 |
| Titel |
Comprehensive evaluation of long-term hydrological data sets: Constraints of the Budyko framework |
| VerfasserIn |
Peter Greve, Boris Orlowsky, Sonia I. Seneviratne |
| Konferenz |
EGU General Assembly 2013
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 15 (2013) |
| Datensatznummer |
250078899
|
|
|
|
|
|
| Zusammenfassung |
| An accurate estimate of the climatological land water balance is essential for a wide range of
socio-economical issues. Despite the simplicity of the underlying water balance equation, its
individual variables are of complex nature. Global estimates, either derived from observations
or from models, of precipitation (P ) and especially evapotranspiration (ET) are characterized
by high uncertainties. This leads to inconsistent results in determining conditions related to
the land water balance and its components.
In this study, we consider the Budyko framework as a constraint to evaluate long-term
hydrological data sets within the period from 1984 to 2005. The Budyko framework is a well
established empirically based relationsship between ET-P and Ep-P , with Ep being the
potential evaporation. We use estimates of ET associated with the LandFlux-EVAL initiative
(Mueller et. al., 2012), either derived from observations, CMIP5 models or land-surface
models (LSMs) driven with observation-based forcing or atmospheric reanalyses. Data
sets of P comprise all commonly used global observation-based estimates. Ep is
determined by methods of differing complexity with recent global temperature
and radiation data sets. Based on this comprehensive synthesis of data sets and
methods to determine Ep, more than 2000 possible combinations for ET-P in
conjunction with Ep-P are created. All combinations are validated against the
Budyko curve and against physical limits within the Budyko phase space. For this
purpose we develop an error measure based on the root mean square error which
combines both constraints. We find that uncertainties are mainly induced by the ET
data sets. In particular, reanalysis and CMIP5 data sets are characterized by low
realism. The realism of LSMs is further not primarily controlled by the forcing,
as different LSMs driven with the same forcing show significantly different error
measures.
Our comprehensive approach is thus suitable to detect uncertainties associated with
individual data sets. Furthermore, combinations performing well within the Budyko phase
space are identified and could be used for future studies, like e.g. to investigate decadal
changes of the land water balance.
B. MUELLER, M. HIRSCHI, C. JIMENEZ, P. CIAIS, P.A. DIRMEYER, A.J. DOLMAN,
J.B. FISHER, Z. GUO, M. JUNG, F. LUDWIG, F. MAIGNAN, D. MIRALLES, M.F.
MCCABE, M. REICHSTEIN, J. SHEELD, K. WANG, E.F.WOOD, Y. ZHANG, S.I.
SENEVIRATNE (2012): Benchmark products for land evapotranspiration: LandFlux-EVAL
multi-dataset synthesis, Hydrol. Earth Syst. Sci., submitted. |
|
|
|
|
|
|