BMEGUI Tutorial 2

Temporal BME

  1. Objective

The primary objective of this tutorial is to introduce the BMEGUI users to theBME analysis of soft data. In this exercise, we will use a purely temporaldataset, i.e. the data consist in a time series of soft data point collected at a unique geographical location. The analysis will consist in an exploratory analysis of the data across time, in the modeling of its temporal covariance, and in obtaining a plot showing the BME estimate of the data as a function of time.

  1. Install BMEGUI 2.1.1

See tutorial 1.

  1. Data

Download the data file “data02.csv” from the Tutorial Data Files and save it in a folder called “work02”. Open the data file using a spreadsheet editor (such as Excel, etc.) or a text editor (such as Notepad, WordPad, Word, etc.) to see the fields (i.e. columns) available in this data file. This file contains a field indicating the “Type” of the data, and two fields providing readings named “Val1” and “Val2”, respectively.

Most of the data in this data file areof type 0, corresponding to hard data with a value given by Val 1. However, note that at times T=2 and 3, the data is of type 1, corresponding to interval soft data with a lower and upper bounds given by Val1 and Val2, respectively. Similarly, note that at times T=8 and 9, the data is of type 2, corresponding to Gaussian soft data with a mean and standard deviation given by Val1 and Val2, respectively.

  1. Operation
  1. Launch ArcMap and add, if needed, the BMEGUI toolbox (see the BMEGUI 2.1.1 user’s manual for more details).
  1. Run the BMEGUI tool and select the following workspace and data file.
  2. Workspace: work02
  3. Data File: data02.csv
  1. Click on the “OK” button. The “Data Field” screen appears.
  1. In the “Data Field Setting” section, check the “Use Datatype” button and select the following column names from the dropdown menu of each field.
  • X Field: X
  • Y Field: Y
  • Time Field: T
  • ID: Automatic ID
  • Data Type: Type
  • Value1 Field: Val1
  • Value2 Field: Val2
  1. In the “Unit/Name” section, input the following units and name of data in each entry box.
  • Space Unit: deg.
  • Time Unit: days
  • Data Unit: ug/L
  • Name of Data: Contaminant B

Figure 1: The “Data Field” screen

  1. Click on the “Next” button. The “Data Distribution” screen appears
  1. Check the basic statistics (mean, standard deviation, coefficient of skewness, and coefficient of kurtosis) of the data and its log-transformed data in the “Statistics” section.
  1. Check on the histograms of raw data and the log-transformed data. By clicking the “Raw Data” and “Log Data” tab in the “Histogram” section, you can switch the histograms

Figure 2: The “Data Distribution” screen showing the Histogram of “Raw Data” (left) and “Log Data” (right)

  1. Click on the “Next” button. The “Exploratory Data Analysis” screen appears.
  1. Click on the “Temporal Evolution” tab and examine the temporal distribution of data

Figure 3: The “Temporal Evolution” tab in the “Exploratory Data Analysis” screen

  1. Click on the “Next” button. The “Mean Trend Analysis” screen appears.

Figure 4: The “Mean Trend Analysis” screen

  1. Since we skip the mean trend analysis in this exercise, click on the “Next” button. The “Space/Time Covariance Analysis” screen appears.
  1. Click on the “Temporal Component” tab. Then click on the “Edit Temporal Lags…”button in “Experimental Covariance” section. A dialogue box will appear to enter the temporal lags.
  1. Input the following values in “Temporal Lag” and “Temporal Lag Tolerance” fields of this dialog box.
  • Temporal Lag: 0.0,1.0,2.0,3.0,4.0,5.0
  • Temporal Lag Tolerance: 0.0,0.5,0.5,0.5,0.5,0.5
  1. Click on the “OK” button. The experimental covariance plot (shown in red dots) is automatically updated.

Figure 5: The “Space/Time Covariance Analysis” screen and the dialog box for modifying the lag size

  1. In the “Covariance Model” section, input “2” in the “Number of Covariance Structure” field.
  1. Input the following model parameters
  • Structure 1
  • Sill: 0.038279
  • Spatial Model: exponentialC
  • Spatial Range: 1
  • Temporal Model: gaussianC
  • Temporal Range: 2.5
  • Structure 2
  • Sill: 0.008
  • Spatial Model: exponentialC
  • Spatial Range: 1
  • Temporal Model: exponentialC
  • Temporal Range: 0.5
  1. Click on the “Plot Model” button. Theplot of covariance model is superimposed on top of the experimental covariance values.

Figure 6: The updated experimental covariance and covariance model

  1. Click on the “Next” button. The “BME Estimation” screen appears.
  1. Click on the “Temporal Distribution” tab. To calculate BME estimatesas a function of time at Station “1”, setthe following estimation parameters in “New Plot” section
  • BME Parameters:
  • Use default settings except for “Order”
  • Order: “Constant Mean”
  • Estimation Parameters:
  • Station ID: 1
  • Display Parameter:
  • Use default settings
  1. Click on the “Estimate” button. A new tab labeled“Plot ID: 0001” appears and anew entry appears on the list in the “Plot List” section.

Figure 7: The “BME Estimation” screen

  1. Click on the “Plot ID: 0001” tab and examinethe plot of the BMEestimates as a function of time.

Figure 8: Time series of the BME estimate

  1. Click on the “Close Tab” button. The tab labeled “Plot ID: 0001” closes.
  1. In “Plot List” tab, select the entry “0001”, then modify the “Scaling Factor” in “Display Parameter” section from 0.1 to 0.5.

Figure 9: The “BME Estimation” screen

  1. Click on the “Show” button in the “Plot List” section. BMEGUI redraws the plot with a new scaling factor. Click on the “Plot ID: 0001” tab to see the effect of new scaling factor

Figure 10: The effect of the “Scaling Factor”

  1. Click on the “Quit” button to close the screen. A dialog box appears. Click on the “OK” button of that dialog box to confirm that you want to quit BMEGUI.