As Calvert (2005) remarked: “Most of the methods which have been introduced to infer the permeability from 4Dseismic based on the saturation change concept are based on the streamline logic” (de Haan, 2001; Kretz et al., 2004; Vasco et al., 2004). Because streamlines do not cross, they partition 3D flow problems into a series of 1D problems. These problems would be very easy to deal with if such streamlines did not change with the pressure field over time‟. In contrast to the streamlines used to recover permeability, finite-difference simulation is the main simulation modelling tool. Also, using a saturation approach to infer permeability has some shortcomings. Its applications have revealed the fact that using a pressure solution could be more useful, due to the fast diffusion of the pressure disturbance across the field, as compared with the saturation profile which is often spread over a particular region of the field. Based on this heuristic, two important developments for determining permeability have been proposed by Vasco (2004) and MacBeth and Al-Maskeri (2006). Vasco (2004) described a technique that uses pressure change estimated from 4Dseismic to infer the reservoir permeability using the finite-difference approach. His methodology starts from the governing simulation equation for a two-phase, compressible fluid, neglecting the gravity term:
The timelapse responses were calculated for the 2006-86/91 and 2011-89/91 differences. The three seismic volumes (86/91, 2006 and 2011) were pre-stack and post-stack parallel reprocessed in 2011 by WesternGeco (processing contractor) with a repeatability estimated (200 ms time window 200 ms above top reservoir) in 25% NRMS (Normalized Root Mean Square). The 4Dseismic analysis and interpretation to evaluate the effect of shale activation in this dataset was performed at two scales: the initial one considers that pressure diffusion may generate its own 4Dseismic response within shales, while in the second, the analysis is performed at the scale of the overall reservoir elastic response (including shales and sands). Due to the good lateral continuity of Heidrun Field shales and their relatively consistent thickness, the analysis was performed by applying surface- based amplitude seismic attributes (RMS) for carefully selected time windows associated with the horizons corresponding to top and base of the studied shales in the observed seismic. The extracted static and dynamic seismic amplitude response for the Not Formation (calculated with the RMS amplitude algorithm applied to a 15-millisecond time window) for the depletion and build up stages (Figure 3-16) shows the strong influence of the reservoir signature, as saturation changes in the adjacent sands (oil-water contact fluid movement and gas exsolution) are easy to recognize in the response extracted for the Not Formation interval. With no changes in the Not Formation saturation (which remains 100% water-saturated, due to capillary forces), and with no correspondence between the predicted shale pressure and the observed elastic response (Figure 3-16), the reservoir sand’s elastic response (static and dynamic) was imprinted in the extracted signal corresponding to the shale intervals, making impossible to determine if the response of the shale pressure diffusion creates a measurable seismic signature within the shale interval.
The potential of the seis2sim exercise has been applied to many fields. In the literature, most of the relevant case studies are not directly under the name of CtL, but in fact they all tend to close some of the selected loops by pre- or post-processing the 3D and 4Dseismic data. Zachariassen et al. (2006) conducted the 3D and 4D elastic inversion for the Oseberg field, the results of which led to a probabilistic classification of the reservoir sand and various facies according to the 3D and 4D data (Figure 1.10). Model realisations generated from the sand probability cubes are ranked according to the visual comparison between the synthetic and observed 4Dseismic responses. Nonetheless, their 4D seis2sim work was only validated by checking the 4D and production matches, leaving the static loop loosely closed. Ingrid et al. (2009) extended the 3D inversion work of Wijngaarden et al. (2007) to 4D and applied it to the Troll West field. The porosity and clay distribution of the geomodel were updated by the 3D inversion results, while the initial and produced oil-water contacts were inferred by the inversion of 3D and 4Dseismic. They also updated the depth location of the model accordingly (Gjerding et al., 2010). However, their 4D seis2sim work did not show the “feedback loop”, therefore the update was not verified. Leguijt (2001, 2009) introduced a probabilistic Bayesian approach to invert for 3D static reservoir properties. Floricich et al. (2010, 2011) applied the method to the Schiehallion field and one other North Sea field to update the NTG and facies distributions in the reservoir models, which led to better static seismic responses. They carried out the inversion of the Schiehallion 4D data, in which the eight time-lapse vintages were simultaneously inverted into pressure and saturations over time and compared with the model predictions. This series
and base RC logs leads to the presence of a trough amplitude at the position of the new oil water contact as well as a peak amplitude at the original oil water contact. Yet the presence of the Not Formation as laterally extensive intra-reservoir shales might impact the final outcome. Indeed, although it has been proven that fluid contact movement between Garn and Ile reservoirs is coupled (Kahar et al., 2006), the position of the shale relative to the fluid contacts might interfere with the signal derived from the fluid information. If gas expansion is assisting the water flooding, the expected RC log for the post-production scenario will also include a decrease of the P-wave impedance at the top of the oil leg leading to a decrease in the spike associated to the original gas oil contact and a new positive spike at the new gas oil contact in the post-production RC log. This will be evidenced in the RC difference log (monitor minus base) which when convolved with a European polarity wavelet will introduce a peak at the original oil gas contact and a trough at the new oil gas contact. As in the water flooding case, interference might be present in each vintage due to the presence of the thin shale beds close to the fluid contacts. This phenomenon might ultimately affect the final 4Dseismic amplitude. Nonetheless, predicted changes seem to be in agreement with the observations at the production sites. They are particularly consistent with the boundaries between compartments as well as with the mapped original fluid contacts.
The ability to image where CO 2 has been injected into an underground reservoir would not only identify any remaining space for the sequestering of additional CO 2 could occur but would also identify unswept portions of the reservoir where more oil reserves remain to be recovered. This knowledge would increase the economic advantage to petroleum industry companies and encourage drilling of additional wells for not only the recovery of these extra oil reserves but also the injection and sequestration of additional CO 2 . The overall objective for this project is to determine if advanced geophysical methods, such as 4Dseismic surveys, can be used to identify those portions of an oil-bearing reservoir that have been flooded with CO 2 , as well as those portions that remain unflooded. The reservoir chosen for this study is the Charleton 30/31 Field, a Niagaran reef field in the northern Michigan basin.
seismic surveys. Usually, such models are used for initializing the dynamic reservoir modeling process, which involves the dynamic simulation of fluid flows within a reservoir. Typical dynamic datasets in hydrocarbon energy applications include historical production data (flow rates or volumes and pressure data) and time-lapse4Dseismic data. Borehole measurements are often the main information source for reservoir modeling; however, boreholes are sparsely distributed, and these measurements are not sufficiently informative to yield accurate and detailed representations of the whole 3D reservoir volume. There are many non-unique models that fit the sparse well data despite the huge challenge posed by extrapolating these data to an entire reservoir volume. Due to their excellent spatial coverage, 3D seismic data play a key role in defining the structure and geometry of the reservoir, and in setting constraints to variations in reservoir properties. To produce realistic models of reservoir lithofacies and corresponding petrophysical properties while avoiding non-physical results at the same time, 3D seismic information should be actively incorporated into the static reservoir property of modeling process. On the other hand, 4Dseismic data are powerful constraints on dynamic reservoir models because of its valuable information relating to production-induced reservoir changes such as fluid movements and pressure and saturation changes.
compared with the impedances obtained from the simulation model through the petro- elastic modelling. This approach, albeit visually compelling, has drawbacks of the previous two: inversion to the impedances is likely to ignore the uncertainties, and petro-elastic modelling for the simulation model will need reliable petro-elastic model parameters. However, the domain of impedances appears to be the most popular in the history matching literature, see e.g. , , , , . While there are reports of better history matching performance in this domain rather than e.g. in the amplitudes domain , it is still not clear whether the popularity of history matching in the impedances domain is due to its robustness, or because this domain is an acceptable compromise which is understood by both engineering and geophysical communities. There are also approaches which avoid the direct comparison (e.g. in the least-squares sense) of the observed and modelled time-lapse reservoir signatures, calculating instead some correlation measures between the quantities in question. Usually they are employed for the reservoirs where petro-elastic modelling is challenging. For example, Waggoner et al.  used the normalised cross-correlation between the observed and modelled maps of acoustic impedance for history matching of a Gulf of Mexico gas condensate reservoir. Kjelstadli et al.  employed the correlation between the observed and modelled attribute maps to history match a North Sea compacting chalk reservoir, where adequate seismic modelling was problematic.
et al. 2000) are injected, to ensure injectant conformance and flood front management, maximizing recovery and minimizing operational costs. The availability of dense areal information from frequent 4Dseismic offers a great opportunity to achieve these goals. It enables better understanding of reservoir sweep and flow patterns, reduction of the uncertainty in the reservoir properties and adjustment of the operational strategy to restore conformance and optimize recovery (Foster 2007; Przybysz-Jarnut et al. 2015; Watanabe et al. 2017). However, it also poses new challenges in terms of dynamic reservoir modeling and seismic history matching to infer changes in the state of the reservoir. The underlying issues for successful monitoring of reservoir fluid-flow systems using time-lapse data were reviewed by Lumley (2001) and Behrens et al. (2002).
Preliminary analysis of the 4-D seismic data shows that the producing reservoir is situated on the crest of a multi-fractured rollover anticlinal structure (Figure 2), bounded by a major growth fault on the North-East of the field. The significant petroleum reservoir sands in the field consist mainly of middle Miocene deltaic sand- stones, which are poorly consolidated, with high effective porosities and permeability’s . This makes the field ideally suited for a time-lapse analysis, as effects due to fluid saturation change could be readily discernible.
and compartmentalisation. The signature around well INJ1 displays features that could be easily interpreted as flooded areas where no saturation change has occurred in the simulation model (red ellipse in Figure 5.18). It also shows a dimming in 4Dseismic amplitude where we had a negative change in water saturation (red ellipse in Figure 5.18), contrarily to what is expected. The 4Dseismic signature around INJ2 could be interpreted as the result of flow barriers within the reservoir at the vicinity of this well as it is divided into two separate features inferring that there is no hydraulic communication between the corresponding areas within the reservoir, which is not correct. This is further highlighted in the amplitude envelope map. As the dominant seismic frequency increases, the seismic resolution increases as a consequence and more subtle details about the swept areas are shown in the 4D RMS amplitude maps. Most importantly, the 62 Hz seismic and the 125 Hz seismic successfully produced continuous drainage maps similar to those computed from the fluids flow simulation. Interpretation of 4D RMS amplitude and 4D RMS amplitude envelope maps yield an accurate mapping of fluids contacts and the highly uneven fluid front. The 125 Hz seismic resolved flooded areas as small as a few meter large. It also shows a less smooth and continuous waterfront, highlighting the strong variations of rock properties inside the reservoir.
2.1 Qualitative Use of 4DSeismic Data
The simplest, most direct method of using time-lapseseismic data is to qualitatively monitor reservoir changes due to production. In this ap- proach, one simply identiﬁes regions in which the amplitude or impedance has changed with time and attributes these changes to changes in satura- tion, pressure, or temperature . Time-lapseseismic is not a new topic in petroleum engineering and geophysics. The pioneering work of time-lapseseismic can be traced back to late 1980s and early 1990s, e.g., Wayland and Lee . Similar studies have been reported by Cooper et al. at the Foin- haven ﬁeld  and by Lumley et al.  at the Meren ﬁeld in Nigeria. The primary objectives at Foinhaven were simply to map ﬂuid movements and to identify by passed oil. The authors of the study concluded that the time- lapse signal qualitatively agreed with the expected reservoir performance. At Meren, the goal was to identify pathways of injected water, sealing faults, and compartments that may contain bypassed oil. 4Dseismic has been very useful on the Gullfaks ﬁeld to identify areas where signiﬁcant gas saturation changes have occurred and to locate ﬂuid communication paths . Also for the Gullfaks ﬁeld, 4Dseismic has been used to ascer- tain depleted areas and so far, 14 inﬁll wells have been drilled based on 4D studies  . For the Gannet C oil and gas ﬁeld in the UK central North sea, 4D data revealed major extensions of reservoir units previously presumed to be absent or thin over much of the reservoir . More recently, for the Heidrun ﬁeld in the North sea, 4Dseismic data improved the understanding of reservoir ﬂuid ﬂow and communication across faults .
uniqueness makes the choice of the “best model” in sim2seis studies challenging. Indeed, any PEM may act as a “prime” for the proxy, which can then be used in sim2seis modelling calculations on its own. Keeping in mind that the goal is searching for a rock physics model that is accurate, robust and simple for timelapse interpretation, judging all the models based on the concept of Occam’s razor seems reasonable: choosing the least complex PEM with the least number of parameters that fits the data well. Based on this philosophy the simplified linear proxy model with two parameters, may work well in practice to resolve this difficulty of building and choosing the right PEM, especially for an oil-water system (i.e. no gas). The presence of gas in any field, increases the uncertainty and predictability of the 4D maps using a simple two parameters mathematical equation. Nevertheless, from a quantitative point of view, considerations and attention must be taken into consideration. The proxy model works for an oil-water system, although big errors were shown in the time-lapse maps in Chapter 5 because of the uncertainty related to the laboratory stress sensitivity term in the PEM. Another uncertainty is the flow simulator model itself, if the reservoir model has a poorly history matched pressure and production and injection fluid data, then errors will be carried on through the entire 4Dseismic study, indifferently of the rock physics model chosen and its associated calibrated input parameters.
guidelines describing the feasibility and applicability of these techniques (Lumley and Behrens, 1998).
There are many technical challenges associated with the use of repeated 3D seismic surveys to infer changes in the state of the reservoir (Lumley, 2001; Behrens et al., 2002). Nonetheless, 4Dseismic data have demonstrated success as a monitoring tool for mapping changes in phase saturations and pressure and, therefore, the inference of areas of bypassed oil, fluid contacts and pressure compartmentalization (Foster, 2007). Appropriately, 4Dseismic is at present most commonly recognized as a reservoir management tool, where a number of successful field experiences of reservoir management have used timelapseseismic information (Behrens et al., 2002; Cooper et al., 2005; Landro et al., 1999). The logical progression of this technology is for the application of the dynamic information, inferred through timelapseseismic interpretation, in reservoir model calibration. Specifically, in high-resolution geologic models the areally dense characteristics of the seismic data are expected to compensate for the lack of production data resolution, particularly away from well locations. Nonetheless, the reconciliation of reservoir model heterogeneity with temporal changes in seismic attributes is a remarkably complex task (Gosselin et al., 2001). Several such dynamic data integration algorithms have been proposed in the literature, which we categorize as two distinct types: (1) methods that use direct seismic attributes (e.g., reflection amplitude) and (2) methods that use seismic inverted properties derived from a geophysical inversion (e.g., elastic acoustic impedance, compressional velocity, saturation changes, etc).
It is common in practice for well production and injection rates to fluctuate or for wells to be shut-in, re-started or turned off during acquisition of seismic data for time-lapse studies. Additionally, in any field numerous wells affect pressure locally and at field- scale while the seismic survey is ongoing. This means that the reservoir is typically not in pressure equilibration during the acquisition (if at all during its lifetime), and pressure will fluctuate in and around wells, and also between wells. As the reservoir always remains active, there are also many fluid changes occurring during the duration of the survey. I therefore believe that the intra-survey effect observed in this study is possibly widespread across many 4Dseismic field datasets but remain at background noise levels due to the dominance of data non-repeatability noise. As data repeatability improves, as is currently achievable by PRM technology (NRMS ~ 2%), such intra- survey effects will become prevalent, and could remain unobserved due to the coarse way in which we compare time-scales between the simulator and seismic domains. This may add a general floor of small-scale fluctuations within the non-repeatability noise attributed to processed 4Dseismic data. For wells with significant responses during or close to the start of the survey, the impact has been demonstrated to be significant. It is typical to match the simulation results to the start, middle or end time of the seismic survey, however, it is suggested in this study that perhaps the best comparison may be achieved by averaging fine-scale simulation predictions between the start and the end of the survey.
In recent years, the magnetotelluric method (MT) has been increasingly used as a cost-effective technique for subsurface resistivity monitoring (see Rees et al. 2016, for a general introduction). As this is a relatively new application for MT, there are few modelling tools avail- able. Peacock et al. (2012) have used differences in MT phase tensors to qualitatively interpret resistivity changes due to fluid injection in an enhanced geothermal system (EGS) at Paralana, South Australia. This technique has also be used by Didana et al. (2017) interpreting change at the Habernero EGS. Recent work includes parameteris- ing resistivity changes as a three-dimensional (3D) plume structure and inverting using Markov Chain Monte Carlo (Rosas-Carbajal et al. 2015). Another approach has been to use 1D layer-stripping, where the effect of over- lying structures is removed to model the time-varying
Analysis of cellular behavior is significant for studying cell cycle and detecting anti-cancer drugs. It is a very difficult task for image processing to isolate individual cells in confocal microscopic im- ages of non-stained live cell cultures. Because these images do not have adequate textural varia- tions. Manual cell segmentation requires massive labor and is a time consuming process. This pa- per describes an automated cell segmentation method for localizing the cells of Chinese hamster ovary cell culture. Several kinds of high-dimensional feature descriptors, K-means clustering me- thod and Chan-Vese model-based level set are used to extract the cellular regions. The region ex- tracted are used to classify phases in cell cycle. The segmentation results were experimentally as- sessed. As a result, the proposed method proved to be significant for cell isolation. In the evalua- tion experiments, we constructed a database of Chinese Hamster Ovary Cell’s microscopic images which includes various photographing environments under the guidance of a biologist.
What are the experiences of using the EmbryoScope™ time-lapse system in a clinical setting? Users at current fertility clinics find that the technology ultimately provides valuable flexibility in the daily routines of the clinic, because embryologists are free to carry out evaluations of embryo development at times that suit them and can easily consult other specialists or colleagues if necessary. Embryologists are able to use more time evaluating embryos, but less time with physical processes such as removing and replacing from the incubator and taking manual notes.
ABSTRACT The chick hindbrain starts from a simple and relatively uniform axis and becomes segmented into repeating units, called rhombomeres. The rhombomeres become sites of cell differentiation into specific neurons and the location from which neural crest cells emerge from the neural tube to form the peripheral nervous system, which has only been analyzed at distinct time points due to the lack of a method to watch the neural tube as it is shaped into segments. We have developed a whole-embryo explant culture system in order to study cell and tissue movements with time-lapse video microscopy. Quantitative analyses of the neural tube during its segmentation show that not all rhombomeres are shaped by the same mechanism. In the rostral hindbrain, or first three segments, rhombomeres are shaped by an expansion in the lateral width of the mid- rhombomere; either a smaller expansion or a constriction takes place at the rhombomere bounda- ries. In the caudal hindbrain, the rhombomere boundaries constrict more than the mid-rhombomere lateral widths increase or decrease, leading to the shaping of the segments. Throughout the segmentation process the rostrocaudal lengths of all rhombomeres remain nearly constant indicating that shape changes are influenced by lateral expansions and constrictions of the neural tube.
Motions can occur over both short and long time scales. We introduce motion denoising, which treats short-term changes as noise, long-term changes as signal, and re- renders a video to reveal the underlying long-term events. We demonstrate motion denoising for time-lapse videos. One of the characteristics of traditional time-lapse imagery is stylized jerkiness, where short-term changes in the scene appear as small and annoying jitters in the video, often ob- fuscating the underlying temporal events of interest. We ap- ply motion denoising for resynthesizing time-lapse videos showing the long-term evolution of a scene with jerky short- term changes removed. We show that existing ﬁltering ap- proaches are often incapable of achieving this task, and present a novel computational approach to denoise motion without explicit motion analysis. We demonstrate promis- ing experimental results on a set of challenging time-lapse sequences.
The three systems differ in the way that they observe embryos. In the Embryoscope, the tray holding the cul- ture dishes is under constant movement to bring each embryo individually into the field of view. When the tray is fully loaded (72 embryos), it takes 20 minutes until the next image of a given embryo is taken. This interval does not allow the embryologist to detect rapid changes accurately (e.g., S1 which should last <30-35 min). The constant movement, electromagnetic effects, heat and volatile organic compounds released from the lubricants related to this technology carry the potential to exert ad- verse effects though no such negative effect has been directly confirmed yet. However, this technology enables the system to maximize resolution. Each Primo Vision microscope is able to monitor up to 16 embryos at the same time without moving them. With this method, the embryos are cultured in a completely undisturbed envir- onment. This system requires significantly less frequent image acquisitions (because all 16 embryos are observed at the same time); hence, the exposure to light, electri- city and electromagnetic effects is even lower than that