Exploration and production operations associated with oil and gas mining are moving further offshore into deeper water. Drilling operations routinely take place now in water depths over 3,000 ft, and water depth drilling records exceed 10,000 ft. The prevailing oceanographic conditions in this ultradeep water environment are severe with large waves, strong currents, and high winds. Estimating environmental design criteria for exploration and production platforms is a common problem. The resilience and efficiency of offshore structures and operations depends upon development of reliable design criteria; yet, there is uncertainty with characterizing the oceanographic environment in areas lacking sitespecific data and experience. Uncertainty is typically balanced in the engineering process by applying conservative factors of safety; however, overly conservative assumptions can result in extremely expensive design packages that may not be supported by the economics of the oil and gas field. Likewise, underestimating design criteria can be catastrophic, resulting in unexpected damage to structures, as well as environmental, health, and safety concerns. Developing reliable and reasonable design criteria is an important topic for a broad variety of maritime applications, and one with which Woods Hole Group has valuable experience. Major criteria include operational and extremal statistics for winds, waves, and currents used to estimate forces acting on an offshore structure.
Woods Hole Group’s approach to the design problem helps achieve more accurate results, while minimizing uncertainty. The method is based on understanding the physical processes at work, and separating the different physical processes represented by the data (see flow chart). A probability distribution function is estimated separately for each physical processes, so that an independent statistical estimate of operational and extreme winds, waves and currents can be obtained. In deep water, extreme and/or operational current estimates may also be separated for different depth regimes. The separate estimates are then combined using joint probability of occurrence statistics. An advantage of this technique is a reduction in the uncertainty of the final estimate since the periods of extrapolation for each process are reduced. The result is a set of design criteria that avoids being overly conservative, and that make physical sense. For instance, the set of physical conditions that produce the largest waves and/or surface currents often differ from the conditions that generate the swiftest currents at depth. Traditional statistical analysis may produce design conditions that assume all processes occur together, which may not be physically possible. The Woods Hole Group approach helps ensure the design criteria make physical sense, improves the reliability of the statistical estimates, and maintains sensitivity to economic implications.
Our understanding of ocean dynamics is based on observations, theory, and numerical modeling. Although the science and technology behind these approaches to solving the problem is advancing rapidly, none of the three approaches is sufficient alone. Theory is based on assumptions; numerical simulations are approximate and based on the existing theory; and observations, unfortunately, always are limited by gaps in time and space. Bridging data gaps in a defensible manner requires knowledge of motion scales and kinematic structure. Processoriented analyses for a specific location allows the results to be described in the context of regional features, and provides an improved understanding of regional oceanography.
One major challenge is the insufficient length of time series data used for estimating extreme events. For example, it is often ‘standard’ practice to use a oneyear dataset to estimate characteristics of a 100year event (i.e., 1% chance of occurring any given year). The observed data usually represent “typical” events (i.e., those having low amplitude but high probability of occurrence); therefore, extrapolation to largeramplitude events of much lower probability can be done only when the underlying physical processes are statistically stationary, and if the shape of the probability function can be estimated with confidence. Analysis of short time series must receive special attention.
Forecasting extremal currents, for example, is difficult because ocean currents can vary on a wide range of time scales, including very long time scales (years to decades or longer in some places). Consequently, timeseries current measurements are unlikely to resolve the full range of natural variability. Ocean currents are the sum of currents driven by diverse forces, which may have widely differing probabilities of occurrence. While tidal currents peak on a daily time scale, low frequency currents may have only a few peaks in an annual time series. Particular care must be used when applying statistical extrapolation techniques to relatively short observational data sets, since just one or two extremes that rise significantly above the rest of the peaks may greatly affect the form of the distribution function, and result in substantial errors when extrapolated to long return periods.
Overall, developing reliable design criteria depends upon strong knowledge of prevailing oceanographic processes, sound science and statistics, and responsible professional judgment.

Leonid Ivanov, Ph.D., Physical Oceanographer
Dr. Ivanov has more than 20 years experience worldwide in open ocean and coastal oceanography. Dr. Ivanov has conducted field studies and research in the eastern Tropical Atlantic focusing on the investigation of physical mechanisms driving seasonal and mesoscale ocean variability, as well as baroclinic and barotropic tides, inertiagravity waves and coastal fronts off West Africa.

