|
Titel |
On the cyclo-stationarity of the time-variable Kaula rule |
VerfasserIn |
B. Devaraju, C. Lorenz, M. J. Tourian, N. Sneeuw |
Konferenz |
EGU General Assembly 2012
|
Medientyp |
Artikel
|
Sprache |
Englisch
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 14 (2012) |
Datensatznummer |
250069513
|
|
|
|
Zusammenfassung |
Stochastic filtering of the monthly fields of GRACE time-variable gravity requires the
computation of a signal covariance model, which is usually computed as a power-law from
the GRACE data itself. The computation of the power-law involves the averaging of the
degree variance models of all the available months of data, and fitting a straight
line to the logarithm of the degree variances. Such signal variance models tend to
under-represent the inherent annual behaviour of the signal. In this contribution,
we demonstrate the need to calculate a cyclo-stationary signal variance model,
which represent the stochastics of the signal more truthfully. Further, we show
that when cyclo-stationarity of the signal variance is ignored, the filtered fields
are under-/over-smoothed. Due to the fact that every month uses a different signal
variance model, the resolution of the filter kernels keeps changing over the time-series.
Therefore, the time-series of maps generated by such filter kernels will not share
a common resolution, and hence, we homogenize the resolution using isotropic
filers. Finally, we will compare the time-series of GRACE fields, filtered using both
stationary and cyclo-stationary signal variance models, with observed hydrological data. |
|
|
|
|
|