 |
| Titel |
Estimation of the Uncertainty in Global Precipitation Observations and Its Propagation to Model Simulations of Evapotranspiration and Runoff |
| VerfasserIn |
Hyungjun Kim, Taikan Oki, Jaeil Cho, Sujan Koirara, Shinjiro Kanae, Pat J.-F. Yeh |
| Konferenz |
EGU General Assembly 2010
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 12 (2010) |
| Datensatznummer |
250041251
|
|
|
|
|
|
| Zusammenfassung |
| In this study, the uncertainties of five (three ground measurements and two satellite hybrid
products) global observed precipitation datasets and its translation into evapotranspiration
and runoff through ensemble hydrological simulations are estimated. A dimensionless index
Ω, is used to quantify the “similarity” among the simulations of ensemble members which is
assumed as a surrogate of “uncertainty” within them. The uncertainty in precipitation is in
general amplified in simulated runoff and damped in evapotranspiration, and global average
of ΩP, ΩET and ΩR are 0.84, 0.89 and 0.65, respectively. A relatively low ΩP is found in
certain mountainous areas, deserts, and Central Africa around the equator and 20Ë N. Zonal
summation of gauge station numbers and zonal mean of ΩP exhibit a similar pattern.
Globally, the spatial distribution of ΩET shows a similar pattern of ΩP but generally with
higher values, which indicates more similar temporal variability among ensemble
members of simulated evapotranspiration. However, some considerable differences
can also be observed in certain regions. Based on ΩP and ΩET, the patterns of
uncertainty propagation from precipitation to evapotranspiration can be classified
into the following four groups: (1) High ΩP (> 0.8) and higher ΩET (> 0.9), (2)
Low ΩP and high ΩET (ΩET – ΩP > 0.3), (3) High ΩP and low ΩET (ΩET –
ΩP) < -0.3, and (4) ΩP (< 0.5) and Low ΩET (< 0.5). Most regions in the
northern Hemisphere, except for arid regions, are included in Group 1, some tropical
regions (e.g., archipelago of South Pacific Ocean) are classified as Group 2, large
regions of Central Africa are shown as Group 3, and Group 4 includes mostly arid
regions and regions where observational networks are sparse. The uncertainty of
simulated runoff (ΩR), in turn, globally marks lower value than ΩP. It means that the
uncertainty in precipitation translates into the amplified uncertainty in runoff simulated,
and, therefore, ΩR – ΩP shows negative value globally. This amplified uncertainty
propagation in runoff simulation is mainly due to the weak uncertainty propagation in
evapotranspiration simulation. In terms of water balance, precipitation is partitioned into
evapotranspiration and runoff through water storage components (here, snow pack
and soil moisture), and the uncertainty also should be allocated through the same
mechanism. In global scale, evapotranspiration is rather insensitive to the precipitation
uncertainty, and it even dampens the uncertainty since plant growth in many of terrestrial
regions is constrained by energy (i.e., radiation and temperature) rather than water
availability. It results in much part of precipitation uncertainty translates into an
uncertainty in runoff, and the greater size of precipitation (about 2 fold) leads to
bigger uncertainty in the standardized index Ω. As a result, global average of ΩR
(0.65) marks a much smaller value than ΩET (0.89) and ΩP (0.84). In regional
scale, it is found that a number of relatively small ΩR – ΩP values are located in
transitional regions (e.g., around 10Ë N and 20Ë S – 40Ë S) between water-limited region
(e.g., near 20Ë N) and water-abundant region (e.g., tropical region), as similar to
the spatial pattern of ΩET – ΩP, and a high peak is observed near the equator. |
|
|
|
|
|
|