 |
| Titel |
Quantifying variogram uncertainty linked to soil moisture at field scale |
| VerfasserIn |
N. Prolingheuer, B. Scharnagl, M. Herbst, H. Vereecken |
| Konferenz |
EGU General Assembly 2010
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 12 (2010) |
| Datensatznummer |
250034946
|
|
|
|
|
|
| Zusammenfassung |
| The spatial dependence of an environmental variable is commonly estimated by
variogram analysis. However, especially for smaller data sets (< 100 sample points) the
resulting sample variogram typically includes large uncertainty. This uncertainty
is often neglected in further analysis and interpretation of the data. In this study,
we used stochastic simulation to estimate the expected variogram uncertainty and
used this information for the quantification of spatial dependence of soil water
content.
We used a data set of soil water content measured weekly from April to October 2009 in 6 cm
depth at 61 locations in a 50 Ã 50 m plot of an bare agricultural field in Jülich, Germany. For
each variogram, 1000 realizations of the underlying random field were generated using
unconditional sequential Gaussian simulation. For each realization we draw 2000 samples
within the sampling domain. From these realizations the ergodic and nonergodic sample
variogram were calculated. The ergodic variogram is the exhaustive variogram including all
2000 realizations and the nonergodic variogram is the sample variogram including only
the 61 actually sampled locations. Variogram uncertainty was calculated from the
differences between ergodic and nonergodic variograms. We fitted various variogram
models to the underlying sample variogram by iterative weighted least squares
using the uncertainty of each variogram lag as weights. We then applied among
others the corrected Akaike Information Criterion (AICc) to select the variogram
model that best described the data. Our results indicate that explicit consideration
of variogram uncertainty may strongly affect the interpretation of geostatistical
data and we propose to apply this approach routinely in geostatistical practice. |
|
|
|
|
|
|