δ18Oppt) isoscapes at a
monthly temporal frequency across Canada and the northern United States (US) utilizing
CNIP (Canadian Network for Isotopes in Precipitation) and USNIP (United States Network
for Isotopes in Precipitation) measurements.
Multiple linear stepwise regressions of CNIP and USNIP observations alongside NARR
(North American Regional Reanalysis) climatological variables, teleconnection indices, and
geographic indicators are utilized to create empirical models that predict the δ18O of
monthly precipitation across Canada and the northern US. Pooling information from
nearby locations within a region can be useful due to the similarity of processes and
mechanisms controlling the variability of δ18O. We expect similarity in the controls on
isotopic composition to strengthen the correlation between δ18Oppt and predictor
variables, resulting in model simulation improvements. For this reason, three different
regionalization approaches are used to separate the study domain into “isotope zones” to
explore the effect of regionalization on model performance. This methodology
results in 15 empirical models, five within each regionalization. A split sample
calibration and validation approach is employed for model development, and parameter
selection is based on demonstrated improvement of the Akaike Information Criteria
(AIC).
Simulation results indicate the empirical models are generally able to capture the overall
monthly variability in δ18Oppt. For the three regionalizations, average adjusted-R2 and
RMSE (weighted to number of observations within each isotope zone) range from 0.70 – 0.72
and 2.76 - 2.91, respectively, indicating that on average the different spatial groupings
perform comparably. Validation weighted R2and RMSE show a larger spread between
models and poorer performance, ranging from 0.45 – 0.59 and 3.28 – 3.39, respectively.
Additional evaluation of simulated δ18Oppt at each station and inter/intra-annually is
conducted to evaluate model performance over various space and time scales. Stepwise
regression derived parameterizations indicate the significance of precipitable water content
and latitude as predictor variables for all regionalizations. Long-term (1981-2010) annual
average δ18Oppt isoscapes are produced for Canada and the northern US, highlighting the
differences between regionalization approaches. 95% confidence interval maps are
generated to provide an estimate of the uncertainty associated with long-term δ18Oppt
simulations.
This is the first ever time-series empirical modelling of δ18Oppt for Canada
utilizing CNIP data, as well as the first modelling collaboration between the CNIP
and USNIP networks. This study is the initial step towards empirically derived
time-series δ18Oppt for use in iso-hydrological modelling studies. Methods and
results from this research are equally applicable to ecology and forensics as the
simulated δ18Oppt isoscapes provide the primary oxygen source for many plants and
foodwebs at refined temporal and spatial scales across Canada and the northern US. |
|
|
|
|
|