Satellite Monitoring of Cyclic-Steam Stimulation Without Corner Reflectors
The authors discuss a new way of extracting deformation information from radar imagery, contributing to improved accuracy of InSAR surface-elevation monitoring.
Accurate monitoring of the ground deformation of a field under cyclic steam stimulation (CSS) can be used to calibrate predictive models, show the effectiveness of steam injection, and demonstrate the effect of the recovery operation on the surface elevation. Interferometric synthetic aperture radar (InSAR) uses radar returns from the ground to calculate highly precise estimates of the ground change. The authors discuss a new way of extracting deformation information from radar imagery, contributing to improved accuracy of InSAR surface-elevation monitoring.
The challenge in the Canadian oil sands is to achieve robust measurements with limited infrastructure. Essentially, the problem is to develop robust measurements of the ground movement without reliance on buildings, pipelines, or even installed targets [such as corner reflectors (CRs)]. The case study described here shows exceptional results for monitoring of regions without traditional point targets.
The operations and mechanism of thermal recovery of steam-assisted gravity drainage (SAGD) and CSS differ. SAGD is a continuous process of balanced injection and production, while CSS is a cyclic process of injection, soaking, and depletion. The methodology provided depends on an understanding of the expected ground movement but does not rely on foreknowledge of the motion itself. The rate of change in ground deformation associated with a CSS operation is key to the extraction of the signal (deformation) over the time required for this monitoring. The relatively short time scales and the high magnitude of the changes experienced in CSS are well-suited to the nontraditional methods described here.
The methods are validated by measurements of surface elevation by use of real-time kinematic (RTK) global-positioning-system (GPS) surveys acquired in the Primrose field coincident to the InSAR observations. The net-over-injection (NOI) metric provided by Canadian Natural Resources Limited (CNRL) is also used to demonstrate the causal relationship between injection and production volumes and the observed ground heave and subsidence, respectively.
Satellite radar achieves ground imaging by sending microwave energy from the satellite to be reradiated from the Earth’s surface and captured as a wave by the satellite. Because the energy travels at the speed of light, the process of sending a pulse from a satellite at 800 km takes only a few milliseconds. The speed at which the satellite travels means that large areas of ground may be imaged almost instantaneously. The microwave imagery provides amplitude and phase information about the radiation returned to the satellite. The amplitude variations can be described as a measurement of the texture of the Earth surface. Key to the capture of ground movement is the precise measurement of the time that the radiation travels. This measurement is accomplished by the displacement between two waves (the phase). Capturing the phase from the satellite at different times provides a measurement that has submillimeter precision under optimal measurement conditions. The process of measuring the differential phase is called interferometry.
Fig. 1 shows the geometry of the measurement made by the satellite. The line labeled LOS (line of sight) shows the typical path of the microwave radiation from the sensor to the ground (and its return path). The measurement of ground deformation is made along this path. In the case of a single-LOS observation, the ground movement can be known only along this projection. Multiple-LOS measurements can be used to independently construct the movement of the ground in both a lateral and a vertical sense.
Imaging. Data collection over the region has been ongoing since 2008 and has been conducted primarily with the RADARSAT-2 satellite. The primary mode of observation has been with the so-called MultiLook Fine beam of the satellite. This beam provides a ground resolution of approximately 8 m and a swath width of 50 km. In December of 2014, additional imaging was initiated with multiple RADARSAT-2 Wide UltraFine beam modes, providing a nominal ground resolution of 3 m and a similar swath width of 50 km. Further details on the satellite and its instrumentation are provided in the complete paper.
Increased Accuracy in Spatial Estimation of Deformation. Recent advances in the effectiveness of InSAR analysis have typically extended the characterization of point sources and neighborhoods of stationary phase information. These methods are most effective when the region to be examined includes objects that can be expected to provide a consistent phase return over time. CRs play an important role in accurately deriving the movement over an SAGD operation. The installation of CRs is required to measure small deformation signals quickly and accurately.
CRs, however, have some significant drawbacks. The installation of CRs requires hardware expense, tree cutting, and dispensations. The drawbacks of CRs are avoided in a CSS operation with a very different surface-movement pattern. The problem for a CSS operation is the speed at which the ground is moving. However, this same factor can also serve as a benefit.
If CRs are used to categorize the movement over an area of interest, then one must consider both the temporal and the spatial unwrapping of the interferometric signal. Unwrapping is required because the phase difference is known only within a circle. The total separation between two waves needs to be unwrapped with reference to the wavelength. In essence, the slope (in either time or space) between two CR observations must be small enough to be unambiguously observed at the pair of CRs. The two CRs must be close enough to experience no more than the maximum deformation. When 25 cm of ground movement over a 1-km×500-m region can be expected in less than 1 month, the required density of observations quickly becomes impossible.
The speed of the deformation, however, is also of benefit, as previously mentioned. That the amount of deformation is on the order of centimeters per observation means that the significant noise factor (atmospheric noise) is much smaller than the expected measurement. The requirement then is to understand the effects of the ground conditions on the signal/noise ratio available from the interferometric pair and extrapolate the motion that can be found in the observation. Figs. 2 and 3 of the complete paper provide information about the methodology of extrapolating ground deformation and coherence.
CNRL provided measurements of the ground deformation from six locations on Pad 43 in the Primrose Field. Measurements of surface elevation by RTK GPS surveys have been taken since 22 October 2014 and have a reported accuracy of 2 cm.
The RTK GPS observations were made on wellhead flanges of two observation wells and the pile caps of four light poles on Pad 43; the coincident InSAR observations are available contiguously throughout multiple drainage areas. The graphic in Fig. 4 of the complete paper was constructed by taking an average of the RTK GPS measurements at the six locations. The InSAR measurements are averaged over a 100×100-m area centered on Pad 43. The excellent agreement between the two instruments is confirmation that they are both measuring the same signal.
There are two periods of missing observations from the InSAR measurements (notably March to April 2015) and a period of missing observations from the RTK GPS (July to October 2015). The first period of missing observations for the InSAR was caused by snow and ice cover from 13 November 2014 to 7 December 2014. The second period of missing observations for the InSAR was caused by spring thaw from 2 March 2015 to 19 April 2015. InSAR results were obtained by processing SAR images captured from 2 September 2014 to 28 October 2015. No RTK GPS surveys were taken between 30 June 2015 and 27 September 2015.
In the periods during which both systems were in operation, the agreement is clear. The nature of the InSAR measurement as a difference between one observation and the next makes periods of missing observations problematic. In this case, the signal from the InSAR has been rebalanced to the RTK GPS signal by estimating the GPS deformation at the restart of the InSAR time series.
Further evidence of the accuracy of the InSAR method described here for CSS operations comes from the comparison of the NOI and the cumulative deformation. Volume above fill-up (VAF) is defined as the cumulative volume of steam injected in a steam cycle after the reservoir pressure has reached the vertical in-situ stress. NOI is defined as injected steam VAF minus the cumulative volume of liquid produced. Increasing NOI occurs during steam injection, and a decreasing NOI occurs during production.
The relationship between the NOI and the actual surface deformation as direct and spatial is clearly validated. The differences between the NOI and ground deformation for a particular well may be related to arching deformation of the overburden and variations in the stratigraphy, overburden thickness, and steam injection or production distribution along the length of horizontal wells. Nevertheless, the surface deformation occurs where predicted by the NOI information.
This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper SPE 184971, “Satellite Monitoring of Cyclic-Steam Stimulation Without Corner Reflectors,” by Michael D. Henschel and Jonathan Dudley, MDA, and Peter Chung, Canadian Natural Resources Limited, prepared for the 2017 SPE Canada Heavy Oil Technical Conference, Calgary, 15–16 February. The paper has not been peer reviewed.