Lecture 19 - Spatial Distribution of Geochemical Data Part 1 PDF

GLGY3764 Lecture #19 Modelling the Spatial Distribution of Geochemical Data Part 1 Dr. John Carranza Professor of Economic Geology Definitions  What is a model (noun)? Concise Oxford...

GLGY3764 Lecture #19 Modelling the Spatial Distribution of Geochemical Data Part 1 Dr. John Carranza Professor of Economic Geology Definitions  What is a model (noun)? Concise Oxford Dictionary (1995)  a representation in three dimensions of an existing person or thing or of a proposed structure  a simplified (often mathematical) description of a system, etc., to assist calculations and predictions Glossary of Geology (1987)  A working hypothesis or precise simulation, by means of description, statistical data, or analogy, of a phenomenon or process that cannot be observed directly or that is difficult to observe directly Sides (1992)  A representation of reality, based on recorded measurements at visited locations, which can be used as a predictive tool to forecast other measurements at unvisited locations 2 Definitions  What is a map (noun)? National Geographic (Online)  a symbolic representation of selected characteristics of a place, usually drawn on a flat surface Axismaps.com  a symbolic interpretation of place and highlights the relationships between elements in space, either perceived or actual Anonymous  a spatial (geographic) representation of one or more features 3 Definitions So, MAPS = MODELS 4 Why do we need to model (or map) the spatial distribution of geochemical data? To recognize the relative importance of and understand underlying geochemical processes;  To document geochemical background and variations therein;  To discover mineral deposits;  To identify and document human impact on natural element distributions;  For land use planning and many other political decisions, e.g., for defining clean-up levels following industrial activities; and  To understand the importance of chemical elements for the onset and distribution of endemic diseases. 5 Why do we need to model (or map) the spatial distribution of geochemical data? To recognize the relative importance of and understand underlying geochemical processes;  To document geochemical background and variations therein;  To discover mineral deposits;  To identify and document human impact on natural element distributions;  For land use planning and many other political decisions, e.g., for defining clean-up levels following industrial activities; and  To understand the importance of chemical elements for the onset and distribution of endemic diseases. 6 General considerations for modelling (or mapping) the spatial distribution of geochemical data 1. Geographical information systems (GIS) versus paper maps special techniques are needed to present geochemical data in a map to extract the full information value, independent of whether paper or computer maps are prepared, and these techniques are still not provided in most GIS. 7 General considerations for modelling (or mapping) the spatial distribution of geochemical data 2. Map scale The choice depends on the size of the mapped area and the purpose of the map  A map should contain just so much information that the user can easily grasp the inherent facts at the selected scale. Too much information and the map becomes hard to read, too little and it is not fit for purpose.  A4 size (210 mm × 297 mm) should normally be the maximum size for a geochemical map  The information value of a map increases when the scale is decreased to give the smallest map possible 8 General considerations for modelling (or mapping) the spatial distribution of geochemical data 3. Background information A geochemical map is usually superimposed on some kind of topographical background map 9 General considerations for modelling (or mapping) the spatial distribution of geochemical data 3. Background information If the background map is a geological map, however well designed for that purpose, the addition of geochemical information will result in a cluttered map  How much of what background information is needed in a geochemical map? (from Banks et al. 2001). 10 General considerations for modelling (or mapping) the spatial distribution of geochemical data 3. Background information When producing a geochemical map, the most important information is the spatial distribution of the chemical elements studied  It is the geochemistry that should govern the mapping and not geography, geology or other ancillary information  Background information not related to geochemistry should be avoided because it distracts attention from the geochemistry 11 Placing geochemical data into a map 1. Write analytical value on map  Disadvantages:  Hard to read once number of samples increases.  No identification of outliers.  Advantages:  Direct relation to the original analytical results.  Can be used to trace contours. 12 Placing geochemical data into a map 2. Arbitrarily chosen class boundaries, e.g., 10, 20, 30,...  Disadvantages:  No relation to data structure.  No definition of outliers.  Advantage:  Classes are easily understood and look ‘logical’ in the legend. 13 Placing geochemical data into a map 3. Calculate mean and standard deviation (sd), set class boundaries to mean ± multiples of sd  Disadvantages:  Requires (log)normal transformation of the data and may result in negative values at the lower end.  Class boundaries will be strongly influenced by outliers.  Advantage:  May have relation to data structure if outliers are detected and removed before calculating the limits. 14 Placing geochemical data into a map 4. Symbol size increases in proportion to analytical value  Disadvantages:  Scaling can be difficult.  High values dominate the visual effect of the map.  Advantage:  Good visual maps IF scaling is done correctly. 15 Placing geochemical data into a map 4. Symbol size increases in proportion to analytical value Combined use of bubble plots and colour with sizes and tints, respectively, varying according to value. 16 Placing geochemical data into a map 4. Symbol size increases in proportion to analytical value  How do we vary the sizes of the bubbles with variation in chemical contents for proper scaling? poor visual map is from Stanley (2005) differentiation cluttered 17 Placing geochemical data into a map 4. Symbol size increases in proportion to analytical value  How do we vary the sizes of the bubbles with variation in chemical contents for proper scaling? 18 Placing geochemical data into a map 4. Symbol size increases in proportion to analytical value  How do we vary the sizes of the bubbles with variation in chemical contents for proper scaling? poor visual map is aesthetic differentiation cluttered from Stanley (2005) 19 Placing geochemical data into a map 5. Percentiles and symbols  Disadvantages:  Choice of symbols is crucial.  No outlier definition (other than ‘the uppermost 2.5 or 5% of the data’).  Advantages:  Direct relation to data structure when classes are evenly spread over the whole data range.  Especially well suited for colour mapping. 20  Placing geochemical data into a map 5. Percentiles and symbols from Reimann (2005) Boxplot classes Percentile classes, Percentile classes, Percentile classes, evenly spread over only one colour for only one colour for the whole data the lower 50% of all the lower 75% of all range. data. data. To calculate percentiles, see: http://web.stanford.edu/class/archive/anthsci/anthsci192/anthsci192.1064/handouts/calculating%20percentiles.pdf 21 Placing geochemical data into a map 6. Exploratory data analysis (EDA) approach (boxplot and EDA symbols)  Disadvantage:  Resulting maps are difficult to read at first glance, high values have no visual impact.  Advantages:  Data structure is directly displayed in the map, geochemical processes become visible.  Lower outliers will be detected if present.  Does not assume a normal distribution. 22 Placing geochemical data into a map 6. Exploratory data analysis (EDA) approach (boxplot and EDA symbols) 23 Placing geochemical data into a map 6. Exploratory data analysis (EDA) approach (boxplot and EDA symbols) Boxplot classes & Arbitrary class Mean + multiples of Growing dots EDA symbols boundaries & standard deviation according to arbitrary symbols percentiles from Reimann (2005) 24 Geochemical data structure … tomorrow’s lecture 25