27
Estimating Mature Plays
a play. If this problem can be overcome, then the estimation of population mean, variance, and correlation among variables can be achieved.
The objective of this chapter is to explain the characterization of the
discovery process by statistical formulation.
To account for sampling bias, Kaufman et al. (1975) and Barouch and
Kaufman (1977) used the successive sampling process of the superpopulation probabilistic model (discovery process model) to estimate the
mean and variance of a given play. Here we shall discuss how to use superpopulation probabilistic models to estimate pool-size distribution.
The models to be discussed include the lognormal (LDSCV), nonparametric (NDSCV), lognormal/nonparametric–Poisson (BDSCV),
and the bivariate lognormal, multivariate (MDSCV) discovery
process methods. Their background, applications, and limitations
will be illustrated by using play data sets from the Western Canada
Sedimentary Basin as well as simulated populations. The steps for estimating undiscovered resources for a mature play involve (1) identifying a play, (2) compiling the data, (3) estimating pool-size distribution
and number-of-pools distribution, (4) estimating pool-size-by-rank, (5)
estimating play resource and play potential distribution, and (6) conducting feedback.
The superpopulation models do not require prior values for the total
number-of-pools, population parameters, exploration effi ciency, or
truncation of large values. However, BDSCV requires a prior Poisson
distribution for the number of pools and the lognormal pool-size distribution for estimating the posterior number-of-pools distribution.
LDSCV requires a lognormal pool-size distribution, and MDSCV also
requires a multivariate lognormal distribution for the reservoir parameters and a bivariate lognormal oil and gas pool-size distribution.
All available data are used to estimate population mean and variance, because an adequate estimate of population variance cannot be
derived from truncated data. Furthermore, the procedure requires
estimation of the population, rather than the fi tting of a distribution to
the discovery sequence.
The Superpopulation Model
Lognormal Discovery Process Model
In the superpopulation approach, the key step is to estimate the parameters of the underlying superpopulation distribution from samples
Estimating Mature Plays
a play. If this problem can be overcome, then the estimation of population mean, variance, and correlation among variables can be achieved.
The objective of this chapter is to explain the characterization of the
discovery process by statistical formulation.
To account for sampling bias, Kaufman et al. (1975) and Barouch and
Kaufman (1977) used the successive sampling process of the superpopulation probabilistic model (discovery process model) to estimate the
mean and variance of a given play. Here we shall discuss how to use superpopulation probabilistic models to estimate pool-size distribution.
The models to be discussed include the lognormal (LDSCV), nonparametric (NDSCV), lognormal/nonparametric–Poisson (BDSCV),
and the bivariate lognormal, multivariate (MDSCV) discovery
process methods. Their background, applications, and limitations
will be illustrated by using play data sets from the Western Canada
Sedimentary Basin as well as simulated populations. The steps for estimating undiscovered resources for a mature play involve (1) identifying a play, (2) compiling the data, (3) estimating pool-size distribution
and number-of-pools distribution, (4) estimating pool-size-by-rank, (5)
estimating play resource and play potential distribution, and (6) conducting feedback.
The superpopulation models do not require prior values for the total
number-of-pools, population parameters, exploration effi ciency, or
truncation of large values. However, BDSCV requires a prior Poisson
distribution for the number of pools and the lognormal pool-size distribution for estimating the posterior number-of-pools distribution.
LDSCV requires a lognormal pool-size distribution, and MDSCV also
requires a multivariate lognormal distribution for the reservoir parameters and a bivariate lognormal oil and gas pool-size distribution.
All available data are used to estimate population mean and variance, because an adequate estimate of population variance cannot be
derived from truncated data. Furthermore, the procedure requires
estimation of the population, rather than the fi tting of a distribution to
the discovery sequence.
The Superpopulation Model
Lognormal Discovery Process Model
In the superpopulation approach, the key step is to estimate the parameters of the underlying superpopulation distribution from samples