- 1 What is Geostatistics?
- 2 Origins of Geostatistics as a science
- 3 Geostatistics vs Space Statistics
- 4 Geostatistics and GIS for spatial data analysis
- 5 Using R for Geostatistics: main packages
What is Geostatistics?
In this article cartogeek.com want to address a very interesting topic within the sector GIS and geotechnological, especially linked to data analysis space: what is geostatistics, what methods does it provide for geospatial analysis and what tools can we use with QGIS, ArcGIS and with him R programming language.
In summary, we will try to define and contextualize this scientific sub-discipline, its methods and applications in projects of important geographic component when dealing with spatial data.
The Geostatistics it is, without a doubt, a powerful tool that has been used for decades and that takes on great relevance in any project or research linked to exploration, the analysis and the display of spatial data.
What is Geostatistics?
We cannot affirm that there is a single definition of Geostatistics: There are multiple certain meanings and all of them provide different nuances.
As its name suggests it is a scientific specialty, a branch of science, which is located at the intersection of two areas of knowledge: the Geography and the Statistics.
The first focuses on the description and analysis of the Earth’s surface and the relationships that exist between diverse elements and phenomena, both natural and anthropic…using Cartography and other graphical display media as preferred tools to represent those objects or phenomena and their relationships.
For its part, Statistics is an area of knowledge that we can locate within the Mathematical Sciences. It focuses on the study and numerical description of all kinds of quantifiable phenomena.
It provides a wide range of tools that allow describing, analyzing, visualizing, estimating, relating, modeling and inferring on any phenomenon regardless of its origin, nature or complexity, always on a mathematical basis.
A definition more or less successful could be the following:
“ The Geostatistics is the branch of Statistics that focuses on predicting the spatial distribution of continuous variables associated with quantifiable phenomena »
Precisely the intersection of two areas of knowledge, both academically and professionally, allows opening new perspectives and applications.
The Geostatistics it is no exception at all … rather the opposite: from its origins it has evolved and has allowed its application to be applied methods to a wide variety of areas, obviously linked to space phenomena.
Among them we can find: Meteorology and Climatology, Geomarketing or geographic analysis of markets ( also in the conjunction of two main areas of knowledge ) as well as Biogeography, in addition to application on other sciences such as Demography, Criminology or Epidemiology, among many other examples.
Origins of Geostatistics as a science
The origins of the development of the Geostatistics, Just as its techniques and methods lie in the need to probably infer, predict or calculate the location of certain elements or phenomena on the Earth’s surface.
The first productive sector that adopted this knowledge and applied it was the mining and hydrocarbon industry, in the mid-20th century. It was from engineers, geologists, and mining industry companies that developed specific methodologies and software to locate prospecting locations.
These companies required predicting more accurately the location of potential new deposits or reservoirs of materials such as hydrocarbons, minerals or precious materials for extraction.
The Geostatistics, combining different methods, tools and techniques based on statistics, the spatial analysis and Cartography allows a huge optimization of resources.
It is only necessary to think about the costs associated with the search for suitable locations or with a high probability of containing materials of interest to the extractive industry.
In this sense, the Geostatistics it allowed to significantly reduce these prospecting costs, by generating probabilistic models that would provide great advantages when carrying out these operations.
Geostatistics vs Space Statistics
Geostatistics and Space Statistics they cannot be used synonyms because they concern different things. It is important not to confuse both concepts or use them indiscriminately as it would lead to error.
The Geostatistics it is part of the Space Statistics, and not the other way around. In other words: Geostatistics is a sub-discipline and a subset of methods and techniques encompassed within the Space Statistics.
In turn, the Space Statistics It is part of the Statistics, specifically focusing on the analysis of spatial data. Broadly speaking, it deals with the analysis and characterization by statistical parameters of spatial data series associated with phenomena that occur in space.
In itself, Space Statistics constitutes a set of techniques exploratory to describe and model distributions, patterns, processes, and spatial relationships.
Space Statistics has a clear link with Tobler’s laws, more specifically with the so-called “ First Law of Geography ” or principle of spatial autocorrelation, by which:
“All things are related to each other, but the closest things have a greater relationship than the distant ones ”.
For its part, the Geostatistics it focuses specifically on the prediction of the distribution of spatial variables associated with different continuous physical or anthropic phenomena.
Geostatistics methods and techniques allow us to know in terms of probability how a quantifiable phenomenon is spatially distributed. For example: precipitation, temperature, altitude, air pollution, soil composition…
The Geostatistics, therefore, it allows to generate space models distribution in the form of continuous surfaces. These models will be able to help us predict what values the spatial variable analyzed will have on the entire surface, although we only know its real values at certain points of it, that is, at measurement or sampling locations.
Here comes into play the role of the extractive industries that we have discussed above. With the use of Geostatistics can be generated prediction models. In this way, the places where a certain material or mineral is most likely to exist can be delimited, reducing the number of prospections: it is unfeasible to prospect the entire territory.
Similarly, Geostatistics allows configuring temperature, humidity or precipitation maps, based on a series of measurements at locations known as meteorological stations. Based on the values recorded by these, a continuous surface is adjusted by interpolating those values to predict the value of the spatial variable in places where there are no measurement points.
Geostatistics and GIS for spatial data analysis
The Geographic Information Systems or GIS are indispensable tools today to carry out analysis based on the Geostatistics. These are programs specifically designed to read, edit, analyze and transform all kinds of spatial data.
Thus, focusing on the two major GIS on the market, both free and commercial, we find that both QGIS as ArcGIS Pro ( o ArcMap) have specific tools to carry out geostatistical analysis.
Geostatistical analysis with QGIS
QGIS has different tools from multiple libraries that allow carrying out processes of spatial interpolation of variables, based on layers of points with known values and obtaining as a result continuous surfaces in raster format.
Tools for Geostatistics in ArcGIS
Alternatively, and for ESRI users who know and have access to ArcGIS Pro or ArcMap, it is worth noting the Geostatistical Analyst extension: a fantastic very complete tool to perform any type of statistical analysis spatial.
Geostatistical Analyst at ArcGIS PRO. Source: https://pro.arcgis.com/
Using R for Geostatistics: main packages
Use R to perform geostatistical analysis not only is it possible, but it is a highly recommended and extended tool for it.
The R programming language has enormous potential for statistical analysis, and also has packages especially suitable for use R in Geostatistics.
Some of the main R packages for Geostatistics:
- gstat: one of the most specific and interesting packages of R to get fully into geostatistical analysis with this programming language. Lets you use tools like creating variograms, kriging and cokriging, graphics generation, etc. You can access from here to gstat manual.
- spatstat: a multipurpose R package for the spatial statistical analysis. Link to spatstat user manual.
- GeoR: another interesting and complete package to carry out geostatistical analysis based on variograms and Bayesian methods. Link to access the manual. There is also this github article with a brief manual in Spanish.
- Georob. Other R package that gives access to geostatistical analysis functions with this programming language. Allows you to generate models and perform interpolations Kriging. Access to Georob manual.
- rgdal, a R package that provides bindings to use the library geospatial GDAL / OGR, one of the main bookstores GIS from Python with the R programming language. Here you have the user manual.
- maptools: other R package on the rgdal line to manipulate spatial information. You can access here the manual.
- ggplot2 the quintessential package for visualize spatial data and configure graphics in R. Link to manual here. There are also plenty of books available regarding this package, one of them free access is this.
- ggmap: a R package to generate maps, mounted on ggplot2. Essential for spatial data visualization and map generation static to show analysis results. Access directly to manual.
Likewise, we also recommend one of the main books specially indicated for Geostatistics and spatial data analysis with R, which is the following:
- A Practical Guide to Geostatistical Mapping by Tomislav Hengl
You can download it from this link: http://spatial-analyst.net/book/system/files/Hengl_2009_GEOSTATe2c1w.pdf
You can also review our article with a free R book compilation for GIS and spatial data analysis. In it we offer download links from the main manuals to learn how to use R and have the basic knowledge to apply in Geostatistics and spatial data analysis
Hopefully the article has been helpful and can clarify certain concepts about Geostatistics and statistical analysis of spatial data. If you have any questions or recommendations for the community, do not hesitate to leave us a comment to open debate.