a b s t r a c tMonitoring of the marine environment for indications of a leak, or precursors of a leak, will be an intrinsic part of any subsea CO 2 storage projects. A real challenge will be quantification of the probability of a given monitoring program to detect a leak and to design the program accordingly. The task complicates by the number of pathways to the surface, difficulties to estimate probabilities of leaks and fluxes, and predicting the fluctuating footprint of a leak. The objective is to present a procedure for optimizing the layout of a fixed array of chemical sensors on the seafloor, using the probability of detecting a leak as metric. A synthetic map from the North Sea is used as a basis for probable leakage points, while the spatial footprint is based on results from a General Circulation Model. Compared to an equally spaced array the probability of detecting a leak can be nearly doubled by an optimal placement of the available sensors. It is not necessarily best to place the first in the location of the highest probable leakage point, one sensor can monitor several potential leakage points. The need for a thorough baseline in order to reduce the detection threshold is shown.

No abstract

a b s t r a c tThe risk and environmental impact assessments required for geological CO 2 storage projects will have to rely on different types of numerical models, which will have to be calibrated and validated against measurements. Available measurements from ongoing demonstration projects are limited, hence it is necessary to turn to analog processes or laboratory experiments to estimate model parameters. In any case, parameter estimates will have uncertainties that will be important to assess when predicting future scenarios.We study a model for the rise velocity of droplets in the ocean, an important process sub-model for simulating gas seeps into marine waters. As the origin we use the parameters estimation study by Bigalke et al. (2010) based on a tank experiment. We illustrate how Linearized Covariance Analysis (LCA) can be used to assess the parameter uncertainties, and how to design a similar experiment that reduces these uncertainties. The linearity assumption underlying LCA is assessed using curvature measures. It is shown that up to ∼63% reduction in uncertainties is achieved by choosing the right droplet size distribution; by extending the range of droplet sizes to include larger droplets the uncertainties are reduced by another ∼88%.

No abstract