USER MANUAL - Fabio Veronesi
Contents
1. ID Lat Lon value depth 1 XXXXXX XX yyyyyy yy ii dd 2 XXXXXX XX yyyyyy yy iii ddd 3 XXXXXX XX yyyyyy yy iii dddd 4 XXXXXX XX yyyyyy yy iii ddddd The format of the two spatial coordinates is not important because within the sp package in R there is the possibility to set the projection of your data according to the proj4string format http spatialreference org It is important tough the format of the depth values they have to be in centimetres and positives not 10 but 10 Once the user insert the correct projection of his data all the interpolation will consider the Euclidean distance between data points Cranfield University f veronesi cranfield ac uk IntR3D Interactive GUI for R Fabio Veronesi The important thing is that the header of the file is exactly the same of the above example Covariates format All the covariates need to be in ascii grid In IntR there is a module that converts the data between the ESRI shape file and the ascii grid format In this module the user can convert a raster into an ascii grid selecting the resolution of the final grid Starting IntR When the user starts IntR the software shows a window with a button for each module of the program The user can very intuitively click on the button that corresponds to the module he wants to start To consult this user manual the user can click on the menu File and on the menu item User Guide Each module corresponds2. the folder c RHOME All the covariates need to be in ascii grid asc format Running the Program The working flow for 3D Universal Kriging is the following select data file input of ID number of the projection select the prediction grid size select number and covariates files input the linear regression model select the variogram model execute the cross validation perform the kriging The first information the program needs after the data file is the ID number of the projection For more information see paragraph Projection below The data file as be a txt and the format has to be the same of the one presented in the Data Format paragraph In IntR3D the Universal Kriging module does not create the prediction grid itself but it imports it from the output of the module 3D Grid Subsequently the program will ask for the covariate number and to select the covarites files The user must be careful during this step because if he needs to insert 2 covariates for example the second file selection window for selecting the second covariate will appear very shortly after the first At this point the user must insert the linear regression model The model has to be in the form covariate 1 covariate2 covariateN Ex if the two covariates are named slope asc and em38h asc the model will be slope asct tem38h asc it is important to insert the entire name of the file extension included When the user finishes inp
3. to a different task that will be compiled interactively in IntR and executed with R Following there are detailed instructions for each module E intR3D Interactive GUI for R 3D Universal Slices Kriging 3D 3D Video Variogram IDW Output 3D 3D Ordinary Anisotropy Kriging Figure 1 starting windows of IntR3D The user here can intuitively select the algorithm that he needs to use Cranfield University f veronesi cranfield ac uk IntR3D Interactive GUI for R Fabio Veronesi 3D Variogram amp 3D Anisotropy In IntR3D there are two modules that are used for performing some preliminary studies upon the study area These two modules are 3D Variogram and 3D Anisotropy 3D Variogram This module studies the spatial variability of the dataset performing the variogram computation The results are a series of three images the omnidirectional variogram four directional variograms a 0 45 90 and 135 degrees a variogram map that can be used to assess the anisotropy of the field and a series of variogram with differents dips and directions Installation Simply run the script START exe and select 3D Variogram from the buttons In order to run the software all the data and covariates that you need for running the interpolation must be in the folder c RHOME Running the Program The working flow for Variogram Plot is the following select data file input of ID number of the projection select number and covari
4. IntR3D Interactive GUI for R Fabio Veronesi USER MANUAL Introductory Notes This software is a simple interface for compiling R script interactively After the script is compiled it runs it in batch mode All the computations are done using packages already present in R and NOT CREATED BY THE AUTHOR OF THIS SOFTWARE INTERFACE The packages with the relative authors are listed below GSTAT Pebesma E J 2004 Multivariable geostatistics in S the gstat package Computers amp Geosciences 30 683 691 sp Pebesma E J R S Bivand 2005 Classes and methods for spatial data in R R News 5 2 http cran r project org doc Rnews Roger S Bivand Edzer J Pebesma Virgilio Gomez Rubio 2008 Applied spatial data analysis with R Springer NY http www asdar book org maptools Nicholas J Lewin Koh and Roger Bivand rgdal Timothy H Keitt Roger Bivand Edzer Pebesma Barry Rowlingson lattice Sarkar Deepayan 2008 Lattice Multivariate Data Visualization with R Springer New York ISBN 978 0 387 75968 5 akima Albrecht Gebhardt 1998 Akima Interpolation of irregularly spaced data Note on R R is a very powerful statistical language However it is also a slow language and this is clear in geostatistical interpolations For this reason especially when it is interpolating with inverse distance it takes some time to finish the process Do not worry too much and let it works System Requirements The system requ
5. Please write the model of the variogram Type 0 for the default model Sph for Spherical Model Exp for Exponential model Gau for Gaussian model Mat for the Matern if the user wants the spherical model he needs to insert Sph with the first letter in capital Subsequently a dialog appears and asks for the two anisotropy parameter The user if an anisotropy is presents needs to insert the two numbers separated by a comma If no anisotropy is present in the field the user needs to insert a single 0 Cranfield University f veronesi cranfield ac uk IntR3D Interactive GUI for R Fabio Veronesi Ex Parameter p is 45 and the parameter s is 0 5 The user has to insert 45 0 5 At this point the cross validation starts After it finishes the cross validation it shows a window with the R Squared value of it and a button that shows the variogram image which is also saved on C If the result of the cross validation is satisfying you can click on YES and automatically proceed with the kriging with the same linear regression model If the result is not good you can click on NO and repeat the cross validation selecting again the linear regression model and all the parameters of the variogram Projection To set the projection of the map you need to insert the ID number of the coordinates system You can find a list of ID number and the relative system here http spatialreference org ref search utm amp
6. ates files input the linear regression model select the variogram model execute the cross validation This module can be used for an ordinary kriging variogram or for computing a regression variogram which uses a linear regression module Even if the user wants to perform a simple ordinary kriging variogram he needs to insert a covariate file that will be used for the variogram boundary Results At the end of the module IntR3D displays a window with four buttons that are used to show the images of the variograms 3D Anisotropy Anisotropy is a module that is used to provide the user with the five values used to assess the geometrical anisotropy of the field A geometrical anisotropy is evident when the directional variograms show the same sill but different ranges To correct this anisotropy the user needs to insert five parameters the angle of the direction of maximum continuity in positive degrees from north the dip in the direction of maximum continuity the rotation angle of the other two axis the last two values are ratios between the main axis and the other two Installation Simply run the script START exe and select 3D Anisotropy from the buttons In order to run the software all the data that you need for running the interpolation must be in the folder c RHOME Running the Program This module asks only for the data file the projection and an ascii grid for bounding the variograms Results At the end of the
7. e The user needs to select the 3D file usually a txt generated by the module 3D IDW Then he needs to select the type of image he wants as output The user has three choices a coloured image a coloured image with contours and finally a contour image The next step is to select the depth latitude and longitude interval of the slicing process The user can type 0 to let the program create a slice for each depth latitude or longitude But he can also select an interval if he is not interested in the entire field 3D Video This module copies the file mencoder exe www mplayerhq hu which is a very powerful encoder that works with command lines into the folders created by the module 3D slices After this step the module starts mencoder and performs the creating of the 3 videos one for depth one for latitude and one for longitude Cranfield University f veronesi cranfield ac uk
8. irements are the same of R In this version of IntR3D is included the version 2 12 0 of R that is used to run the scripts This version of R is compatible with 32bit machines On these machines R can run but it has some memory limitations It can use a maximum of 4 Gb of RAM even though in some machines the amount is limited at 2 or 3 Gb of RAM This means that sometimes R will not be able to complete the process because it is not able to allocate the data onto the RAM However there are some ways to limit the amount of memory necessary for complete the process The common tip is to reduce the size of data and covariates In IntR3D the user can choose in every step of the process the resolution of the ascii grid and the resolution of the map that he wants to create This way if the user notices that R does not complete the process because of a memory related problem he can just decrease the resolution and try again The only way to increase the memory that R can use for its computation is to use a 64bit machine and IntR3Dx64 Another limitation regards the use of the CPU By default R can use only one core Therefore in the task manager when R is performing the prediction will indicate that the CPU usage is low This is because is using a fraction of the virtual power of the machine Data Format IntR works with data file arrange in a textual table in txt format The structure of the file must be like the following example
9. module IntR3D displays a window with the values of the five parameters used to correct the geometrical anisotropy Those 5 values are saved in a txt file called parameters txt within the RHOME folder To correct the anisotropy the user needs to insert manually the five parameters when asked in one of the kriging modules Cranfield University f veronesi cranfield ac uk IntR3D Interactive GUI for R Fabio Veronesi 3D Grid IntR3D has been created to be used like a unique working flow rather than a series of separate modules like IntR For this reason the creation of the 3D grid to be used as new data for the final prediction in the kriging modules is performed in a separate module Installation Simply run the script START exe and select 3D Grid from the buttons In order to run the software all the data that you need must be in the folder c RHOME Running the Program This module asks for the data file the horizontal size of the prediction grid the vertical resolution is 1m absolute or relative altitude an ESRI shp file of the area The user can choose between absolute or relative altitude If he select relative altitude the software creates a grid that start from a depth of 0m and proceed by one till reaching the maximum depth of the data file If the user select absolute he needs to import also a DTM file in ascii grid and the program will correct all the depth based on the real altitude of the point Results At the end
10. nterpolation Running the Program The working flow for Inverse Distance is the following select the SHP file with the prediction area select the size of the prediction grid select data file perform the inverse distance This module was included for increasing the number of point in the prediction grid The main purpose is to visualize the results properly This module can be used to perform an inverse distance interpolation on a 3D dataset but it lacks the cross validation part So the user can not be sure about the precision of the prediction This module is very easy to use because it asks only for few information and performs the interpolation Results The result is a file called 3Dinverse txt which can be used with the subsequent module 3D Slices Cranfield University f veronesi cranfield ac uk IntR3D Interactive GUI for R Fabio Veronesi 3D Slices This module creates a series of slice images from the 3D dataset This module works by selecting from the 3D file a single value of depth extrapolating all the values from that depth and creating an interpolated image of the selected depth The same procedure is done also for latitude and longitude Running the Program The working flow for Inverse Distance is the following select the 3D file select the type of image select the depth interval select the latitude interval select the longitude interval The execution of this module is very simpl
11. of ID number and the relative system here http spatialreference org ref search utm amp srtext Search At the moment IntR supports only EPSG projections so you need to insert a number of an EPSG projection If your data are unprojected insert 0 in the apposite space It is important to insert the projection ID if the data are projected because the package GSTAT has been designed to work with projections This means that if the projection is available the interpolation is done using Euclidean Distances Output The output of the script are variogram plot in jpg prediction map in jpg kriging error map in jpg cross validation of the last used model in txt kriging results table in txt ascii grid of the map shp file with predected points Cranfield University f veronesi cranfield ac uk IntR3D Interactive GUI for R Fabio Veronesi 3D Ordinary Kriging The Ordinary kriging script is similar to the other two kriging script The format of the data file is the same coordinates columns called Lat and Lon and the data column called value The difference of course is that with the ordinary kriging you don t need to insert covariates However you still need to insert an ascii grid for interpolating the final prediction map The ascii grid can be the result of the SHPtoASC script The series of information needed by the software are similar to the one needed for universal kriging Installation Simply run the
12. of the module it does not show any summary windows it just close itself after it finished creating a file called Pred_Grid txt into the RHOME folder This file can be used with the modules 3D Universal Kriging and 3D Ordinary Kriging Cranfield University f veronesi cranfield ac uk IntR3D Interactive GUI for R Fabio Veronesi 3D Universal Kriging This python script will help you compiling and running a script able to perform a 3D Universal kriging on your data The kriging is performed with the package Gstat In the 3D Universal Kriging the cross validation is not performed with the default function implemented in GSTAT This function in fact performs an n fold cross validation extracting n data and repredicting their value But this approach when the dataset has been created resampling a spline is not appropriate because the kriging algorithm will use neighbourhood values from the same samples to repredict the missing data And the neighbourhood values are related to the missing data by a nonlinear regression equation that simplifies the prediction For this reason in this module has been implemented a method that extract an entire sample and repredict it using all the others This way the result of the validation is more accurate Installation Simply run the script START exe and select 3D Universal Kriging from the buttons In order to run the software all the data and covariates that you need for running the interpolation must be in
13. script START exe and select Ordinary Kriging from the buttons In order to run the software all the data and covariates that you need for running the interpolation must be in the folder c RHOME All the covariates need to be in ascii grid asc format To convert a table with Latitude Longitude and covariate value into asc you can use the Inverse Distance algorithm in IntR or any other GIS software To convert an ESRI raster into an ascii grid use the module from SHP to ASC Running the Program The working flow for Universal Kriging is the following input of ID number of the projection select data file select the prediction grid size select number and covariates files input the linear regression model select the variogram model execute the cross validation perform the kriging The first information the program needs is the ID number of the projection For more information see paragraph Projection below After that the program will ask you to choose the data file The data file can be a txt and the format has to be the same of the one presented in the Data Format paragraph In the next input windows IntR asks to insert the grid size of the prediction grid If the user inserts 10 for example it means that IntR will predict a value on a square grid of dimension 10x10 m Subsequently the program will ask for an ascii grid This will be used to bound the variogram Ex on the screen you will see a message like
14. srtext Search At the moment IntR supports only EPSG projections so you need to insert a number of an EPSG projection If your data are unprojected insert 0 in the apposite space It is important to insert the projection ID if the data are projected because the package GSTAT has been designed to work with projections This means that if the projection is available the interpolation is done using Euclidean Distances Output The output of the script are variogram plot in jpg prediction map in jpg kriging error map in jpg cross validation of the last used model in txt kriging results table in txt ascii grid of the map shp file with predicted points Cranfield University f veronesi cranfield ac uk IntR3D Interactive GUI for R Fabio Veronesi 3D Inverse Distance This program performs an inverse distance interpolation in 3D The data file can be a txt and the format has to be the same of the one presented in the Data Format paragraph The main purpose of this module is to create a continuous lattice from a series of data points in 3D which are the result of the 3D kriging At the end of the 3D kriging in fact the program performs an interpolation in each point of the prediction grid The result is a series of points with a predicted value for the property of interest However there is no way in R to visualize this prediction as a continuous surface In order to do so the only way is to perform a further i
15. ut the linear regression model IntR asks for the variogram model It important to insert the name as it is shown on the screen Ex on the screen there will be a message like Please write the model of the variogram Type 0 for the default model Sph for Spherical Model Exp for Exponential model Cranfield University f veronesi cranfield ac uk IntR3D Interactive GUI for R Fabio Veronesi Gau for Gaussian model Mat for the Matern if the user wants the spherical model he needs to insert Sph with the first letter in capital Subsequently a dialog appears and asks for the five anisotropy parameters The user if an anisotropy is presents needs to insert the five numbers separated by a comma If no anisotropy is present in the field the user needs to insert a single 0 At this point the cross validation starts After it finishes the cross validation it shows a window with the R Squared value and a button that shows the variogram image which is also saved on C If the result of the cross validation is satisfying you can click on YES and automatically proceed with the kriging with the same linear regression model If the result is not good you can click on NO and repeat the cross validation selecting again the linear regression model and all the parameters of the variogram Projection To set the projection of the map you need to insert the ID number of the coordinates system You can find a list
Download Pdf Manuals
Related Search