.png){kind=icon}
.png){kind=icon}
Return
The Dhahran Probability Density Function for Improved Parameter Estimation
.png){kind=icon}
PRE
AI
DOI：10.1007/s13369-026-11432-6.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Reliable estimates of reservoir parameters are essential for proper reservoir management. One of the means to estimate reservoir parameters is through history matching. History matching involves finding the estimates of model parameters that ensure an acceptable match between the output of a model and the physical observations the model is meant to replicate. History matching is often posed as a minimization of an objective function and the most common objective function used in parameter estimation is the Euclidean norm (also called the $$l_{2}$$ –norm) of the model errors (i.e., the difference between the modeled response and the measured/observed response). Minimization of the Euclidean norm of the model errors is a direct consequence of the assumption that these errors follow a Gaussian distribution. However, this minimization is still fraught with several problems including the nonuniqueness of solutions and the inability to reproduce the true parameters of the system in most cases. We propose a new multivariate probability density function called the Dhahran probability density function (DPDF). The proposed probability density function is built by combining spatial deviation from a location parameter with angular deviation from the same parameter. The Dhahran probability density function was used as a replacement for the multivariate Laplacian and Gaussian probability density functions in model parameter estimation. Sample illustrations were presented to show that this probability distribution provides more reliable estimates of the model parameters than the Gaussian and Laplacian probability distributions.
Keywords:
Dhahran probability distribution
Parameter estimation
Inverse modeling
Angular deviation
Minkowski distance
Journal
A
IF:
2.9
Papers:
962
Citations:
0