DSpace Collection:http://hdl.handle.net/11368/28293532020-05-27T06:56:25Z2020-05-27T06:56:25ZRecursive Polynomial Chaos Co-Kriging for Reliability-based Design Optimisationhttp://hdl.handle.net/11368/29627582020-04-27T10:54:48Z2019-01-01T00:00:00ZTitle: Recursive Polynomial Chaos Co-Kriging for Reliability-based Design Optimisation2019-01-01T00:00:00ZSearching for an optimal and reliable design under epistemic modelling uncertaintyhttp://hdl.handle.net/11368/29627602020-04-27T08:16:34Z2019-01-01T00:00:00ZTitle: Searching for an optimal and reliable design under epistemic modelling uncertainty2019-01-01T00:00:00ZUncertainty budget of solid Earth data reductions to global gravity modelshttp://hdl.handle.net/11368/29517172020-03-17T16:27:41Z2019-01-01T00:00:00ZTitle: Uncertainty budget of solid Earth data reductions to global gravity models
Abstract: Solid Earth applications of satellite gravity models commonly involve some type of data reduction - i.e. forward
modelling the gravity effect of known mass distributions to isolate an anomaly from the observed field, which is
then attributed to the enquired phenomenon. The adopted "known masses" suffer from the uncertainties arising
from the non-modelled variance in the shape of geological bodies and the density distribution therein. These
uncertainties are propagated to the reduced gravity field, superimposed to the formal errors provided with the
gravity model. Given the different origin between formal errors of satellite global gravity models (GGM), arising
from observation and noise models, and the contribution of geophysical data reductions, we aimed at assessing
the comprehensive error characteristics of reduced-GGMs.
In order to do so, we computed a set of common reductions (topography, crustal layers, mantle inhomogeneities)
using a combination of spectral- and space-domain forward modelling. Uncertainties in the input
quantities (depths and densities) were propagated trough Monte Carlo methods.
Geometries were constrained by a topography-bedrock-ice model (Earth2014), by a global layered model of the
lithosphere (LITHO1.0), and by local higher detail models of the crust and sediments, where available. Depth
uncertainties, if not provided with the input data, were assigned according to method-specific assumptions. Estimates of density and its variance come from probability distributions fitted to literature data, from petrophysical
relationships (e.g. velocity-composition-temperature) and from worst-case assumptions where no sufficient data is
available.
We report the outcome of a set of global models, at a resolution and spectral content coherent with the
currently available satellite-only GGMs. We resort to global uncertainty maps and to the familiar representations
employed in GGM sensitivity assessments (e.g. degree error curves). Different combinations of data reductions
were applied, simulating the interest in different anomalies (e.g. by correcting either for the crust or the mantle).2019-01-01T00:00:00ZBounding First Passage Times in Chemical Reaction Networkshttp://hdl.handle.net/11368/29551002020-02-20T08:35:31Z2019-01-01T00:00:00ZTitle: Bounding First Passage Times in Chemical Reaction Networks2019-01-01T00:00:00Z