TP Model Options

Workflow context
For the workflow context, see 📘 T-PROGS (borehole-derived geology) · 📘 Stochastic Modeling

Model Grid

Use the text fields to define the number of cells (Num Cells) in the x-, y-, and z-directions. The size of each model cell is automatically determined from the domain size and the number of grid cells. (NOTE: TP Model uses a structured grid where each model grid cell is the same size.

More >> Model Domain (plan view):

The values in the input fields are intended to be percentages.

DataXlimit=Data extent in X- (west-east) direction DataYlimit=Data extent in Y- (north-south) direction

Datamin=coodinate of origin of data bounding box (lower-lleft corner) Datamax=coordinate of corner opposite of origin

By default, Xmin=DataMin(X), Ymin=Datamin(Y), so on.

But if a non-zero percentage is used, it will shift Xmin, Ymin, etc. based on a percentage of the data extent in that particular direction.

Vertical Model Extent

Be default, the vertical boundaries of the TP model are determine by the data extremes (global maximum and minumum elevation found in the borehole dataset).

To change the top extent of the model to the maximum land elevation in the model area, select Max DEM (DEM = Digital Elevation Model). The maximum DEM is automatically extracted from the MAGNET Data Center for the area of interest.

Users can also use the minimum 'RockTop' elevation (interface between unconsolidated and consolidated deposits) from the MAGNET Data Center to define the bottom extent by selecting 'Min RockTop elevation.

Data quality / resolution depends on geographic location.

More >> Compression and Virtualization Options

Compression - In some cases, the horizontal extent of the model is much larger than the vertical extent (e.g., when modeling across a large area). In such cases, it may be necessary to compressthe data in the horizontal direction during simulation. The value in the text field embedded in phrase "If ratio X/Z is greated than ..." is used i) to determine if the ratio is large enough such that compression should occured; ii) as the compressoin ratio for compressing the horizontal data during numerical simulation.

The default value is a very large number (99999) so that compression does not occur unless the user changes it to a small number (e.g., 5, 10, or 15).

Virtualization - In some cases, multiple wells occur in one model cell. Users can "virtualize" or aggregate multiple wells into a single well with effective lithologies (material category), in order to reduce the number of wells. This aggregation method may be necessary to enable large-scale (e.g., countywide) geostatistical simulation.

Simulation Options

Seed - A random number for initiating the stochastic simulation process. The seed input affects the random drawing of a value from the probability distribution.

Number of realizations - The number of 3D model representations to simulate. If more than one realization is created, AquaNET will calculate and output an ensemble mean model by assigning the most frequently occurring material at each grid cell across different realizations.

Quenching - Quenching is an interative solution process that involves improving the match between the simulated spatial variability and the varaibility inferred from the data. Quenching is accomplished by looping through every cell in the realization and determining whether change in material type will improve the match between the data and simulated spatial variability; if so, the change is accepted.

The number can specify "how much" quenching should occur by:
1 - assigning a maximum number of quenching iterations (how many times to cycle through each cell in the realization to evaluate potential improvements)
2 - assigning a quenching tolerance (quantitative measure of an "acceptable" match between simulated varaibility and data) You can potentially improve accuracy of the simulation by increaseing the maximum number of quenching iterations and/or decreasing the quenching tolerance. Typically, default values are sufficient (especially when experimenting with other simulation parameters, e.g., compression ratio or length ratios.

Markov Chain Analysis

The Transition Probability approach uses a Markov Chain analysis to infer spatial patterns (transitions, material proportions, and typical lens/interval lengths)in the vertical (z-direction). The user can specify a lag spacing (spacing between measurement points) consistent with the data and desired resolution of the simulation (e.g., generally speaking, a smaller lag spacing should be used for shallow / short boreholes, and vice versa). The default lag spacing is 1ft (0.3048m).

Length Ratios

The Transition Probability approach offered in AquaNET does not include horizontal (x- and y-direction) Markov Chain analyses, so the user assumes a ratio of horizontal extent of a material to its vertical extent — or an anisotropy ratio — to create a 3D realization of the material distribution. This is done for each material type, for both the x-direction (West-East, W-E) and y-direction (N-S). The default length ratios for each material type is 10, but the user can change this value using the provided text fields to better reproduce known large-scale varaibility / geologic intuition.

TP Z-direction Ratio Results (after analysis)

This option shows the transition probability matrix of auto- and cross-correlations between the material types as a function of vertical lag spacing. Graphical depictions of the spatial correlations vs. vertical lag distance are generated and geostatistical models are fit to the data using Markov chain analysis. (NOTE: this option is not available until after completing TP simulation (1st "run").