USING GEOSOFT OASIS MONTAJ AND CHIMERA - Bert De Waele (for IMC Consulting) 1 8/07/2005 USING GEOSOFT OASIS MONTAJ AND CHIMERA Course developed for IMC Group Consulting Limited To

©Bert De Waele (for IMC Consulting) 8/07/2005 1

USING GEOSOFT OASIS MONTAJ AND CHIMERA

Course developed for IMC Group Consulting Limited To be lectured at Nouakchott

October 2003


TABLE OF CONTENTS

PREPARATIONS 1

STARTING GEOSOFT 1

CREATING A DATABASE 2

IMPORTING DATA FROM EXCEL (.CSV FILE) 2

WORKING WITH GEOSOFT DATABASES 4

MERGING LOCATION AND VALUES DATABASES BASED ON COMMON SAMPLE CHANNEL 5

PRELIMINARY CHECKS 7

REARRANGING THE DATA INTO LINES 7

WORKING WITH LINES 8

CREATING A MAP 10

PLOTTING SAMPLE LOCALITIES 14

EXAMINING THE DATABASE IN CONJUNCTION WITH THE MAP 15

COLOUR RANGE AND PROPORTIONAL SIZE PLOTS 17

MINIMUM CURVATURE GRIDDING 19

KRIGING 23

CONTOURING 25

HISTOGRAM ANALYSIS 26

ADVANCED MAPPING TECHNIQUES 28

MASKING DATA (CHEMMASK) BASED ON CODE 29

MASKING DATA USING DRAWING TOOLS 30


USING MATHEMATICAL EXPRESSIONS 31

THE MODIFIED BOXPLOT 32

ADVANCED USE OF THE HISTOGRAM 32

CORRELATION ANALYSIS 34

SCATTER ANALYSIS 34

THE GEOSOFT HELP SYSTEM 35


Preparations Before starting to work on a project, it is necessary to create a location on the local hard drive in which all data, related to the project, will be stored. A new directory can be created on the drive, and the necessary datafiles can be copied into that directory. It is good practice not to work from a location on the harddrive that contains your vital information…make a copy instead to safeguard vital data. Assume we had created a directory on the root of the C drive, which contains all project related information, and called it C:\Mauretania. We can create a subdirectory in that, and call it “season#1”. We will prepare our raw data (excel sheets and/or access databases), and copy the necessary files into that workspace for processing in Geosoft. Although Geosoft’s import facilities include the possibility to import directly from Excel and Access, we will prepare Comma Separated Values files (.CSV) with excel, which is the preferred import format, and avoids complications with formatting issues. We will save csv files for standards, repeat analyses, assays and locational data into the C:\Mauretania\season#1\ directory… Starting Geosoft Double click on the icon and the Geosoft programme window comes up, with all but a few options dimmed out:

Click on File…New workspace

Browse to C:\Mauretania\season#1 and type in the name of the new workspace (season#1). Geosoft now creates a “season#1.gws” file that contains all information pertinent to out work on this project. Whenever we change something in our project (add/modify databases, add maps, …) Geosoft will automatically reflect that change into this file…so that we can pick up from where we left off by loading the .gws file. The menu bar will now look a bit different as new functionality becomes available:

At this stage we also can load other extra menus we will need later in the program. These menus are not automatically loaded by Geosoft to save time. We find the menus under the GX pop down window, Load menus…

The following dialog box appears:


We can simultaneously select Chimera.omn; ChimeraDPA.omn and oasis.omn by holding the ctrl key. The menu bar will now look like this:

Creating a database The first thing we’ll need to do, is to create a database in which we can import our data. All data can be placed into one single database, or we can create separate databases. Let’s start with one database. Click on Data…New database…:

Geosoft databases are engineered to deal with geophysical and geochemical data. Such data, especially airborne geophysics, are often collected along a grid, where sampling happens sequentially along a line. The database structure reflects this, and the terminology used in Geosoft databases can be compared to Excel as follows: Geosoft Excel Lines/Groups Sheets Channels/Fields Columns

In other words, each Geosoft Line actually represent a set of data collected along a physical line, and is kept on one sheet in the database. Each channel represents a collected value field for each station (e.g. x, y, total magnetics, radiometrics…) and is kept on a column within each sheet. Each row in the database represents the results for one sampling station. In the dialog box, we can fill out the name for our database, and we can leave the recommended setting for Lines and Channels as is… These settings can be changed later in the Database maintenance commands… After creating this (empty) database, the screen looks like this:

What we are looking at is an empty sheet (line 0:0). The values below L0:0 are called fiducials, and will be automatically generated for each line with increment 1.0. These values cannot be changed manually, and are controlled by the database. The way fiducials are attributed can be changed through the Data…Fiducials… command. Importing data from Excel (.CSV file) All database related utilities reside under the data menu:


We will select Data…Import…Ascii… to import the CSV files…

Geosoft allows the creation of templates to automate import rules. This is handy if repeat imports are anticipated of identically structured input files. We will go through the Wizard to have control over the input parameters every step of the way. Browse to the file “All_data_season#1_loc.csv” and click on the Wizard button.

We can already see a preview of the data contained in the file. Select Delimited on the radio buttons, as a

csv file has a comma delimited structure.

Select the Microsoft Excel CSV radio button and hit Next…

The last step of the Wizard allows us to define the data-type for each column. We can click on the first column, which holds the sample numbers. The column highlights, and we can select String as data type. If we want, we can change the Channel (column) name and Label (the way the database displays the column header). Dummy indicates which values were entered in the csv file to denote either missing data or values below Minimum Detection Limit (MDL). After import, the database looks as follows:


We can now similarly create a database called “season#1_assay” with the assay values and laboratory name, “season#1_duplicates” with all duplicate analyses, and “season#1_rocks” with all rock assays. Working with Geosoft databases Let’s go over a few basics when working with Geosoft Databases. As explained before, the structure of the database assumes the data to be in lines, where each sheet contains line data, and each column contains the values collected along a line (typically X, Y, Z and measurements or assays). When working with randomly collected data (e.g. stream sediments or randomly collected ground data), only one line will be created with an index D, signifying random data. Data collected along a line will be indexed with L (line data). In case the database is collected along a linear grid, it may be useful and/or necessary to use tools to split the data into lines either based on a Line channel (column), or by letting Geosoft look for X-Y breaks. We will perform these operation later

on when all databases have been merged with our locational data. Once the database is organised into lines and channels, the data can be inspected in various ways. First note that the column (channel) header has a black triangle in the upper left corner. This indicates that the channel is protected. This is a very useful characteristic of Geosoft, in that it tries to avoid accidental modifications. We can unprotect and protect a channel by right-clicking the header:

Note that Geosoft allows us to protect all, or none. That means that the protection status is normally only changed for the current channel of the current line. Lines below or above are unaffected unless one chooses protect all/none. If you left-click the column header, the channel gets selected. The


first click selects only the channel of the active line, a second click selects this particular channel for all the lines (header white=only current line; header black=channel in all lines). We will get back to this feature later. Protection of data has also led to the inability to select more than one channel at a given time, again to avoid accidental modifications. This makes Geosoft secure, but also takes away some of the powerful tools a normal spreadsheet can offer. It is therefore advised to prepare the data as best as possible in Excel and/or Access before importing into Geosoft, so that not too many changes need to be done once the Geosoft database is created. Notice that it is not directly possible to change the order (from left to right) of appearance of the channels. To rearrange the channels, it is necessary to remove them all from the sheet, and then display them again, one by one, in the correct channel. If you right-click on the “Nr” channel, and then choose “Remove all”, all columns disappear. Note that the actual data is not lost, but we only changed the way the data is displayed (or not). If we now right-click on the leftmost channel, we can choose “List…”. A listing of available, undisplayed channels appears, from which we can now choose “Nr” to be displayed in the leftmost channel.

We can do the same to display from left to right next to the Nr channel, X, Y, Au_ppb and Lab.

Deleting a channel can also be done using the right-click menu on a channel header. Deleting a channel actually erases the selected channel irretrievably from the database. The operation will not be allowed if the selected channel is protected. It is advised to leave all original data channels in protected mode and copy their contents to a new working channel for manipulation. This avoids accidental loss of data… Merging Location and Values databases based on common Sample Channel After importing the various databases, the first thing that needs to be done is to merge the survey data (x and y location) to the sample assays, rock assays and duplicate databases, based on the sample numbers we used. For the merge procedure, Geosoft assumes that there is a channel (column) called “Sample”, in all databases, and will use that as the common link to merge the data. In our case, the channel “Sample” does not exist, and is instead named “Nr”. We will change the channel name “Nr” into “Sample” for all databases. Geosoft also expects the location database to contain a channel called “X” and another “Y”. In our case, these channels are called “X_UTM” and Y_UTM”. We’ll rename these to “X” and “Y”. Open the “season#1_assay” database, and right-click on the column header “Nr”. A pop-up menu appears as follows:


Select Edit… and the following dialogue box appears:

We can change the channel name into “Sample”. Note that some other settings can be adapted here as well, including the data-type.

We should change the name of the channel “Nr” into “Sample” in all databases. We can now also open the “season#1_loc” database and change the “X_UTM” channel in “X” and the “Y_UTM” channel in “Y”. The “season#1_rock” database had been prepared with a sample column, and columns X and Y, so no modifications are necessary… We are now ready to merge the location data with the assays and other attributes. Click on the Chemimport… menu and choose Merge and Verify…:


We browse and select the assay database, and the location database, and request for a report after merging. After clicking OK, Geosoft merges the databases, using the Sample channel as common denominator, and shows a report of the operation. If at this stage, a lot of errors are reported for missing data, duplicates and other problems, it is necessary to carefully look at the reported errors, and correct the problems if possible. The “season#1_assay” database has now been updated, and all assays now


have an X and Y channel with location data. We can now do the same for any other database that has no location data… Preliminary Checks Before we get into the actual data processing of the geochemical database, we can perform a few routine data integrity checks, so that if any problems exist, they are dealt with immediately. The data in a database quite commonly has wrongfully entered (anomalous) data, which could either be typing errors in the X or Y channel, or mistyped data in other channels. In the locational data, these errors lead to significant problems when attempting to plot data, or calculating grids, as typing errors can often introduce points that lie well outside the area actually sampled. Likewise, mistakes in the assay channels can lead to non-existent anomalies, and significantly skew statistical properties of a dataset. One handy tool to check the data for possible outliers is the “profile” tool. Geosoft allows the visual inspection of data along a profile line. Channel values are plotted on the Y-axis, while stations are plotted along an X-axis. We’ll plot the X, Y and Au_ppb channels on a profile, so that abnormally high or low values immediately become apparent. Right-click on the header of the X-channel of the assay database and select “show profile…”. The bottom of the window now opens up to display the profile of the channel. The semi-regular nature of the profile line indicates that no serious errors exist in the X channel. We can display a second profile for the Y channel

Again, the trend of the line indicates that also the Y-data is not encumbered by serious errors. Rearranging the data into Lines The step-like nature of both profile lines demonstrates that our database was collected along lines. We’ll attempt to split the database into its respective lines, using the X,Y breaks. Once we have the database split into its lines, we can save it as “season#1_assay_lines.gdb” First, we’ll have to expand the database, which in its present state can only hold 50 lines. The data was collected along more lines than that, and we’ll grow the database to be able to hold a maximum of 1000 lines. Go to “Data…Maintenance…Grow…”.


Choose the “season#1_assay” database in the dialog box, and answer YES when notified you should not interrupt the process. Now fill in 1000 lines, and leave the 50 channels limit as is…

The database is now expanded to have room for up to 1000 lines. We can now split the data along line breaks as follows: Go to the “Utility” menu, and select “Split Line…Split line on X,Y breaks…”.

Based on prior knowledge of the approximate line spacing and spacing between stations, we can fill out the dialog box as follows:

Geosoft now looks for the breaks according to these tolerances, and

transfers the data to their corresponding lines… As soon as Geosoft has finished, we can go to “Data…Save database as…” and fill out the new name as “season#1_assay_lines.gdb”, and click OK. The database gets saved, and immediately opens up in a new window. Working with Lines The new database now has the same data organised into lines, one line per sheet. The line currently in view is indicated in the leftmost column. The mark next to it indicates that it is also selected (highlighted), meaning that operations carried out on the database will include this line. By right-clicking on the Line header we can change the selection (default is all lines) by using the selection tool.

Other selection options include selecting directly by line number, selecting by flight number (the number after the colon in each line, in our case all flight numbers are 0), by direction or by line type. The selection tool is an interactive tool, in which we manually select a set of lines:


It is important to have all lines selected when working with the database, as results of statistical operations and queries will only be valid for the selected Lines. Another way to rapidly step through the different lines is by using the toolbar buttons:

The blue triangle up/down moves up or down to the first or last line. The red triangles move up or down one at a time… We can display the profile of the “Au_ppb” channel on the database, and then move down line per line to see the gold values plotted along each sampling line.

This is a very powerful method to see the trends of the values along the line. If you want to see the actual value in the database that corresponds to a “peak” in the profile, you can click on the peak, and the corresponding value

in the database is automatically highlighted. This also works in the other direction, so that selecting a point in the database will show where the point plots on the profile. We’ll see more interactive linking of database values with profiles, maps or even statistical plots such as scatterplots, later on. Lets explore the statistical tools available in Geosoft. As mentioned earlier, we can select either an entire channel of a database (across lines) or a channel within the current line. With the “season#1_assay_lines” database open, click on the “Au_ppb” header twice. This select the Au_ppb channel for the line we are in. We can now right-click and choose Statistics…

The following window comes up with basic statistics. The most important values for us are the total number of items (in the line we had selected), the number of Dummy values, the Minimum and Maximum:

Notice that the minimum is actually 0, because all assays below Detection Limit (< 5ppb) were made 0. We can change that background to the actual Detection Limit later on.


If we click OK and now select the same channel across the entire database we can call up statistics for all data in that Channel.

Creating a map We will now prepare a blank base map, which will serve as a template for all our maps of the Oassat-Sfariates area. Geosoft can work with two different layout versions of a map. One is called “FIGURE” and is a more simple way of presenting data, the other is called “MAP”, and is a more professional version for full scale plotting of final maps. We will plot all our small maps (A4) into a figure. Creating a map can either be based on a database, where Geosoft scans the X,Y limits of the data to produce a map, or can be done on the basis of a grid. We only have a database at the moment, and will use that as a base for the figure. Click on “Map…New map…”.

Click on the button “Scan data”, and Geosoft scans the data in the open

database and returns with the X and Y limits. Click Next…

Fill out the name of the map as “base.map”, size as portrait A4 and click on the Scale button. Geosoft calculates the scale that fits the map onto A4 portrait paper, and reports a number 1,075,425 as possible scale. To make that a round value we will change that number to 1,250,000. We can create a basemap by using the menu “Mapping…Base map…Draw basemap…”

We get the following dialog box:

We can leave the defaults and click next:


Here we change the Reference grid option to crosses, change the spacing to 5000m and leave the rest as default

In the last dialog box, we can fill out the titles for the map. When we click “Finish” we get the following map:

It appears that the grid spacing chosen (5000m) is too close, so we can go back through the process, and change that particular setting to 20,000 m. Notice that Geosoft remembers all our previous entries, which makes modifications of the map on a trial and error basis a viable option… After the change to 20,000m grid spacing, the map will now look as follows:

We will now manually change the layout a little, so that the scale bar does not overlap with the titles… Notice that with the map as active window, a menu becomes available at the right edge of the screen:


View/Group Manager Layer selection Group selection Shadow cursor buttons Pan tool Zoom tools: Interactive zoom Zoom Shrink 50% Last view Redraw map Zoom to full extent Map selection tools

A Geosoft layout is organised into two layers (VIEWS), onto which “Groups” of data and objects can be arranged. The BASE view is the map base we just created, and can contain other objects that form part of the basic layout (e.g. additional textboxes, logo of the company etc…). The DATA view contains the data we want to plot into a particular map. When we click on the topmost button “View/Group Manager”, we get the following window:

Notice that the Base view, which contains Titles, North Arrow, Scale bar and Surround sits on top, while the Data view, which only holds a layer called Coordinates, sits below. In fact the Base view is always displayed BELOW the Data view. The View/Group manager is a convenient way to organise our data layers. We can simply click a layer, and move it in front or to the back either step by step, or move it completely down or up. We can also select a Group and start editing it, hide it or simply delete it. Another way of selecting Groups is directly using the Group selection button. We’ll select the north arrow, resize it and move it on top of the scalebar: Click the “Group Select” button. We can now select the Title and move it (drag and drop) closer to the surround of our map:

Once satisfied, we can also lock it’s position by using the right-click menu: Status…Movable…


We can also use this menu to hide a selected Group, to select and delete a Group, and to perform various zoom functions. Via the Windows clipboard, it is also possible to copy (or cut) and paste Groups from one map to another. Now move the scalebar just below the Title, and resize the northarrow, and move it on top of the map in the lower right corner. The map should now look like this:

The last thing we need to change is the text of the Title. Geosoft, which is an Anglophonic programme, and does not use the same fonts as windows, cannot directly produce typically French fonts

like é, ë or ê. To get such text into Geosoft we’ll type it into Word, copy and paste it into a new text box… Open Word and type : “Première Campagne d’échantillonnage du sol”. Then select this text, and use COPY. Now, in the Geosoft map, right-click the Title and select “Edit this Group” from the menu.

Now select the text we want to change and delete it. Then click the text button

and click below the second line of the title. A new textbox is created, and we can now paste the text into it.

To end editing we can right-click on the Title Group and select “End editing” from the pop-up window:


We will now save the changes to the map through the menus: Map…Save changes…

Plotting sample localities It would be nice at this stage to take a look at the actual distribution of samples across the area. We’ll duplicate our base map, and plot sample localities with sample numbers into the new map: First go to the menu “Map…Duplicate map…”

We’ll fill the new name of the duplicate map as “Sample_localities_A4.map”. Also make sure to select “Copy current contents” so that we get an exact duplicate of the map.

The new map opens, and we can close the original Base map. Firstly we’ll need to change the title of the map a bit to be as follows:

The best way to do that is to first move the scalebar a bit down to make room for the extra line of text, then to start editing the Title Group (right-click menu), select, copy and paste the third line, move the pasted version a bit down, and retype that line to say “Localisation d’échantillons”. With care you can re-use the “échantillon” of the original text (avoiding a more


complicated procedure to recreate the accented letter é. Now we are ready to plot the samples. We’ll use the Chemmap menu, which apart from general mapping tools available in Geosoft, also includes the Chimera extension commands:

Choose Chemmap…Point plots…Location plot…

Leave Mask Channel as “None”, change the symbol to “dot”, size to 1mm, and leave other options as default.

We can now zoom into the central area for a closer look. Notice that a zoom selection requires a window, which we can then move across to the area to be enlarged… Examining the database in conjunction with the map After choosing the area we need to look at closer, we can arrange the database to be shown on a tile next to the map (i.e. having both windows open, tiled next to each other:

What we can do now, is to first display the profile line below the database, and then link up database and map so that when selecting a value in the database, the cursor on the map moves to the selected point. Likewise, we’d like to


see that selecting a point on the map, moves the highlight in the database to the point and line selected. We use the “Shadow cursor” and “Shadow cursor with datalink” buttons. The profile will allow easy recognition of higher values, so that finding them in the database is made easy… First click on the map to make it the active window. Then click on the Shadow cursor button . After doing that, when selecting a value in the database, the shadow cursor on the map will move to the point selected. We can now easily move through a line in the database, inspecting the location of each sample selected on the map. To activate the datalink, we can click on the “Shadow cursor with datalink” button and get the following window:

We can select the Data Group we want the cursor to link to, which is a handy tool if we were to plot different datasets, from different databases onto the map. We can select “SYMB_season#1_assay_lines”, and now have an active link between the data points on the map and the database. We can now post the sample numbers next to the data points on the map, to produce a sample locality map.

Go to Chemmap…Point plots…Post assay values…

We get the following dialog box:

Choose Sample as the data channel to be posted, apply No Mask Channel, and change the posting size to 1mm. Leave other values as is. The sample numbers are now posted on top of the map, and make up a new data layer:


When zooming into an area of the map, we can see that the posting size was too large, so that labels overlap. We can change these setting by running through the procedure again, and changing the values in the dialog box.

Notice that in the View/Group manager, there is still only one Group with Posted values, in other words, anytime we change the setting, the same Group is adapted. If we want to keep the Group displayed, while we create another Post, we need to rename the Group in the Manager, so that it is not automatically overwritten when posting another set of values. This is important when posting the channel repeatedly, while a Chemmask is applied (we’ll see that later). Posting another channel creates a new Post Group. We can post the Au_ppb values to the “Left Centered” with an X offset of –0.1mm. The Au_ppb values will be posted in a separate Group

Colour range and proportional size plots Another type of plot, which conveys to a certain extent the values of the data, without having to post the actual numbers, is the colour range or the proportional size plot. In a colour range plot, we can set threshold values, and plot samples between these threshold classes in different colours. Alternatively, we can plot higher values as larger symbols, using thresholds and symbol size parameters. Let’s first save the changes to our sample location map, and open the base map again to make a duplicate called “Au_colour_range_A4.map”

We’ll now first have to change the title a bit:

Remember that we’ll have to move the scalebar first, then edit the Title box, copy the third line, move the copied line down and change the text to what we want… There is an accent again, so that copying of the existing line is the fastest way … Then we can access the Colour ranges plot tool in “Chemmap…Point plots…Colour range symbols…”


We get the first of a series of dialog boxes:

We fill out the Au_ppb channel to base the classification on, and choose 4 ranges. We best change the symbol edge colour to white, after which Geosoft ask us whether we want it white in colour, or whether we want no edge for the symbols, answer NO so that no outline will be drawn for the circles…

We can then change the Ranges by clicking the “Ranges” button, and get the following dialog box:

We can fill out threshold values of 5, 15 and 50 ppb, which means classes 0-5 ppb (i.e. background), 5-15 ppb, 15-50 ppm and > 50 ppb. We can click on Symbols and can change the symbols appliedto each class if necessary. Let’s leave all symbols as circle:

Click Ok and we get back to the previous dialog box. We can now choose Sizes and get the following box:

Fill out the sizes 0.2, 0.3, 0.4 and 0.5 and click OK and then choose Colours:

Here we’ll change the colours to have a colour ramp from white over yellows to red. Click OK, and again OK, and we can press PLOT to plot our map:


We can now zoom into the areas with (high value) red dots, link the map to the data, and the data to the map, and explore the anomalous values.

The map is not complete as it lacks a legend to explain the symbology used. A legend can be generated automa-tically through Chemmap…Point plots…Colour range symbol legend…

Which brings up the following dialog box:

Click plot and the legend appears on the map. Manually select the legend, and drag it to the desired location on the plot. Minimum Curvature Gridding A common, and very useful operation is gridding. Gridding basically imposes a grid, the grid spacing and dimensions of which have been predefined by the user, and recalculates the sample values, based on a chosen algorithm, to assign virtual values to the grid nodes. Naturally, depending on the type of data, sampling distances, line spacing and a lot more, the gridding parameters have to be adapted to suite certain needs. In our case, data has been collected along lines, spaced at about 4000m (2000m along the Sfariates


ridge), but with station spacing of 400m or less. In this case, the grid is inequidistant, which is not well suited for simple gridding algorithms. Gridding needs to take into account the different sampling density according to direction…such gridding is called BIDIRECTIONAL LINE GRIDDING, and is unfortunately not included in the Chimera module, as it is a typical gridding method to be applied on airborne geophysical data. The gridding options available to us are “Minimum Curvature” and “Kriging”. Let’s first attempt to grid the data using the method that produces the smoothest possible surface that will fit the given data values. Choose “Grid…Minimum Curvature…Dialog Controls…”


Fill in the name of the Channel (Au_ppb), the Grid name (“Au_mc_A4”) and choose a cell size, which is ¼ of the maximum line spacing. Click on Advanced so we can fill in a few more parameters:

We need to set a blanking distance, which is just larger than the maximum line separation (5000m). Click Finish, and the grid is calculated. Next we can display the grid on a new map. Let’s open the Base map, make a duplicate called “Au_mc_A4” and change the title text to:

We can click “Grid…Display Grid…Colour Shaded Grid…”



We can select the grid file, leave all options as is and click the “Current map” button. To put the legend we can go back into the menu and select “Horizontal Colour Legend bar…”

A dialog box appears in which we can leave all parameters as is, and click OK. We have to move the colour bar to a desired location and the map looks something like this:

Notice the the Minimum curvature gridding algorithm has evened out all anomalous values, so that the maximum value in the grid (according to the colour bar, is only 4.6 ppb. Note also that the gridding has introduced negative values… To overcome these problems, it may be advised to introduce “false” values to replace all assay results that are too close to Detection limit. We will sort our database, and assign a value of 5ppb to all assays that are 5ppb or below. This will even out the data, so that detected values higher than 5ppb will play a greater role in the gridding. To do that we will work with the Random database, rather than the Line database. Close all windows and plots, saving all content when asked… Then open de “season#1_assay.gdb” database. We’ll sort all data according to the Au_ppb channel. Click on “ChemUtilities…Sort all by 1 Channel…”

Select the Channel “Au_ppb” as a sort key:


The database is now sorted in ascending order on the Au_ppb channel. Instead of changing the actual Au_ppb values, we’ll create a new channel, in which we’ll recalculate and modify the values. First we’ll have to create a new Channel called “Au_recalc”. Right-click on the first empty header to the right of the data, and select New…

Type in the following fields in the dialog box:

and click OK. The new Channel is created, and filled with Dummy values (asterisk). We’ll now copy the Au_ppb channel into that new Au_recalc Channel: Go to the ChemUtilities…Copy Channel… menu…

and the following dialog box appears:

Fill out the fields and click OK. We now have a new Channel with all data. We can now scroll down until we find the last value of 5 ppb, click, hold SHIFT, and then scroll all the way back up and click on the first value (while holding SHIFT). Now all values up to 5ppb are selected. To change these values to 5ppb, simply type 5 in the Cells field at the bottom of the screen and hit enter. All values below 5ppb are now 5ppb. We can now re-grid the data using Minimum Curvature (call the grid “Au_recalc_mc_A4”), create a Duplicate map based on the Base map called “Au_mc_recalc_A4”, plot the Grid with legend:


The grid looks a lot smoother, but we can see that our background value may have to be increased to smoothen it further. Notice that the Grid values are now all positive, but there is still a maximum of 6ppb, while we know that there are a limited number of samples in excess of 15 ppb… The Minimum curvature method smoothens out the data, so that these anomalous values get lowered to fit the surface… Kriging Kriging is a special approach to creating a grid. Where Minimum Curvature tries to model a smooth surface based on all data, usually using an inverse distance algorithm, Kriging uses a more rigid statistical approach that takes into account the correlation of the data in function of the distance. The method is applied in two steps: firstly one creates a semi-variogram that shows the relation of the data in function of the distance as well as the number of data-pairs versus distance. On the basis of that, the model is chosen (Gaussian, Sphere, Power, Exponential etc…), values are determined for nugget (intersection of variogram with Y-axis=average error

in each data point), sill (Y-axis value where variogram reached the plateau (over that, the values are uncorrelated)), slope (corresponding distance at which sample pairs are uncorrelated) etc… and then the dialog controls are used again, all values entered, and the actual gridding performed. Kriging requires a lot of experience before it can be applied correctly… Let’s start by making the semi-variogram for the Au_recalc Channel. Go to Grid… Gridding … Kriging … Dialog Controls…


We’ll choose the Au_recalc Channel, type in the name of the Grid file as “Au_krg_A4.grd”, and specify a cellsize of 500m. Click Advanced> and we get into the advanced dialog box:


We can fill out the name of the output Variogram file, type in a blanking distance of 5000m and choose a spherical model. Click on Variogram only, and Geosoft calculates the variogram. Next we can plot the variogram file into a blank map. We load the Base.map, and duplicate that into a “variogram.map” leaving the map blank (as opposed to copying the current contents). Once the blank map opens we can plot the variogram through Grid… Gridding… Kriging… Plot Variogram…

Visual inspection of the variogram allows the determination of nugget, sill and range/slope values, and we can go back into the Kriging tool to perform the actual gridding:

Type in the name of the error gridfile (Au_krg_recalc), model = spherical, power=1, Range/Slope=7500, nugget=0.5 and sill=8.5. Click back and then OK and the grid is created. Let’s plot up the grid into a Duplicate map from Base.map, naming it “Au_krg_recalc_A4.map”. We can also plot up a horizontal colourbar, change the titles a bit, so that the map looks as follows:

When compared to the Minimum Curvature grid, it is immediately apparent that Kriging manages to perform the gridding process in a more “clustered” way. This is because the procedure ignores values beyond a


certain (statistically defined) distance. Away from values in excess of 10ppb, values quickly return to the artificially introduced background of 5ppb (green). It is important to remember that a grid is a model, in which all values are recalculated based on our data. This means that the actual values will differ from the calculated grid values. To see just how much change has been effected through the Kriging process, we can sample the grid, and place the gridded values for X and Y positions, into a new Channel we’ll call “KrigridAu”. First open the “season#1_assay.gdb” database (if not open, then create a new Channel called “KriGridAu” (floating point). We can now use Grid… Utilities… Sample a Grid…

We fill in the parameters in the dialog box:

The grid gets samples, and corresponding values placed in the new Channel. Let’s compare the original Au_ppb Channel with the sampled grid values by plotting the profiles for both in a profile window:

Contouring Contouring is a procedure that can be useful for displaying trends in the data, and for delineating anomalies. The procedure can only be carried out when a grid has been created. To apply a contouring operation, we use Mapping… Contour… Quick…

A dialog box asks for the grid file to use, and contours will be immediately created.

Because of the lack of interactivity, this “Quick” tool is best left aside, and


we’ll perform a more interactive contouring algorithm. First we’ll need to open the map with the Kriged grid (Au_Krg_A4). We now access the Contouring menu through Mapping… Contour… Contour… and get the following dialog box:

We can click on options, and get the following dialog box:

Change setting to your liking and click OK. Clicking Line styles in the first dialog box brings up a series of dialog boxes to adapt the appearance of the lines:

Change all these settings to obtain a desired result. With our gold database, contouring is a difficult way to show data, as a lot of anomalous values occur on their own. This makes contouring a tedious and unproductive exercise in this case, and point plots confer the message a lot better. Histogram analysis Geosoft Chimera has a range of powerful statistical tools that help in


the analysis of a dataset. In our case of the gold dataset, the use of these tools will be limited to univariate statistics, as there is only one variable (Au). Many datasets will actually have a range of element, as known associations of elements in nature are used to better detect possible anomalies. The associated elements, which often have a better detection Limit are called PATHFINDER elements, and are very important in mineral exploration. For gold, pathfinder elements commonly analysed in conjunction with Au include As, Sb and W. A powerful statistical tool for univariate datasets is the histogram. A histogram plots the number of analysis that fall within a chosen bin, and does that for all data in a dataset. The histogram plot is often combined with a cumulative curve, which can be used to detect different populations within a dataset. To start up the histogram tool, we will first make a new Channel in which we’ll copy the Au_recalc Channel. We will need to mask out the most prominent background population of 5ppb so that we can make any sense of the important data (> 5ppb). After having created the new Channel, and copied the Au_recalc data into it, we can select all data of 5ppb, and type in the Dummy value for them (Spacebar). This replaces all 5ppb values by *. We can now start the Histogram tool “ChemAnalyse...Histogram analysis…”

We immediately get a histogram plot of all data…(Dummy’s are not included):

We can change the Minimum and Maximum on the X-axis, as well as the number of classes (division) with a maximum of 1024. The histogram is directly linked to the datasheet, so that changing a value in the database will result in an automatic updating of the histogram plot. I have changed the Divisions to 160 (which is just over the maximum value in our database), so that every step accounts for 1ppb. We can now move the cursor, step by step across the histogram, and can immediately read off the value and the percentile for that value.


A summary of the statistics is displayed at all times, and the histogram tool tells us that 90.34% of all data has a gold value of 13ppb or less. With all values <5ppb as a Dummy, we only have 290 values remaining…only 10% of those are in excess of 13ppb (about 30 points). For this particular dataset, this means that most statistical tools often used in exploration, applying percentiles of 25%, 75%, 90% and 95% will be quite useless in our case. Only very few data points have anomalous values, but the data is too abnormally distributed to be able to rely on classical statistics to calculate values such as thresholds, anomalous values, lower and upper fences etc… We will work on advanced statistics using another database (NW Zambia) later in the course. Advanced mapping techniques So far, we have only worked with the geochemical dataset, without using geological or other criteria to analyse the data. In many cases, quite a lot of information will be available in either vector or image formats, which can be used in conjunction with geochemical data to make sound decisions. We will load in a scanned, and georeferenced base map (toposheet), and will also

briefly load in and work with a aeromagnetic map. To load a geocorrected image into our maps, we can use Grid… Display grid… Image (bmp,tif…)…

First load the Base map, and Duplicate it into “Topomap.map” (with the surround). The image we want to load is called “sfariates_topo2_utm29wgs84.tif”:

Choose “Default registration” as the image is geocorrected, and should automatically plot correctly with respect to our data. The map should look like this:


We can also load in total magnetics, preprocessed over the area, in four panels, and see whether there is a relation between magnetic vertical gradient and the anomalous values in the geochemistry (files: ####_GRADIENT_VERTICAL, with ### 2310, 2311, 2410 and 2411).

Masking data (Chemmask) based on CODE Another important tool in analysing data is the ability to subset the dataset. Geosoft offers the ability to create a Mask channel, which effectively temporarily removes data from the database. Masking can be done based on spatial criteria (X,Y limits, or features selected from polygons), or on alphanumerical criteria. In our case we have a dataset that contains assays

done from OMAC, and assays from another laboratory OMRG. We will create a Mask Channel (Chemmask), and mask out all samples from OMAC, and then subset the database (extract that data into a new database). We can then do the same for the OMRG data. Make sure you have the “season#1_assay’ database open. Select Chemanalyse… Classify by Code…

The following dialog box comes up:

Specify the Channel that contains the code on which we classify (Lab), and type in the code “OMAC”. Make sure the maskchannel is Chemmask (if it does not yet exists, it will be created), and choose New (not append). Notice that the Chemmask Channel is NOT immediately displayed in our database. We need to right-click on the empty column header to the right of the last column, choose LIST and then select Chemmask… OK.


The Chemmask channel now contains a value 1 for all OMAC samples, and a Dummy (*) for all other values. Now go to Chemanalyse… Create subset database…

And fill in the following:

A database with only OMAC assays will be created. We can repeat these steps to mask all OMAC values out, and create an OMRG database… Masking data using drawing tools Let’s load the topomap, and plot the sample localities (colour ranges) on top. We’d like to create a subset of data collected immediately northeast of Sebkha Oumm el Drous Guebli in the centre of the area.

We’ll select the area of the Sfariates ridge immediately NE of the Sebkha using a polygon: Go to Chemanalyse… Classify by region … Create Polygon mask…

The following dialog box pops up:

Fill in a name for the polygon file (.ply), and use inclusive (other option is exclusive…). Notice that one can append the selection to a polygon mask, or make a new file. Choose New. Now digitise the polygon, and right-click when done. The polygon has been created but is not automatically shown. To load the polygon into the map we have to go back through the


menu Chemanalyse… Classify by Region… Draw polygon outline…

Select the polygon we just created and click OK.

Next we can apply the polygon mask to the data through Chemanalyse… Classify by region… Apply Mask to data…

Make sure you select New, or the mask is appended to the pre-existing selection. Using mathematical expressions In order to demonstrate the use of mathematical expressions, we will load another database of a multi-element stream sediment survey in NW Zambia. The database is called “NW_Geochem_RC.gdb” We will concentrate on the assays for Au, As and Mo. Notice that the assay columns for Au, As and Mo are not

displayed, so we need to right-click on the first empty column header, and list the channels. Please pick Au, As and Mo. We’ll make a new Channel called “Sum_As_Au_Mo” and calculate the sum of As, Au and Mo in it. We can then run a few statistical tools, and finally make a colour point plot. First create the new Channel. Then we’ll use a mathematical expression to calculate the value for that channel based on values in the As, Au and Mo Channels. Click Chemutilities… expression…

Type in the following formula:

and the values are automatically calculated into our new Channel. Now we can create a map based on the new database, plot a new basemap with surround, and then plot the values of the Channel “Sum_As_Au_Mo” as colour ranged symbols.


The modified BOXPLOT The boxplot method is a statistical method of defining thresholds and splitting a population of data into classes, which can be used as a base for symbol plots, and definition of “anomalous” values. The following statistical values are calculated: 1st quartile (25th percentile): The maximum value of the lowest 25 % of the total sample population. 3rd quartile (75th percentile): The minimum value of the highest 25 % of the total sample population. Lower fence: 3rd quartile-1st quartile-1st quartile * 1.5) Upper fence: 3rd quartile-1st quartile+(3rd quartile*1.5) Values above upper fence are regarded as anomalous (statistically). Advanced use of the Histogram The histogram tool allows the direct delineation of classes, roughly following the guidelines of the BOXPLOT method. The Quantiles are directly shown on the tool, and the shape of the histogram can allude to “natural” breaks in the dataset, on which ranges can be extracted. Let’s use the histogram tool on the

Au_f_1ppb channel (Fire Assay gold assay). In excel I calculated the various statistical parameters of the BOXPLOT as follows: 1st Quartile = 2 3rd Quartile = 6 Lower Fence = -5 Upper Fence = 13 The histogram plots as follows:

We can move the cursor on the tool, and notice that 90% of our data has values < 8ppb. We are again faced with the fact that the data for gold is too close to background to be able to define different populations. Using the threshold that makes a value anomalous (13ppb) we can see that 97.93% of our data actually plot below that, leaving only 2.07% of our data as potentially anomalous. Let’s try and use the histogram tool on the Channel “Fe_0_02_”.


Here we can clearly see that the data contains several populations. This is a normal feature for iron, which can exist as Ferric and Ferrous. Using the histogram tool, we can now select populations, mask the data, and interactively plot the various population in different colours on a map. We’ll be using the following buttons on the histogram tool:

Define masking limits Apply the mask Reset mask

Refresh

Plot symbol legend Plot data Plot histogram on map

First we can define a masking limit using the mouse, directly on the histogram. The selected range is highlighted in pink, and we can apply the selected range to the database:

A dialog box appears:

After clicking OK, we can plot the symbols to the map and get the following dialog box:

Since the Chemmask channel now contains the mask based on the selected range, plotted symbols will fall within that range, while samples outside the mask will not be plotted.

It is best to immediately rename the data layer on the map (using “Views/Groups manager”), so that when we plot our next range, the already plotted data is retained…


Correlation analysis In multi-element geochemical datasets, it is common practice to assess the correlation of various elements. Strongly correlated elements can be used together, or as pathfinder elements. Geosoft allows the calculation of the correlation coefficients, and simultaneously can plot the correlation diagrams (scatterplots) for easy access to these statistical parameters. We will remove all data columns of the database, and then load the Cu, Pb, Zn and Fe Channels, and calculate/plot the correlation characteristics. First remove all Channels from the database (right-click menu):

Then we can load in Cu, Pb, Zn and Fe. We then access the Correlation tool in “Chemanalyse… Correlations…”

The following dialog box appears. We select “Displayed Channels”, type in a title and leave all options as default.

The following plot is created:

The colours indicate a moderate correlation between Zn and Fe, but weak correlations for all other elements. The scatterplots also show relatively smaller scatter for the Fe-Zn plot than for the other… Scatter Analysis More in depth correlation analysis can be done using the Scattertool. A scatter plot displays the relation between two parameters directly. We can access the


tool in Chemanalyse… Scatter analysis…

The scatterplot immediately appears, and we can select the Channels for both axes using the button.

Buttons from left to right top row: Define masking limit, define polygon mask, set plotting symbol From left to right second row: Apply masking limit, select all points, plot symbols on map. Lowest row: refresh plot, create multi-scatter plot and plot scatterplot on map. Using the polygon selection button ( ) we can select any cluster of points on the diagram, then assign the selection, which automatically creates a Chemmask channel, a plotting symbol ( ). This brings up a symbol selection tool as follows:

Once the symbol has been selected, we will be prompted with the following dialog box:

Here we can enter the name for the group, allowing us to perform the same set of steps over and over again, plotting various populations one at a time on the map, in different symbols, and in their own Groups.

The Geosoft Help system Although most features commonly used in Geosoft are addressed in this course, there will always be features that are left to the individual user to discover. Geosoft, with Chimera extension, has enormous depth, and offers a lot more features than could be fitted in this manual. To advance your understanding of the software and its applications, you can visit Geosoft online at http://www.geosoft.com, or you can access the help system and tutorials provided with the software. An excellent tutorial dataset is included in the installation CD’s, and PDF files guide you every step of the way to discover the vast capabilities of the software.

Documents

USING GEOSOFT OASIS MONTAJ AND CHIMERA - Bert De Waele (for IMC Consulting) 1 8/07/2005 USING GEOSOFT OASIS MONTAJ AND CHIMERA Course developed for IMC Group Consulting Limited To