Datasets:

---

LLMDH

/

OpenScience

like 0

Follow

LLM - Digital Humanities 15

ERROR: type should be string, got "https://doi.org/10.1007/s10109-020-00336-0\nJournal of Geographical Systems (2021) 23:515–546 https://doi.org/10.1007/s10109-020-00336-0\nJournal of Geographical Systems (2021) 23:515–546 ORIGINAL ARTICLE Abstract Twenty years have passed since Bivand and Gebhardt (J Geogr Syst 2(3):307–317, \n2000. https​://doi.org/10.1007/PL000​11460​) indicated that there was a good match \nbetween the then nascent open-source R programming language and environment \nand the needs of researchers analysing spatial data. Recalling the development of \nclasses for spatial data presented in book form in Bivand et al. (Applied spatial data \nanalysis with R. Springer, New York, 2008, Applied spatial data analysis with R, \n2nd edn. Springer, New York, 2013), it is important to present the progress now \noccurring in representation of spatial data, and possible consequences for spatial \ndata handling and the statistical analysis of spatial data. Beyond this, it is imperative \nto discuss the relationships between R-spatial software and the larger open-source \ngeospatial software community on whose work R packages crucially depend. Keywords  Spatial data analysis · Open-source software · R programming language JEL Classification  C00 · C88 · R15 JEL Classification  C00 · C88 · R15 JEL Classification  C00 · C88 · R15 Progress in the R ecosystem for representing and handling \nspatial data Roger S. Bivand1 Received: 9 October 2019 / Accepted: 8 September 2020 \n© The Author(s) 2020\n/ Published online: 16 October 2020 1\t\nDepartment of Economics, Norwegian School of Economics, Helleveien 30, 5045 Bergen, \nNorway *\t Roger S. Bivand \n\t\nroger.bivand@nhh.no *\t Roger S. Bivand \n\t\nroger.bivand@nhh.no\n1\t\nDepartment of Economics, Norwegian School of Economics, Helleveien 30, 5045 Bergen, \nNorway 1  Introduction While Bivand and Gebhardt (2000) did provide an introduction to R as a statistical \nprogramming language and to why one might choose to use a scripted language like \nR (or Python), this article is both retrospective and prospective. It is possible that \nthose approaching the choice of tools for spatial analysis and for handling spatial \ndata will find the following less than inviting; in that case, perusal of early chapters \nof Lovelace et al. (2019) will provide useful context. Two further pointers include \nthe fact that R and most R add-on packages are open-source software and so with-\nout licence fees or other such restrictions. The second pointer is that all scripting 012341 456789)\n3 012341 456789)\n3 516 R. S. Bivand languages provide the structures needed for reproducible research, and open-source \nsoftware gives the interested researcher the means to run the scripts needed to repli-\ncate results without cost, given access to adequate hardware. Hence, an overview of \nthe development of the use of R for handling spatial data can cast light on how and \nwhy steps fashioning today’s software were taken. Of course, an overview of the use \nof R for analysing spatial data would also be tempting, but, with about 900 R pack-\nages using spatial data handling classes and objects, would far exceed the bounds of \na single article. The R statistical programming language and environment has been used for han-\ndling and analysing spatial data since its inception, partly building on its heritage \nfrom S and S-Plus. When the conceptualization of spatial data was introduced in \nthe sp package (Pebesma and Bivand 2005, 2020, on the Comprehensive R Archive \nNetwork (CRAN) since 2005), it was expected that some packages would adopt its \nclasses. Some years later, adoption rates had picked up strongly, as had use of the \nsp-based packages rgdal (Bivand et al. 2020, on CRAN since 2003) for input/output \nand rgeos (Bivand and Rundel 2020, on CRAN since 2011) for geometric manipula-\ntion of vector data. Packages depending on sp classes continue to require support as more mod-\nern data representations have been introduced in the sf (Pebesma 2018, 2020a, on \nCRAN since 2016) and stars (Pebesma 2020c, on CRAN since 2018) packages. 1 3 1  Introduction The sf package provides the data reading and writing, and geometry manipulation \nfunctionalities found in rgdal and rgeos, and an alternative class representation \nfor vector data based on the Simple Features standard (Herring 2011; ISO 2004). The stars package adds facilities for handling spatial and spatio-temporal raster and \nvector data, building in part on work with the spacetime package (Pebesma 2012, \n2020b, on CRAN since 2010) and on raster handling in the raster package (Hij-\nmans 2020a, on CRAN since 2010). The raster package will not be discussed in this \nreview, mostly because information in the Github “rspatial” organization (https​://\ngithu​b.com/rspat​ial) repositories suggests that development is in flux and that raster \nis becoming a new package called terra (Hijmans 2020b). Discussion here will con-\ncentrate on packages developed and maintained by the Github “r-spatial” organiza-\ntion (https​://githu​b.com/r-spati​al) of which I am a member. Newer visualization packages, such as tmap (Tennekes 2018, 2020, on CRAN \nsince 2014), mapview (Appelhans et al. 2020, on CRAN since 2015) and cartog-\nraphy (Giraud and Lambert 2016, 2017, 2020, on CRAN since 2015), give broader \nscope for data exploration and communication. Modelling and analysis packages \ndemonstrate the considerable range of implementations now available and are often \nsupported with additional code provided as supplementary material to journal arti-\ncles, for example in the Journal of Statistical Software spatial statistics special issue \n(Pebesma et al. 2015). The availability of software and scripts provides a helpful \nmechanism supporting reproducible research and hands-on reviewing in which \nreaders can read the code and scripts used in calculating the results presented in \npublished work (see, for example, Sawicka et al. 2018; Lovelace and Ellison 2018; \nEvangelista and Beskow 2018, in one issue of the R Journal). Some packages are used by others in turn forming dependency trees; because of \nthese dependencies, we can speak of an ecosystem. Class representations of data 1 3 Progress in the R ecosystem for representing and handling spatial… 517 are central, with the data frame conceptualization shaping much of the whole R \necosystem. For the modelling infrastructure to perform correctly, the relationships \nbetween objects containing data and the formula representations of models are cru-\ncial. Because both sp and sf provide similar interfaces for the data= arguments for \nmodel fitting functions by behaving as data.frame objects, transition from sp \nto sf representations is convenient by design. 1  Introduction The spdep package (Bivand 2020b, \non CRAN since 2002) for exploratory spatial data analysis and the new spatialreg \npackage (Bivand and Piras 2019, on CRAN since 2019) for spatial regression (split \nout of spdep) have been adapted to accommodate geometries held in sf classes, so \nboth approaches are viable. Other packages, such as mapview, tmap or stplanr \n(Lovelace et al. 2020) have been revised to permit use with both sp and sf objects. In \norder to retain backward compatibility, other central packages may choose to handle \nthe coexistence of sp and sf classes in the same way. Developers of new packages should choose to use sf and stars rather than sp, \nrgdal and rgeos (and terra rather than raster), but existing packages are free to \nadapt, or indeed to stay with sp, rgdal and rgeos, hoping that they may continue to \nbe maintained. Naturally, contributions to maintenance and development from those \nusing software on which one’s work depends are among the “prices” to be “paid” for \nopen-source software, so “hope” may involve commitment to take responsibility for \nmaintenance. If a maintainer is unable to continue in service, software like R add-on \npackages is termed “orphaned”, but may be adopted. This occurred very recently \nwith an R linear programming package lpSolve, which has been adopted by Gábor \nCsárdi to widespread relief and gratitude. Because there is no corporation tasked \nwith maintaining most R add-on packages, their future use has to depend on the user \ncommunity. One key reason why sf is much easier to maintain is that it was written using the \nRcpp package (Eddelbuettel et al. 2011, 2020; Eddelbuettel 2013; Eddelbuettel and \nBalamuta 2018, on CRAN since 2008) to interface C++ code in GDAL and now \nPROJ (both C++11), and GEOS code through the C API, whereas rgdal and rgeos \nhave more fragile handcrafted interfaces originally written for C99 and C++98. Maintaining Rcpp/C++11 interfaces is very much easier than older adaptations not \nas well understood by younger developers. However, the choice of C++ interfaces is \nnot necessarily robust (Kalibera 2019). Since there is a viable alternative to rgdal and rgeos, a fallback in the future for \nsp-dependent packages will be to use sf for reading and writing data, and geometry \nmanipulation, and to coerce to sp classes before passing to existing modelling code \nif sp classes are needed. 1  Introduction Maintainers of actively developed packages using sp vec-\ntor classes intensively and that are also impacted by changes in coordinate reference \nsystems (Sect. 4) are advised to consider transitioning to sf, as substantial revisions \nwill be needed anyway. In this review and prospect, the progress seen over the last 20  years will be \ndescribed, together with the unexpected consequences it engendered. The emer-\ngence of new technologies and standards has also led to the desirability of re-imple-\nmentation of data representations and analysis functionality, some of which is now \nin place and which will be described. Changes also impact the open-source librar-\nies on which core spatial functionality is built, leading to challenges in ensuring 1 3 3 3 518 R. S. Bivand backward compatibility. This will be shown using the example of coordinate refer-\nence systems. In terms of positionality, much of what follows will be presented from \nthe point of view of the author, documented as far as possible from email exchanges \nand similar sources. It is more than likely that other participants in the development \nof the R-spatial community may recall things differently, and of course, I acknowl-\nedge that this description is bound to be partial. 1  Typically plastic boxes containing static software licences attached to a computer port, often the paral-\nlel port otherwise used for printers in the pre-USB era. 2  \u0007Spatial data classes in the sp package CRAN listings show tripack (Renka and Geb-\nhardt 2020) and akima (Akima and Gebhardt 2020)—both with non-open-source 1 3 Progress in the R ecosystem for representing and handling spatial… 519 licences—available from August 1998 ported by Albrecht Gebhardt; ash and sgeo-\nstat (Majure and Gebhardt 2016) followed in April 1999. The spatial package was \navailable as part of MASS (the software supporting the four editions of Venables \nand Ripley 2002), also ported in part by Albrecht Gebhardt. In the earliest period, \nCRAN administrators helped practically with porting and publication. Albrecht and \nI presented an overview of possibilities of using R for research and teaching in spa-\ntial analysis and statistics in August 1998, subsequently published in this journal as \nBivand and Gebhardt (2000). Rowlingson and Diggle (1993) describe the S-PLUS version of splancs (Rowl-\ningson and Diggle 2017) for point pattern analysis. I had contacted Barry Rowl-\ningson in 1997, but only moved forward with porting as R’s package mechanism \nmatured. In September 1998, I wrote to him: “It wasn’t at all difficult to get things \nrunning, which I think is a result of your coding, thank you!” However, I added this \nspeculation: “An issue I have thought about a little is whether at some stage Albrecht \nand I wouldn’t integrate or harmonize the points and pairs objects in splancs, spa-\ntial and sgeostat—they aren’t the same, but for users maybe they ought to appear to \nbe so”. This concern with class representations for geographical data turned out to \nbe fruitful. A further step was to link GRASS and R, described in Bivand (2000), and fol-\nlowed up at several meetings and working closely with Markus Neteler. The inter-\nface has evolved, and its current status is presented by Lovelace et al. (2019, chap-\nter 9). A consequence of this work was that the CRAN team suggested that I attend a \nmeeting in Vienna in early 2001 to talk about the GRASS GIS interface. The meet-\ning gave unique insights into the dynamics of R development, and very valuable \ncontacts. Later the same year Luc Anselin and Serge Rey asked me to take part in a \nworkshop in Santa Barbara, which again led to many fruitful new contacts; my talk \neventually appeared as Bivand (2006), but the contacts made at the workshop were \nvery useful. 2  \u0007Spatial data classes in the sp package In the early and mid-1990s, those of us who were teaching courses in spatial analysis \nbeyond the direct application of geographical information systems (GIS) found the \npaucity of software limiting. In institutions with funding for site licences for GIS, it \nwas possible to write or share scripts for Arc/Info (in AML), ArcView (in Avenue) \nor later in Visual Basic for ArcGIS. If site licences and associated dongles1 used \nin the field were a problem (including problems for students involved in fieldwork \nin research projects), there were few alternatives, but opportunities were discussed \non mailing lists. One of these was the AI-Geostats listserve/mailing list started by \nGregoire Dubois in 1995; AI meant Arc/Info. Another community gathered around \nGRASS GIS and its transition to open-source development; it hosted, among other \nthings, src.contrib and src.garden directory trees with analysis tools (see \ncode stored in the https​://githu​b.com/OSGeo​/grass​-legac​y repository). From late 1996, the R programming language and environment began to be seen \nas an alternative for teaching and research involving spatial analysis by a few people \nincluding the author and Albrecht Gebhardt. R uses much of the syntax of S, then \navailable commercially as S-Plus, but was and remains free to install, use and extend \nunder the GNU General Public License (GPL). In addition, it could be installed \nportably across multiple operating systems, including Windows and Apple Mac OS. At about the same time, the S-Plus SpatialStats module was published (Kaluzny \net al. 1998), and a meeting occurred in Leicester in which many of those looking \nfor solutions took part. (My contribution was published in the meeting special issue \nBivand 1998.) Much of the porting of S code to R for spatial statistics was begun by Albrecht \nGebhardt as soon as the R package mechanism matured. The library() func-\ntion was upgraded in R 0.60 published in December 1997, and the Comprehensive \nR Archive Network was operating in January 1998. An exchange between Albre-\ncht Gebhardt and Thomas Lumley on the R-beta mailing list (https​://stat.ethz.ch/\npiper​mail/r-help/1997-Novem​ber/00188​2.html) shows that the package mecha-\nnism was not yet working predictably before 0.60 for contributed packages. Since \nteachers moving courses from S to R needed access to the S libraries previously \nused, porting was a crucial step. 2  \u0007Spatial data classes in the sp package Further progress during the intervening 4 years in the use of R in spa-\ntial econometrics was reported in Bivand (2002), building on Bivand and Gebhardt \n(2000), but preceding the release of the spdep package. During the second half of 2002, it seemed relevant to propose a spatial statis-\ntics paper session at the next Vienna meeting to be held in March 2003 (known as \nDistributed Statistical Computing (DSC) and led to useR! meetings), together with \na workshop to discuss classes for spatial data. I had reached out to Edzer Pebesma \nas an author of the stand-alone open-source program gstat (Pebesma and Wes-\nseling 1998); it turned out that he had just been approached to wrap the program for \nS-Plus. He saw the potential of the workshop immediately, and in November 2002 \nwrote in an email: “I wonder whether I should start writing S classes. I’m afraid \nI should”. Virgilio Gómez-Rubio had been developing two spatial packages, RAr-\ncInfo (Gómez-Rubio and López-Quílez 2005; Gómez-Rubio 2011) and DCluster \n(Gómez-Rubio et al. 2005, 2015), and was committed to participating. Although he \ncould not get to the workshop, Nicholas Lewin-Koh wrote in March 2003 that: “I \nwas looking over all the DSC material, especially the spatial stuff. I did notice, after \nlooking through peoples’ packages that there is a lot of duplication of effort. My \nsuggestion is that we set up a repository for spatial packages similar to the Biocon-\nductor mode, where we have a base spatial package that has S-4-based methods and 1 3 3 520 R. S. Bivand classes that are efficient and general.” Straight after the workshop, a collaborative \nrepository for the development of software using SourceForge was established, and \nthe R-sig-geo mailing list (still with over 3600 subscribers) was created to facilitate \ninteraction. classes that are efficient and general.” Straight after the workshop, a collaborative \nrepository for the development of software using SourceForge was established, and \nthe R-sig-geo mailing list (still with over 3600 subscribers) was created to facilitate \ninteraction. So the mandate for the development of the sp package emerged in discussions \nbetween interested contributors before, during and especially following the 2003 \nVienna workshop; most of us met at Pörtschach am Wörthersee in October 2003 \nat a meeting organized by Albrecht Gebhardt. 2  \u0007Spatial data classes in the sp package Coding meetings were organized by \nBarry Rowlingson in Lancaster in November 2004 and by Virgilio Gómez-Rubio \nin Valencia in May 2005, at both of which the class definitions and implementa-\ntions were stress-tested and often changed radically; the package was first published \non CRAN in April 2005. The underlying model adopted was for S4 (new style) \nclasses to be used, for \"Spatial\" objects, whether raster or vector, to behave like \n\"data.frame\" objects, and for visualization methods to make it easy to show the \nobjects. Package developers could choose whether they would use sp classes and \nmethods directly, or rather use those classes for functionality that they did not pro-\nvide themselves. The spatstat package (Baddeley and Turner 2005; Baddeley et al. 2015, 2020) was an early example of such object conversion (known as coercion in \nS and R) to and from sp classes and classes, with the coercion methods published in \nthe maptools package (Bivand and Lewin-Koh 2020).i Reading and writing ESRI Shapefiles had been possible using the maptools \npackage (Bivand and Lewin-Koh 2020) available from CRAN since August 2003, \nbut rgdal, on CRAN from November 2003, interfacing the external GDAL library \n(Warmerdam 2008) and first written by Tim Keitt, initially only supported accessing \nand reading raster data. Further code contributions by Barry Rowlingson for han-\ndling projections using the external PROJ.4 library and the vector drivers in the then \nOGR part of GDAL were folded into rgdal, permitting reading vector and raster \ndata into sp-objects and writing from sp-objects. For vector data, it became possible \nto project coordinates, and in addition to transform them where datum specifications \nwere available. Until 2019, the interfaces to GDAL and PROJ had been relatively \nstable, and upstream changes had not led to breaking changes for users of pack-\nages using sp classes or rgdal functionalities, although they have involved signifi-\ncant maintenance effort. The final part of the framework for spatial vector data han-\ndling was the addition of the rgeos package interfacing the external GEOS library \nin 2011, thanks to Colin Rundell’s 2010 Google Summer of Coding project. The \nrgeos package provided vector topological predicates and operations typically found \nin GIS such as intersection; note that by this time, both GDAL and GEOS used the \nSimple Features vector representation internally. By the publication of Bivand et al. 1 3 readxl\nwithr\nleaflet\ntidyr\nglue\nhtmlwidgets\nrjson\nstringi\nshiny\nrlang\ncrayon\nrvest\ncovr\nDT\ntibble\ndplyr\ntidyverse\nbroom\ntidyselect\nmaps\nhttr\nxml2\nassertthat\nhtmltools\nknitr\nrmarkdown\ndigest\nR6\nspelling\npurrr\nmagrittr\ncurl\nlubridate\nRCurl\nreadr\nsf\ntestthat\ndata.table\nXML\ndevtools\nrstudioapi\nkableExtra\nforeign\ntinytest\nroxygen2\nstringr\nyaml\njsonlite\nvdiffr\nshinyjs readxl\nwithr\nleaflet\ntidyr\nglue\nhtmlwidgets\nrjson\nstringi\nshiny\nrlang\ncrayon\nrvest\ncovr\nDT\ntibble\ndplyr\ntidyverse\nbroom\ntidyselect\nmaps\nhttr\nxml2\nassertthat\nhtmltools\nknitr\nrmarkdown\ndigest\nR6\nspelling\npurrr\nmagrittr\ncurl\nlubridate\nRCurl\nreadr\nsf\ntestthat\ndata.table\nXML\ndevtools\nrstudioapi\nkableExtra\nforeign\ntinytest\nroxygen2\nstringr\nyaml\njsonlite\nvdiffr\nshinyjs geosphere\npedometrics\ndeldir\nstabs\nakima\nspacetime\nmaptools\nspData\nFRK\nspam\nRStoolbox\nbiomod2graph4lg\nRandomFields\nspatialreg\ndismo\nINLA\nspatstat\nSSDM\nsp\ngeoR\nENMTools\nCARBayes\nGSIF\ngstat\nmuHVT\nBiodiversityR\nrnaturalearth\necospat\nfields\nspdep\nagricolae\nrgeos\nsplancs\nadehabitatLT\nrdwd\nrasterVis\ntree\nrgdal\ntrajectories\nspatialEco\ninlabru\nraster\nrworldmap\ngdistance\nmapdata\nRgoogleMaps\nCSTools\nplotKML\nsnowfall Fig. 1   Wordclouds of CRAN and BioConductor package dependencies, August 2020, left panel: cluster \n6 (pagerank range 0.002196–0.000070), right panel: cluster 2 (pagerank range 0.022419–0.000390, sf = \n0.000972) Fig. 1   Wordclouds of CRAN and BioConductor package dependencies, August 2020, left panel: cluster \n6 (pagerank range 0.002196–0.000070), right panel: cluster 2 (pagerank range 0.022419–0.000390, sf = \n0.000972) among the cumulated packages held on CRAN and those published by the Biocon-\nductor project. The left panel shows cluster 6, the spatial cluster with sp having a \npagerank of 0.002196, while the right panel shows cluster 2, which is dominated by \npackages developed by RStudio, a commercial company. The sf package is in cluster \n2, with a pagerank score of 0.000972, most likely in that cluster because it itself uses \nmany of the functionalities of RStudio packages. The two word clouds are scaled by \nthe largest pagerank of included packages, so the scales differ by almost an order of \nmagnitude. 2  https​://githu​b.com/rwinl​ib/gdal2​, https​://githu​b.com/rwinl​ib/gdal3​. 2  \u0007Spatial data classes in the sp package (2008), a few packages not written or main-\ntained by the book authors and their nearest collaborators had begun to use sp \nclasses. By the publication of the second edition (Bivand et al. 2013), we had seen \nthat the number of packages depending on sp, importing from and suggesting it (in \nCRAN terminology for levels of dependency) had grown strongly. In late 2014, de \nVries (2014) looked at CRAN package clusters from a page rank graph and found a \nclear spatial cluster that we had not expected. Figure 1 shows word clouds with char-\nacter sizes proportional to pagerank scores for two clusters found in August 2020 1 3 Progress in the R ecosystem for representing and handling spatial… 521 3  \u0007Spatial data classes in the sf and stars packages The raster package complemented sp for handling raster objects and their inter-\nactions with vector objects. It added to input/output using GDAL through rgdal, \nand better access to NetCDF files. It may be mentioned in passing that thanks to \nhelp from CRAN administrators and especially Brian Ripley, CRAN binary builds \nof rgdal for Windows and Apple Mac OSX became available from 2006, but with \na limited set of vector and raster drivers. Support from CRAN adminstrators and \nthe maintainers of the Github rwinlib/gdal2 and rwinlib/gdal3 repositories2 remains \ncentral to making packages available to users who are not able to install R source \npackages themselves, particularly linking to external libraries. Initially, raster was \nwritten in R using functionalities in sp and rgdal with rgeos coming later. It used a \nfeature of GDAL raster drivers permitting the successive reading of subsets of ras-\nters by row and column, allowing the processing of much larger objects than could \nbe held in memory. In addition, the concepts of bricks and stacks of rasters were \nintroduced, diverging somewhat from the sp treatment of raster bands as stacked \ncolumn vectors in a data frame. 1 3 522 R. S. Bivand As raster evolved, two other packages emerged raising issues with the ways in \nwhich spatial objects had been conceptualized in sp. The rgeos package used the C \napplication programming interface (API) to the C++ GEOS library, which is itself a \ntranslation of the Java Topology Suite (JTS). While the GDAL vector drivers did use \nthe standard Simple Features representation of vector geometries, it was not strongly \nenforced. This laxity now seems most closely associated with the use of ESRI Shape \nfiles as a de facto file standard for representation, in which some Simple Features \nare not consistently representable.3 Both JTS and GEOS required a Simple Features \ncompliant representation and led to the need for curious and fragile adaptations. For \nexample, these affected the representation of sp \"Polygons\" objects, which were \noriginally conceptualized after the Shapefile specification: ring direction determined \nwhether a ring was exterior or interior (a hole), but no guidance was given to show \nwhich exterior ring holes might belong to. As R provides a way to add a character \nstring comment to any object, such comments were added to each \"Polygons\" \nobject encoding the necessary information. 3  \u0007Spatial data classes in the sf and stars packages In this way, GEOS functionality could \nbe used, but the fragility of vector representation in sp was made obvious.f Another package affecting thinking about representation was spacetime, stacking \ncolumns for regular spatio-temporal objects with space varying faster than time. So \na single earth observation band observed repeatedly would be stored in a single col-\numn in a data frame, rather than in the arguably more robust form of a four-dimen-\nsional array, with the band taking one position on the final dimension. The second \nedition of Bivand et al. (2013) took up all of these issues in one way or another, but \nafter completing a spatial statistics special issue of the Journal of Statistical Soft-\nware (Pebesma et al. 2015), it was time to begin fresh implementations of classes for \nspatial data. 3  For recent examples see https​://githu​b.com/OSGeo​/gdal/issue​s/1787 and https​://githu​b.com/r-spati​al/\nsf/issue​s/1121. Progress in the R ecosystem for representing and handling spatial… 523 > df <- data.frame(a=letters[1:3], b=1:3)\n> df$c <- list(d=1, e=\"1\", f=TRUE)\n> str(df)\n’data.frame’: 3 obs. of\n3 variables:\n$ a: chr\n\"a\" \"b\" \"c\"\n$ b: int\n1 2 3\n$ c:List of 3\n..$ d: num 1\n..$ e: chr \"1\"\n..$ f: logi TRUE > df <- data.frame(a=letters[1:3], b=1:3)\n> df$c <- list(d=1, e=\"1\", f=TRUE)\n> str(df)\n’data.frame’: 3 obs. of\n3 variables:\n$ a: chr\n\"a\" \"b\" \"c\"\n$ b: int\n1 2 3\n$ c:List of 3\n..$ d: num 1\n..$ e: chr \"1\"\n..$ f: logi TRUE ’data.frame’: 3 obs. of\n3 variables: $ a: chr\n\"a\" \"b\" \"c\" $ b: int\n1 2 3 $ c:List of 3 ..$ d: num 1 ..$ e: chr \"1\" ..$ f: logi TRUE ..$ f: logi TRUE At useR! in 2016, list columns were declared “tidy”, using examples including \nthe difficulty of encoding polygon interior rings in non-list columns. The decision \nto accommodate “tidy” workflows as well as base-R workflows had already been \nmade, as at least some users only know how to use “tidy” workflows. Pebesma \n(2018) showed the status of the sf towards the end of 2017, with a geometry list \ncolumn containing R wrappers around objects adhering to Simple Features speci-\nfication definitions. Note also that from R 4.0.0, data.frame() does not con-\nvert character columns to factor as it did previously (Hornik 2020); column \n\"a\" is character in R 4 or later, and factor before R 4. 3.1  \u0007Simple Features in R It was clear that vector representations needed urgent attention, so the sf package \nwas begun, aiming to implement the most frequently used parts of the specification \n(ISO 2004; Kralidis 2008; Herring 2011). Development was supported by a grant \nfrom the then newly started R Consortium, which brings together R developers and \nindustry members. A key breakthrough came at the useR! 2016 conference, follow-\ning an earlier decision to re-base vector objects on data frames, rather than as in sp \nto embed a data frame inside a spatial object. Although data frame objects in S and \nR have always been able to take list columns as valid columns, such list columns \nwere not seen as “tidy” (Wickham 2014): 3 Progress in the R ecosystem for representing and handling spatial… > library(sf) The feature geometries are stored in numeric vectors, matrices or lists of \nmatrices and may also be subject to arithmetic operations. Features are held in \nthe \"XY\" class if two-dimensional, or \"XYZ\", \"XYM\" or \"XYZM\" if such coor-\ndinates are available (\"Z\" is usually treated as height and \"M\" as some meas-\nure, perhaps accuracy; both need to have specified units); all single features are \n\"sfg\" (Simple Features geometry) objects, with arithmetic and other operators: > pt1 <- st_point(c(1,3))\n> pt2 <- pt1 + 1\n> pt3 <- pt2 + 1\n> str(pt3)\n’XY’ num [1:2] 3 5 > pt1 <- st_point(c(1,3))\n> pt2 <- pt1 + 1\n> pt3 <- pt2 + 1\n> str(pt3)\n’XY’ num [1:2] 3 5 Geometries may be represented as “Well-Known Text” (WKT): > st_as_text(pt3)\n[1] \"POINT (3 5)\" > st_as_text(pt3)\n[1] \"POINT (3 5)\" > st_as_text(pt3)\n[1] \"POINT (3 5)\" > st_as_text(pt3)\n[1] \"POINT (3 5)\" or as “Well-Known Binary” (WKB) as in database “binary large objects” \n(BLOBs), resolving the problem of representation when working with GDAL vector \ndrivers and functions, and with GEOS predicates and topological operations: 1 3 3 524 R. S. Bivand _\n_\ny(p\n)\n[1] 01 01 00 00 00 00 00 00 00 00 00 08 40 00 00 00 00 00 00 14 40 A column of Simple Features geometries (\"sfc\") is constructed as a list of \"sfg\" \nobjects, which do not have to belong to the same Simple Features category; here, we \nassign the Web Mercator CRS to indicate that the points are projected to the plane, with \nposition measured in metres: > pt_sfc <- st_as_sfc(list(pt1, pt2, pt3), crs=3857)\n> str(pt_sfc)\nsfc_POINT of length 3; first list element:\n’XY’ num [1:2] 1 3 When sf was written, the units package was available (Pebesma et al. > library(sf) 2016, 2020) \nand could utilize the metric of the coordinates given in the declared CRS, so here inter-\npoint distances are measured in metres: > st_distance(pt_sfc)\nUnits: [m]\n[,1]\n[,2]\n[,3]\n[1,] 0.000000 1.414214 2.828427\n[2,] 1.414214 0.000000 1.414214\n[3,] 2.828427 1.414214 0.000000 [3,] 2.828427 1.414214 0.000000 If we re-specify the points as geographical coordinates in decimal degrees on the \nWGS84 ellipsoid, the distances will be given as metres, but measured over the ellipsoid: > pt_sfc1 <- st_as_sfc(list(pt1, pt2, pt3), crs=4326)\n> st_distance(pt_sfc1)\nUnits: [m]\n[,1]\n[,2]\n[,3]\n[1,]\n0.0 156759.1 313424.7\n[2,] 156759.1\n0.0 156665.6\n[3,] 313424.7 156665.6\n0.0 The most recent R Consortium grant covers the extension of analysis and data handling \nto global data representation; Pebesma and Dunnington (2020) present a roadmap for the \nuse of the s2 library for operations on geographical coordinates and its integration in sf. > mat <- matrix(0, 3, 3)\n> mat[1,2] <- mat[2,1] <- s2::s2_distance(st_as_text(pt1), st_as_text(pt2))\n> mat[1,3] <- mat[3,1] <- s2::s2_distance(st_as_text(pt1), st_as_text(pt3))\n> mat[2,3] <- mat[3,2] <- s2::s2_distance(st_as_text(pt2), st_as_text(pt3))\n> mat\n[,1]\n[,2]\n[,3]\n[1,]\n0.0 157106.0 314116.3\n[2,] 157106.0\n0.0 157010.4\n[3,] 314116.3 157010.4\n0.0 1 3 1 3 Progress in the R ecosystem for representing and handling spatial… 525 Finally, an \"sfc\" object, a geometry column, can be added to a data.frame \nobject using st_geometry(), which sets a number of attributes on the object and \ndefines it as also being an \"sf\" object (the \"agr\" attribute if populated shows \nhow observations on non-geometry columns should be understood): > st_geometry(df) <- pt_sfc\n> str(df)\nClasses ‘sf’ and ’data.frame’: 3 obs. of\n4 variables:\n$ a\n: chr\n\"a\" \"b\" \"c\"\n$ b\n: int\n1 2 3\n$ c\n:List of 3\n..$ d: num 1\n..$ e: chr \"1\"\n..$ f: logi TRUE\n$ geometry:sfc_POINT of length 3; first list element:\n’XY’ num\n1 3\n- attr(*, \"sf_column\")= chr \"geometry\"\n- attr(*, \"agr\")= Factor w/ 3 levels \"constant\",\"aggregate\",..: NA NA NA\n..- attr(*, \"names\")= chr [1:3] \"a\" \"b\" \"c\" Classes ‘sf’ and ’data.frame’: 3 obs. 4  util=\"ogr2ogr\", options=\"-nlt CONVERT_TO_LINEAR\".\n5  https​://www.ogc.org/stand​ards/geopa​ckage​. > library(sf) of\n4 variables: $ a\n: chr\n\"a\" \"b\" \"c\" :List of 3 c\n:List of 3 ..$ d: num 1 ..$ e: chr \"1\" ..$ f: logi TRUE - attr(*, \"agr\")= Factor w/ 3 levels \"constant\",\"aggregate\",..: NA NA NA\n..- attr(*, \"names\")= chr [1:3] \"a\" \"b\" \"c\" The sf package implements all of the Simple Features geometry categories, but \nsome geometries need be converted to be used inside R, with, for example, the gdal_\nutils() function to convert curve geometries in an input file to linear geometries.4 \nMany of the functions in the sf package begin with st_ as a reference to the same \nusage in PostGIS, where the letters were intended to symbolize “spatial type”.i The vector file creation and reading functionality found in rgdal is also avail-\nable in sf, with substantial improvements with regard to creating spatial database \ntables and reading from databases. Writing a GeoPackage5 is as easy or easier than \nin rgdal, as the layer= and driver= arguments can be inferred from the given \nfile name. List columns cannot be written, because R list columns are not bound to \nbe consistently of the same type. > tf <- tempfile(fileext=\".gpkg\")\n> st_write(df, dsn=tf, quiet=TRUE)\nWarning message:\nIn clean_columns(as.data.frame(obj), factorsAsCharacter) :\nDropping column(s) c of class(es) list > tf <- tempfile(fileext=\".gpkg\")\n> st_write(df, dsn=tf, quiet=TRUE)\nWarning message:\nIn clean_columns(as.data.frame(obj), factorsAsCharacter) :\nDropping column(s) c of class(es) list > tf <- tempfile(fileext=\".gpkg\")\n> st_write(df, dsn=tf, quiet=TRUE)\nWarning message:\nIn clean_columns(as.data.frame(obj), factorsAsCharacter) :\nDropping column(s) c of class(es) list The standard reading method is st_read, providing a similar functionality to \nthat in rgdal, with a number of differences related to character string encoding that \nwill cease to matter when users migrate to modern formats such as GeoPackage. 3 R. S. > library(sf) Bivand 526 > df1 <- st_read(dsn=tf, quiet=TRUE)\n> df1\nSimple feature collection with 3 features and 2 fields\ngeometry type:\nPOINT\ndimension:\nXY\nbbox:\nxmin: 1 ymin: 3 xmax: 3 ymax: 5\nprojected CRS:\nWGS 84 / Pseudo-Mercator\na b\ngeom\n1 a 1 POINT (1 3)\n2 b 2 POINT (2 4)\n3 c 3 POINT (3 5) In a “tidy” workflow, read_sf() can be used, returning a \"tibble-sf\" \ninheriting from a \"tbl_df\" object rather than from a data frame, and not con-\nverting character string columns into \"factor\" categorical variables by default \n(before R 4.0.0): > read_sf(dsn=tf) %>% dplyr::filter(a == \"c\")\nSimple feature collection with 1 feature and 2 fields\ngeometry type:\nPOINT\ndimension:\nXY\nbbox:\nxmin: 3 ymin: 5 xmax: 3 ymax: 5\nprojected CRS:\nWGS 84 / Pseudo-Mercator\n# A tibble: 1 x 3\na\nb\ngeom\n* <chr> <int> <POINT [m]>\n1 c\n3\n(3 5) sf also integrates GEOS topological predicates and operations into the same \nframework, replacing the rgeos package for access to GEOS functionality. The pre-\ncision and scale defaults differ between sf and rgeos slightly; both remain fragile \nwith respect to invalid geometries, of which there are many in circulation. From \nGEOS 3.8, both rgeos and sf offer functions to create valid geometries from invalid \nones; prior to GEOS 3.8, a function in lwgeom was used. Most recently, measure-\nments and operations on non-planar geometries have been moved from lwgeom to s2 \n(Dunnington et al. 2020; Pebesma and Dunnington 2020). Native support for units \nin sf objects carries through to measurements on outputs of topological operations: > buf_df1 <- st_buffer(df1, dist=0.3)\n> st_area(buf_df1)\nUnits: [m^2]\n[1] 0.2826142 0.2826142 0.2826142\n> plot(st_geometry(buf_df1))\n> plot(st_geometry(df1), add=TRUE, pch=4) 1 3 1 3 3 Progress in the R ecosystem for representing and handling spatial… 527 Fig. 2   Three points and three buffers plotted with geometry-specific plot methods for \"sfc\" objects Fig. 2   Three points and three buffers plotted with geometry-specific plot methods for \"sfc\" objects The sf package provides simple base graphics plotting methods. Those for just \nthe \"sfc\" column retrieved by st_geometry() do not take over the layout of \nthe graphics device, but for the \"sf\" object, they do, showing multiple non-geom-\netry columns in separate displays. Figure 2 shows a simple plot of the three buffer \npolygons overplotted with the three points. 3.2  \u0007Raster representations Like sf, the stars package for scalable, spatio-temporal tidy arrays was supported \nby an R Consortium grant. Spatio-temporal arrays were seen as an alternative way \nof representing multivariate spatio-temporal data from the choices made in the spa-\ncetime package, where a two-dimensional data frame contained stacked observation \npositions in space within stacked time points or intervals. The proposed arrays might \ncollapse to a raster layer if only one variable was chosen for one time point or inter-\nval. More important, the development of the package was extended to accommodate \na backend for earth data processing in which the data are retrieved and re-sampled \nas needed from servers, most often cloud-based servers. In most cases, these would \nbe raster geometries, but the array representation also handles irregular geometries \nthrough time. The R Consortium support was chiefly used to let contributors meet to \nmake progress on concepts and implementation. This example only covers a multiband raster taken from a Landsat 7 view of a small \npart of the Brazilian coast. In the first part, a GeoTIFF file is read into memory, using \nthree array dimensions, two in planar space, the third across six bands: 1 3 1 3 R. S. Bivand 528 > library(stars)\n> fn <- system.file(\"tif/L7_ETMs.tif\", package = \"stars\")\n> L7 <- read_stars(fn)\n> L7\nstars object with 3 dimensions and 1 attribute\nattribute(s): stars object with 3 dimensions and 1 attribute\nattribute(s): L7_ETMs.tif\nMin. :\n1.00\n1st Qu.: 54.00\nMedian : 69.00\nMean\n: 68.91\n3rd Qu.: 86.00\nMax. :255.00\ndimension(s):\ns\ne\nu\nl\na\nv\nt\nn\ni\no\np\ns\ny\ns\nf\ne\nr\na\nt\nl\ne\nd\nt\ne\ns\nf\nf\no\no\nt\nm\no\nr\nf\nx\n1 349\n288776\n28.5 UTM Zone 25, Southern Hem... FALSE\nNULL [x]\ny\n1 352 9120761 -28.5 UTM Zone 25, Southern Hem... FALSE\nNULL [y]\nL\nL\nU\nN\nA\nN\nA\nN\nA\nN\nA\nN\n6\n1\nd\nn\na\nb L7_ETMs.tif\nMin. :\n1.00\n1st Qu.: 54.00\nMedian : 69.00\nMean\n: 68.91\n3rd Qu.: 86.00\nMax. :255.00\n( ) dimension(s):\ns\ne\nu\nl\na\nv\nt\nn\ni\no\np\ns\ny\ns\nf\ne\nr\na\nt\nl\ne\nd\nt\ne\ns\nf\nf\no\no\nt\nm\no\nr\nf\nx\n1 349\n288776\n28.5 UTM Zone 25, Southern Hem... 3.2  \u0007Raster representations FALSE\nNULL [x]\ny\n1 352 9120761 -28.5 UTM Zone 25, Southern Hem... FALSE\nNULL [y]\nL\nL\nU\nN\nA\nN\nA\nN\nA\nN\nA\nN\n6\n1\nd\nn\na\nb The bands can be operated on arithmetically, for example to generate a new object \ncontaining values of the normalized difference vegetation index through a function \napplied across the x and y spatial dimensions, using the st_apply abstraction: > ndvi <- function(x) (x[4] - x[3])/(x[4] + x[3])\n> (s2.ndvi <- st_apply(L7, c(\"x\", \"y\"), ndvi))\nstars object with 2 dimensions and 1 attribute\nattribute(s): > (s2.ndvi <- st_apply(L7, c(\"x\", \"y\"), ndvi))\nstars object with 2 dimensions and 1 attribute\nattribute(s):\nndvi\nMin. :-0.75342\n1st Qu.:-0.20301\nMedian :-0.06870\nMean\n:-0.06432\n3rd Qu.: 0.18667\nMax. : 0.58667\ndimension(s):\ns\ne\nu\nl\na\nv\nt\nn\ni\no\np\ns\ny\ns\nf\ne\nr\na\nt\nl\ne\nd\nt\ne\ns\nf\nf\no\no\nt\nm\no\nr\nf\nx\n1 349\n288776\n28.5 UTM Zone 25, Southern Hem... FALSE\nNULL [x]\ny\n1 352 9120761 -28.5 UTM Zone 25, Southern Hem... FALSE\nNULL [y] ( )\ns\ne\nu\nl\na\nv\nt\nn\ni\no\np\ns\ny\ns\nf\ne\nr\na\nt\nl\ne\nd\nt\ne\ns\nf\nf\no\no\nt\nm\no\nr\nf\nx\n1 349\n288776\n28.5 UTM Zone 25, Southern Hem... FALSE\nNULL [x]\ny\n1 352 9120761 -28.5 UTM Zone 25, Southern Hem... FALSE\nNULL [y] The same file can also be accessed using the proxy mechanism, which creates a link \nto the external entity, here a file: > L7p <- read_stars(fn, proxy=TRUE)\n> L7p\nstars_proxy object with 1 attribute in file:\n$L7_ETMs.tif\n[1] \"[...]/L7_ETMs.tif\"\ndimension(s): > L7p <- read_stars(fn, proxy=TRUE)\n> L7p\nstars_proxy object with 1 attribute in file:\n$L7_ETMs.tif\n[1] \"[...]/L7_ETMs.tif\" [1] \"[...]/L7_ETMs.tif\" dimension(s): ( )\ns\ne\nu\nl\na\nv\nt\nn\ni\no\np\ns\ny\ns\nf\ne\nr\na\nt\nl\ne\nd\nt\ne\ns\nf\nf\no\no\nt\nm\no\nr\nf\nx\n1 349\n288776\n28.5 UTM Zone 25, Southern Hem... FALSE\nNULL [x]\ny\n1 352 9120761 -28.5 UTM Zone 25, Southern Hem... 3.2  \u0007Raster representations FALSE\nNULL [y]\nL\nL\nU\nN\nA\nN\nA\nN\nA\nN\nA\nN\n6\n1\nd\nn\na\nb s\ne\nu\nl\na\nv\nt\nn\ni\no\np\ns\ny\ns\nf\ne\nr\na\nt\nl\ne\nd\nt\ne\ns\nf\nf\no\no\nt\nm\no\nr\nf\nx\n1 349\n288776\n28.5 UTM Zone 25, Southern Hem... FALSE\nNULL [x]\n1 352 9120761\n28 5 UTM Z\n25\nS\nh\nH\nFALSE\nNULL [ ] a\nt\nl\ne\nd\nt\ne\ns\nf\nf\no\no\nt\nm\no\nr\nf s\ne\nu\nl\na\nv\nt\nn\ni\no\np\ns\ny\ns\nf\ne\nr\na\nt\nl\ne\nd\nt\ne\ns\nf\nf\no\no\nt\nm\no\nr\nf\nx\n1 349\n288776\n28.5 UTM Zone 25, Southern Hem... FALSE\nNULL [x] x\n1 349\n288776\n28.5 UTM Zone 25, Southern Hem... FALSE\nNULL [x]\ny\n1 352 9120761 -28.5 UTM Zone 25, Southern Hem... FALSE\nNULL [y]\nL\nL\nU\nN\nA\nN\nA\nN\nA\nN\nA\nN\n6\n1\nd\nn\na\nb 3 Progress in the R ecosystem for representing and handling spatial… 529 The same function can also be applied across the same two spatial dimensions \nof the array, but no calculation is carried out until the data are needed and the out-\nput resolution known, with the command needed to create the output stored in the \nobject: > (L7p.ndvi = st_apply(L7p, c(\"x\", \"y\"), ndvi))\nstars_proxy object with 1 attribute in file:\n$L7_ETMs.tif\n[1] \"[...]/L7_ETMs.tif\" > (L7p.ndvi = st_apply(L7p, c(\"x\", \"y\"), ndvi)) > (L7p.ndvi = st_apply(L7p, c(\"x\", \"y\"), ndvi)\nstars_proxy object with 1 attribute in file:\n$L7 ETMs tif p\npp y\np,\n,\ny\n,\nstars_proxy object with 1 attribute in file:\n$L7_ETMs.tif [1] \"[...]/L7_ETMs.tif\" dimension(s):\ns\ne\nu\nl\na\nv\nt\nn\ni\no\np\ns\ny\ns\nf\ne\nr\na\nt\nl\ne\nd\nt\ne\ns\nf\nf\no\no\nt\nm\no\nr\nf\nx\n1 349\n288776\n28.5 UTM Zone 25, Southern Hem... FALSE\nNULL [x]\ny\n1 352 9120761 -28.5 UTM Zone 25, Southern Hem... 3.2  \u0007Raster representations FALSE\nNULL [y]\nL\nL\nU\nN\nA\nN\nA\nN\nA\nN\nA\nN\n6\n1\nd\nn\na\nb\ncall list:\n[[1]]\nst_apply(X = X, MARGIN = c(\"x\", \"y\"), FUN = ndvi) dimension(s): dimension(s):\ns\ne\nu\nl\na\nv\nt\nn\ni\no\np\ns\ny\ns\nf\ne\nr\na\nt\nl\ne\nd\nt\ne\ns\nf\nf\no\no\nt\nm\no\nr\nf\nx\n1 349\n288776\n28.5 UTM Zone 25, Southern Hem... FALSE\nNULL [x]\ny\n1 352 9120761 -28.5 UTM Zone 25, Southern Hem... FALSE\nNULL [y]\nL\nL\nU\nN\nA\nN\nA\nN\nA\nN\nA\nN\n6\n1\nd\nn\na\nb\ncall list:\n[[1]]\nst_apply(X = X, MARGIN = c(\"x\", \"y\"), FUN = ndvi) a\nt\nl\ne\nd\nt\ne\ns\nf\nf\no\no\nt\nm\no\nr\nf 1]]\n_apply(X = X, MARGIN = c(\"x\", \"y\"), FUN = ndvi) The array object can also be split, here on the band dimension, to yield a repre-\nsentation as six rasters in list form: > (x6 <- split(L7, \"band\"))\nstars object with 2 dimensions and 6 attributes\nattribute(s): stars object with 2 dimensions and 6 attributes\nattribute(s):\nX1\nX2\nX3\nX4\nMin. : 47.00\nMin. : 32.00\nMin. : 21.00\nMin. :\n9.00\n1st Qu.: 67.00\n1st Qu.: 55.00\n1st Qu.: 49.00\n1st Qu.: 52.00\nMedian : 78.00\nMedian : 66.00\nMedian : 63.00\nMedian : 63.00\nMean\n: 79.15\nMean\n: 67.57\nMean\n: 64.36\nMean\n: 59.24\n3rd Qu.: 89.00\n3rd Qu.: 79.00\n3rd Qu.: 77.00\n3rd Qu.: 75.00\nMax. :255.00\nMax. :255.00\nMax. :255.00\nMax. :255.00\nX5\nX6\nMin. :\n1.00\nMin. :\n1.00\n1st Qu.: 63.00\n1st Qu.: 32.00\nMedian : 89.00\nMedian : 60.00\nMean\n: 83.18\nMean\n: 59.98\n3rd Qu.:112.00\n3rd Qu.: 88.00\nMax. :255.00\nMax. :255.00\ndimension(s):\ns\ne\nu\nl\na\nv\nt\nn\ni\no\np\ns\ny\ns\nf\ne\nr\na\nt\nl\ne\nd\nt\ne\ns\nf\nf\no\no\nt\nm\no\nr\nf\nx\n1 349\n288776\n28.5 UTM Zone 25, Southern Hem... FALSE\nNULL [x]\ny\n1 352 9120761 -28.5 UTM Zone 25, Southern Hem... FALSE\nNULL [y] X1\nX2\nX3\nX4\nMin. : 47.00\nMin. : 32.00\nMin. : 21.00\nMin. 3.2  \u0007Raster representations :\n9.00\n1st Qu.: 67.00\n1st Qu.: 55.00\n1st Qu.: 49.00\n1st Qu.: 52.00\nMedian : 78.00\nMedian : 66.00\nMedian : 63.00\nMedian : 63.00\nMean\n: 79.15\nMean\n: 67.57\nMean\n: 64.36\nMean\n: 59.24\n3rd Qu.: 89.00\n3rd Qu.: 79.00\n3rd Qu.: 77.00\n3rd Qu.: 75.00\nMax. :255.00\nMax. :255.00\nMax. :255.00\nMax. :255.00\nX5\nX6\nMin. :\n1.00\nMin. :\n1.00\n1st Qu.: 63.00\n1st Qu.: 32.00\nMedian : 89.00\nMedian : 60.00\nMean\n: 83.18\nMean\n: 59.98\n3rd Qu.:112.00\n3rd Qu.: 88.00\nMax. :255.00\nMax. :255.00\ndimension(s):\ns\ne\nu\nl\na\nv\nt\nn\ni\no\np\ns\ny\ns\nf\ne\nr\na\nt\nl\ne\nd\nt\ne\ns\nf\nf\no\no\nt\nm\no\nr\nf dimension(s):\ns\ne\nu\nl\na\nv\nt\nn\ni\no\np\ns\ny\ns\nf\ne\nr\na\nt\nl\ne\nd\nt\ne\ns\nf\nf\no\no\nt\nm\no\nr\nf\nx\n1 349\n288776\n28.5 UTM Zone 25, Southern Hem... FALSE\nNULL [x]\ny\n1 352 9120761 -28.5 UTM Zone 25, Southern Hem... FALSE\nNULL [y] These rasters may also be subjected to arithmetical operations, and as may be \nseen, explicit arithmetic on the six rasters has the same outcome as applying the \nsame calculation to the three-dimensional array: 1 3 530 R. S. Bivand > x6$mean <- (x6[[1]] + x6[[2]] + x6[[3]] + x6[[4]] + x6[[5]] + x6[[6]])/6\n> xm <- st_apply(L7, c(\"x\", \"y\"), mean)\n> all.equal(xm[[1]], x6$mean)\n[1] TRUE > x6$mean <- (x6[[1]] + x6[[2]] + x6[[3]] + x6[[4]] + x6[[5]] + x6[[6]])/6\n> xm <- st_apply(L7, c(\"x\", \"y\"), mean)\n> all.equal(xm[[1]], x6$mean)\n[1] TRUE The extension to a gridded temporal dimension or to irregular spatial and tem-\nporal entities is not particularly difficult, but it remains to document the possibili-\nties of the package more fully. 3.3  \u0007Visualization The classInt package (Bivand 2020a) for finding thematic mapping class inter-\nvals is used directly in plot methods in sf and stars, and in the tmap (Tennekes \n2018, 2020) and cartography (Giraud and Lambert 2016, 2017, 2020) packages. Lapa et al. (2001) (Leprosy surveillance in Olinda, Brazil, using spatial analysis \ntechniques) made available the underlying data set of Olinda census tracts (setor) \nin the Corrego Alegre 1970-72 / UTM zone 25S projection (EPSG:22525); we \nwill use the data set and the deprivation variable to point to visualization alterna-\ntives now available. > olinda <- st_read(\"olinda.gpkg\", quiet=TRUE) The style= argument gives the choice of method used for finding the number \nof classes specified, with \"fisher\" being a natural breaks method, and \"bclust\" \nbagged clustering from the e1071 package (Meyer et al. 2019): > library(classInt) > library(classInt)\n> cI_fisher <- classIntervals(olinda$\"DEPRIV\", n=7, style=\"fisher\")\n> set.seed(1)\n> cI_bclust <- classIntervals(olinda$\"DEPRIV\", n=7, style=\"bclust\")\nCommittee Member: 1(1) 2(1) 3(1) 4(1) 5(1) 6(1) 7(1) 8(1) 9(1) 10(1)\nComputing Hierarchical Clustering > cI_bclust <- classIntervals(olinda$\"DEPRIV\", n=7, style=\"bclust\")\nCommittee Member: 1(1) 2(1) 3(1) 4(1) 5(1) 6(1) 7(1) 8(1) 9(1) 10(1)\nComputing Hierarchical Clustering > cI_bclust <- classIntervals(olinda$\"DEPRIV\", n=7, style=\"bclust\")\nCommittee Member: 1(1) 2(1) 3(1) 4(1) 5(1) 6(1) 7(1) 8(1) 9(1) 10(1) Committee Member: 1(1) 2(1) 3(1) 4(1) 5(1) 6(1) 7(1) 8(1) 9(1) 10(1)\nComputing Hierarchical Clustering For a long time, the RColorBrewer package (Neuwirth 2014) colour palettes \nwere the obvious choice for thematic cartography, but more recently other packages, \nsuch as rcartocolor (Nowosad 2019), have become available, often as supersets of \nthe RColorBrewer palettes: > pal <- rcartocolor::carto_pal(7, \"SunsetDark\")\n> plot(cI_fisher, pal, xlab=\"DEPRIV\", ylab=\"\")\n> plot(cI_bclust, pal, xlab=\"DEPRIV\", ylab=\"\") > pal <- rcartocolor::carto_pal(7, \"SunsetDark\") > plot(cI_fisher, pal, xlab=\"DEPRIV\", ylab=\"\")\n> plot(cI_bclust, pal, xlab=\"DEPRIV\", ylab=\"\") Figure 3 shows plots of the two class interval schemes, with obvious differences \nbetween the two styles. Both styles are attempting to balance low within-class vari-\nance and high between-class variance. 1 3 1 3 Progress in the R ecosystem for representing and handling spatial… 531 0.0\n0.2\n0.4\n0.6\n0.8\n1.0\n0.0\n0.2\n0.4\n0.6\n0.8\n1.0\nfisher\n0.0\n0.2\n0.4\n0.6\n0.8\n1.0\n0.0\n0.2\n0.4\n0.6\n0.8\n1.0\nbclust\nFig. 3   Deprivation by census tract in Olinda, Brazil; empirical cumulative distribution function and class \nintervals for two class intervals: left panel: natural breaks; right panel: bagged clustering Fig. 3   Deprivation by census tract in Olinda, Brazil; empirical cumulative distribution function and class \nintervals for two class intervals: left panel: natural breaks; right panel: bagged clustering Figure 4 shows maps of the same variable with the same class intervals chosen \ninternally using classInt functionality, with the same palette. The syntax is a little \ndifferent, and here, the sf plot method default key has been turned off to permit \nside-by-side display. This method is really best at providing glimpses of the data, \nrather than at creating complete figures, in contrast to the richer functions in cartog-\nraphy; both use base graphics. Fig. > plot(olinda[,\"DEPRIV\"], nbreaks=7, breaks=\"fisher\", pal=pal, key.pos=NULL,\n+\nmain=\"\")\n> library(cartography)\n> choroLayer(olinda, var=\"DEPRIV\", method=\"fisher-jenks\", nclass=7, col=pal,\n+\nlegend.values.rnd=3) > olinda <- st_read(\"olinda.gpkg\", quiet=TRUE) 6   Thematic maps with ggplot2: left panel: default map of a continuous variable; right panel: grati-\ncule removed and palette modified Fig. 7   Thematic maps with ggplot2, natural breaks class intervals: left panel: natural breaks class inter-\nvals; right panel: bagged clustering class intervals Fig. 7   Thematic maps with ggplot2, natural breaks class intervals: left panel: natural breaks class inter-\nvals; right panel: bagged clustering class intervals We can also display the bagged clustering class intervals beside the natural \nbreaks map, again using ggplot2; Figure 7 shows the maps, but because a con-\ntinuous scale is still enforced, all that changes is the position of the breaks on \nthe key. > g3 <- g1 + theme_void() + scale_fill_gradientn(colours=pal,\n+\nbreaks=round(cI_bclust$brks, 3)) > g3 <- g1 + theme_void() + scale_fill_gradientn(colours=pal,\n+\nbreaks=round(cI_bclust$brks, 3)) > olinda <- st_read(\"olinda.gpkg\", quiet=TRUE) 4   Deprivation by census tract in Olinda, Brazil, natural breaks class intervals; left panel: sf plot \nmethod; right panel: cartography choropleth map Fig. 4   Deprivation by census tract in Olinda, Brazil, natural breaks class intervals; left panel: sf plot \nmethod; right panel: cartography choropleth map 3 532 R. S. Bivand Fig. 5   tmap output, natural breaks class intervals; left panel: object o1 without boundaries; right panel: \nobject o2 with boundaries Fig. 5   tmap output, natural breaks class intervals; left panel: object o1 without boundaries; right panel: \nobject o2 with boundaries Figure 5 shows how tmap functions use grid graphics to permit graphics objects to \nbe updated and then plotted using tmap_arrange(). Once again, the same class \nintervals are chosen internally using classInt, with the same palette. The left panel is \nupdated on the right to add thin boundaries between census tracts. tmap also offers \nsmall multiples of facets, for example thematic maps of the same variable observed at \nsuccessive time periods using the same class intervals. An extensive discussion of the \nuse of tmap is provided by Lovelace et al. (2019). > library(tmap)\n> o1 <- tm_shape(olinda) + tm_fill(\"DEPRIV\", style=\"fisher\", n=7, palette=pal)\n> o2 <- o1 + tm_borders(lwd=0.8) > library(tmap)\n> o1 <- tm_shape(olinda) + tm_fill(\"DEPRIV\", style=\"fisher\", n=7, palette=pal)\n> o2 <- o1 + tm_borders(lwd=0.8) > library(tmap) The ggplot2 package (Wickham et al. 2020) provides the geom_sf() facil-\nity for mapping sf objects: > library(ggplot2)\n> g1 <- ggplot(olinda) + geom_sf(aes(fill=DEPRIV))\n> g2 <- g1 + theme_void() + scale_fill_gradientn(colours=pal,\n+\nbreaks=round(cI_fisher$brks, 3)) This approach also builds on grid graphics. It is possible to set a theme that \ndrops the arguably unnecessary graticule, but there is a lot of intervention \nrequired to get a simple map. To get proper class intervals involves even more \nwork, because the package takes specific, not general, positions on how graphics \nare observed. ColorBrewer, for example, eschews continuous colour scales based \non cognitive research, but ggplot2 enforces them by default for continuous vari-\nables. Figure 6 shows the default choice of palette, updated to remove the unnec-\nessary graticule and to use the user-specified palette and class intervals: 1 3 1 3 3 Progress in the R ecosystem for representing and handling spatial… 533 Fig. 6   Thematic maps with ggplot2: left panel: default map of a continuous variable; right panel: grati-\ncule removed and palette modified Fig. 3.4  \u0007Reverse dependencies of the sp and sf packages R packages can possess forward or reverse dependencies. Forward or upstream \ndependencies are typically on R itself, a small number of packages whose func-\ntionalities are used in the package in question (by loading and attaching the pack-\nage (dependencies) or just loading the namespace of the package (imports)), and \npossibly external software libraries. Reverse or downstream dependencies are \npackages that themselves use the package in question by loading and attaching \nit, only loading its namespace or using it on demand (suggests). sp and sf were \nwritten carefully to minimize forward dependencies, with sp only depending on \nand importing packages included in every R distribution by default and sf adding \nCRAN contributed packages Rcpp and units required to build the package and 1 3 3 534 R. S. Bivand classInt for class intervals, DBI for interfacing spatial databases and magrittr \nfor piped operations, where none of these extra forward dependencies draws in \nmany other packages. classInt for class intervals, DBI for interfacing spatial databases and magrittr \nfor piped operations, where none of these extra forward dependencies draws in \nmany other packages. In sp, the compiled code (written in C) is self-contained and is made available to \nother packages, chiefly rgdal and rgeos, to link to their compiled code. rgdal links \nto sp and to the external libraries PROJ and GDAL. GDAL itself links to PROJ and \ncan link to GEOS and many other libraries needed for specific drivers. The external \nsoftware versions used may be reported using *_extSoftVersion, here using \nthe :: operator to avoid attaching the packages being queried: > rgdal::rgdal_extSoftVersion()\nGDAL GDAL_with_GEOS\nPROJ\nsp\n\"3.1.2\"\n\"TRUE\"\n\"7.1.0\"\n\"1.4-4\" The versions vary between platforms and by the installation method used; as \npackage maintainer, I often run with pre-release or latest versions of external soft-\nware to attempt to detect and mitigate changes before they impact users’ workflows. For rgdal, the versions of GDAL, PROJ and sp are reported, together with a test \nshowing whether GDAL was built linking to GEOS, something that affects the \nbehaviour of some drivers. The report for rgeos is simpler, only listing the versions \nof GEOS itself and sp. 3.4  \u0007Reverse dependencies of the sp and sf packages > rgeos::rgeos_extSoftVersion()\nGEOS\nsp\n\"3.8.1\" \"1.4-4\" In the case of sf, and because it brings together access to the GDAL and GEOS \nexternal libraries through Simple Features representation for vector objects, the \nexternal software versions supported are the union of those seen above, omitting \nlinkage to a separate package defining classes for objects. In addition, it reports \nwhich API is used for PROJ, either proj.h or not (the earlier proj_api.h). > sf_extSoftVersion()\nGEOS\nGDAL\nproj.4 GDAL_with_GEOS\nUSE_PROJ_H\n\"3.8.1\"\n\"3.1.2\"\n\"7.1.0\"\n\"true\"\n\"true\" Table 1 shows the structure of reverse dependency counts for sp and sf. Recursive \ndependencies traverse through the whole CRAN dependency tree; the first column \nof the table shows counts of “depends” and “imports” dependencies counted across \nthe whole tree. These split into 1285 only involving sp, 232 involving both pack-\nages and 65 only involving sf. If we additionally include “suggests” dependencies, \nboth packages may be used at least indirectly by all CRAN packages. The two right \ncolumns show the same counts, but only for packages’ first-order dependencies on \nsp, sf or both. We can note that of these for “depends” and “imports” dependencies, \n459 only involve sp, 63 involve both packages and 121 only involve sf. In the first 1 3 Progress in the R ecosystem for representing and handling spatial… Progress in the R ecosystem for representing and handling spatial…\nTable 1   Reverse dependency \ncounts for sp and sf, August \n2020, for recursive and non-\nrecursive reverse dependencies \ntaken as “Depends” and \n“Imports” only, and with \n“Suggests”\nRecursive\nRecursive \nw/suggests\nNot recursive\nNot recursive \nw/suggests\nSum sp\n1517\n16,619\n522\n629\nSum sf\n297\n16,619\n184\n277\nOnly sp\n1285\n0\n459\n513\nOnly sf\n65\n0\n121\n161\nBoth\n232\n16,619\n63\n116\n535 535 column, the number of packages only depending on sf is less than when we ignore \nrecursive dependencies in the third column, which is packages using sf that also use \nsp. The number of packages only using sf is encouraging, given that it first entered \nCRAN in October 2016. It is also encouraging that a fair number of these packages use both sp and sf, \nshowing existing packages are preserving legacy workflows, but also opening up for \nmore modern object representations. It takes time and effort to communicate the \ndesirability of migrating from sp representations to sf and probably stars. 3.4  \u0007Reverse dependencies of the sp and sf packages Although \nkeeping the R code running is feasible, including compiled code not using external \nsoftware, migration to sf is advisable. 4  \u0007Upstream software dependencies of the R‑spatial ecosystem When changes occur in upstream external software, R packages using these libraries \noften need to adapt, but package maintainers try very hard to shield users from any \nnegative consequences, so that legacy workflows continue to provide the same or \nat least similar results from the same data. The code shown in Bivand et al. (2008, \n(2013) is almost all run nightly on a platform with updated R packages and external \nsoftware. This does not necessarily trap all differences (figures are not compared), \nbut is helpful in detecting impacts of changes in packages or external software. It is \nalso very helpful that CRAN servers using the released and development versions of \nR, and with different versions of external software, also run nightly checks. Again, \nsometimes changes are only noticed by users, but quite often checks run by main-\ntainers and by CRAN alert us to impending challenges. Tracking the development \nmailing lists of the external software communities, all open source, can also show \nhow thinking is evolving. However, sometimes code tidying in external software \ncan have unexpected consequences, breaking not sf or sp with rgdal or rgeos, but a \npackage further downstream. Bivand (2014) discusses open-source geospatial soft-\nware stacks more generally, but here we will consider ongoing changes in PROJ and \nlinked changes in GDAL. We will use the example of the location of the Broad Street pump in Soho, London, \nrelated to the 1854 Cholera epidemic and Dr John Snow’s intervention Brody et al. 1 3 536 R. S. Bivand (2000), distributed with sf (from version 0.8–1). Although it was known that changes in \nupstream software would impact workflows, the extent of those impacts became clear \nusing a standard example following upgrading to PROJ 6 and GDAL 3 in 2019: (2000), distributed with sf (from version 0.8–1). Although it was known that changes in \nupstream software would impact workflows, the extent of those impacts became clear \nusing a standard example following upgrading to PROJ 6 and GDAL 3 in 2019: > bp_file <- system.file(\"gpkg/b_pump.gpkg\", package=\"sf\")\n> b_pump_sf <- st_read(bp_file, quiet=TRUE) > bp_file <- system.file(\"gpkg/b_pump.gpkg\", package=\"sf\")\n> b_pump_sf <- st_read(bp_file, quiet=TRUE) Before R packages upgraded the way coordinate reference systems were represented \nin early 2020, our Proj4 string representation suffered degradation. Taking the Proj4 \nstring defined in PROJ 5 for the British National Grid, we can see a +datum=OSGB36 \nkey–value pair. 4  \u0007Upstream software dependencies of the R‑spatial ecosystem But when processing this input with PROJ 6 and GDAL 3, this key is \nremoved. Checking, we can see that reading the input string appears to work, but the \noutput for the Proj4 string drops the +datum=OSGB36 key–value pair, introducing \ninstead the ellipse implied by that datum: > proj5 <- paste0(\"+proj=tmerc +lat_0=49 +lon_0=-2 +k=0.9996012717\",\n+\n\" +x_0=400000 +y_0=-100000 +datum=OSGB36 +units=m +no_defs\")\n> legacy <- st_crs(proj5)\n> proj6 <- legacy$proj4string\n> proj5_parts <- unlist(strsplit(proj5, \" \"))\n> proj6_parts <- unlist(strsplit(proj6, \" \"))\n> proj5_parts[!is.element(proj5_parts, proj6_parts)]\n[1] \"+datum=OSGB36\"\n> proj6_parts[!is.element(proj6_parts, proj5_parts)]\n[1] \"+ellps=airy\" > proj5 <- paste0(\"+proj=tmerc +lat_0=49 +lon_0=-2 +k=0.9996012717\n+\n\" +x_0=400000 +y_0=-100000 +datum=OSGB36 +units=m +no_defs\")\n> legacy <- st_crs(proj5)\n> proj6 <- legacy$proj4string\n> proj5_parts <- unlist(strsplit(proj5, \" \"))\n> proj6_parts <- unlist(strsplit(proj6, \" \"))\n> proj5_parts[!is.element(proj5_parts, proj6_parts)]\n[1] \"+datum=OSGB36\"\n> proj6_parts[!is.element(proj6_parts, proj5_parts)]\n[1] \"+ellps=airy\" We can emulate the problem seen following the release in May 2019 of GDAL 3.0.0 \nusing PROJ 6, by inserting the degraded Proj4 string into the Broad Street pump object. The coordinate reference system representation is now ignorant of the proper datum \nspecification: > b_pump_sf1 <- b_pump_sf > b_pump_sf1 <- b_pump_sf _p\np_\n_p\np_\n> st_crs(b_pump_sf1) <- st_crs(st_crs(b_pump_sf1)$proj4string) p\np\np\np\n> st_crs(b_pump_sf1) <- st_crs(st_crs(b_pump_sf1)$proj4string) > st_crs(b_pump_sf1) <- st_crs(st_crs(b_pump_sf1)$proj4string) Why does this matter? For visualization on a web map, for example using the \nmapview package, the projected geometries are transformed to the same WGS84 \nellipse and datum (EPSG:4326) that were used in PROJ 4 as a transformation hub. In leaflet, these are projected to Web Mercator (EPSG:3857). In mapview(), the \nsf::st_transform() function is used, so we will emulate this step before hand-\ning on the geometries for display. 4  \u0007Upstream software dependencies of the R‑spatial ecosystem The projected British National Grid point location is \nunchanged: 1 3 1 3 Progress in the R ecosystem for rep\n> all.equal(st_coordinate\n+\nst_coordinate\n[1] TRUE\nHowever, because the \nresentation of its coordin\nformed identically to WG\nFigure 8:\n> b_pump_sf_ll <- st_tran\n> b_pump_sf1_ll <- st_tra\n> st_distance(b_pump_sf_l\nUnits: [m]\n[,1]\n[1,] 125.0578 Progress in the R ecosystem for representing and handling spatial… 537 > all.equal(st_coordinates(st_geometry(b_pump_sf)),\n+\nst_coordinates(st_geometry(b_pump_sf1)))\n[1] TRUE However, because the one of the objects now has a degraded Proj4 string rep-\nresentation of its coordinate reference system, the output points, apparently trans-\nformed identically to WGS84, are now some distance apart, as is also shown in \nFigure 8: > b_pump_sf_ll <- st_transform(b_pump_sf, 4326)\n> b_pump_sf1_ll <- st_transform(b_pump_sf1, 4326)\n> st_distance(b_pump_sf_ll, b_pump_sf1_ll)\nUnits: [m]\n[,1]\n[1,] 125.0578 > b_pump_sf_ll <- st_transform(b_pump_sf, 4326) [,1]\n[1,] 125.0578 Once PROJ 6 and GDAL 3 had stabilized in the summer of 2019, we iden-\ntified the underlying threat as lying in the advertised degradation of GDAL’s \nexportToProj4() function. When reading raster and vector files, the coordinate ref-\nerence system representation using Proj4 strings would often be degraded, so that fur-\nther transformation within R (also using GDAL/PROJ functionality) would be at risk of \nmuch greater inaccuracy than with PROJ 5 and GDAL 2. Since then, sf, sp with rgdal Fig. 8   Displays made using mapview displays of the Broad Street pump, with the green point within 2 m \nof the pump location, and the red point in Ingestre Place because of the loss of the datum specification Fig. 8   Displays made using mapview displays of the Broad Street pump, with the green point within 2 m \nof the pump location, and the red point in Ingestre Place because of the loss of the datum specification Fig. 8   Displays made using mapview displays of the Broad Street pump, with the green point within 2 m \nof the pump location, and the red point in Ingestre Place because of the loss of the datum specification 1 3 3 538 R. S. 7  https​://cran.r-proje​ct.org/web/packa​ges/rgdal​/vigne​ttes/CRS_proje​ction​s_trans​forma​tions​.html. 6  https​://www.r-spati​al.org/r/2020/03/17/wkt.html.\n7 8  State of https​://githu​b.com/rsbiv​and/sp of 18 August 2020 or later. 4  \u0007Upstream software dependencies of the R‑spatial ecosystem Bivand and raster have adopted the 2019 version of the “Well-Known Text” coordinate refer-\nence system representation WKT2-2019 (ISO 2019) instead of Proj4 strings to contain \ncoordinate reference system definitions.6 Accommodations have also been provided so \nthat the S3 class \"crs\" objects used in objects defined in sf, and the formal S4 class \n\"CRS\" objects used objects defined in sp and raster, can continue to attempt to support \nProj4 strings in addition, while other package maintainers and workflow users catch up.7 l\nFollowing an extended campaign of checking about 900 reverse dependencies \n(packages depending on sp, rgdal and others) and dozens of github issues, most of \nthe consequences of the switch to WKT2 among packages have now been addressed. Most recently (late August 2020), 115 packages have been offered rebuilt stored \nobjects that had included \"CRS\" objects without WKT2 definitions. i\nThis approach has ensured that spatial objects, whether created within R, read in from \nexternal data sources or read as stored objects, all have WKT2 string representations of \ntheir coordinate reference systems, and for backward compatibility can represent these in \naddition as Proj4 strings. Operations on objects should carry forward the new represen-\ntations, which should be written out to external data formats correctly. There is a minor \ndivergence between sf and sp (and thus rgdal): In sf, the axis order of the CRS is pre-\nserved as instantiated, but objects do not have their axes swapped to accord with authori-\nties unless sf::st_axis_order() is set TRUE. This can appear odd, because \nalthough the representation records a northings–eastings axis order, data are treated as \neastings–northings in plotting, variogram construction and so on: > st_crs(\"EPSG:4326\")\nCoordinate Reference System:\nUser input: EPSG:4326\nwkt:\nGEOGCRS[\"WGS 84\",\nDATUM[\"World Geodetic System 1984\",\nELLIPSOID[\"WGS 84\",6378137,298.257223563,\nLENGTHUNIT[\"metre\",1]]],\nPRIMEM[\"Greenwich\",0,\nANGLEUNIT[\"degree\",0.0174532925199433]],\nCS[ellipsoidal,2],\nAXIS[\"geodetic latitude (Lat)\",north,\nORDER[1],\nANGLEUNIT[\"degree\",0.0174532925199433]],\nAXIS[\"geodetic longitude (Lon)\",east,\nORDER[2],\nANGLEUNIT[\"degree\",0.0174532925199433]], ANGLEUNIT[\"degree\",0.0174532925199433]], ANGLEUNIT[\"degree\",0.0174532925199433]] g\nUSAGE[\nSCOPE[\"unknown\"],\nAREA[\"World\"],\nBBOX[-90,-180,90,180]],\nID[\"EPSG\",4326]] g\nUSAGE[\nSCOPE[\"unknown\"],\nAREA[\"World\"],\nBBOX[-90,-180,90,180]],\nID[\"EPSG\",4326]] USAGE[\nSCOPE[\"unknown\"],\nAREA[\"World\"],\nBBOX[-90,-180,90,180]],\nID[\"EPSG\",4326]] USAGE[\nSCOPE[\"unknown\"],\nAREA[\"World\"],\nBBOX[-90,-180,90,180]],\nID[\"EPSG\",4326]] ID[\"EPSG\",4326]] In sp/rgdal, attempts are made to ensure that axis order is in the form termed \nGIS, traditional or visualization that is always eastings–northings: 1 3 Progress in the R ecosystem for representing and handling spatial… > library(sp)\n> cat(wkt(CRS(\"EPSG:4326\")))\nGEOGCRS[\"WGS 84\",\nDATUM[\"World Geodetic System 1984\",\nELLIPSOID[\"WGS 84\",6378137,298.257223563,\nLENGTHUNIT[\"metre\",1]],\nID[\"EPSG\",6326]],\nPRIMEM[\"Greenwich\",0,\nANGLEUNIT[\"degree\",0.0174532925199433],\nID[\"EPSG\",8901]],\nCS[ellipsoidal,2],\nAXIS[\"longitude\",east,\nORDER[1],\nANGLEUNIT[\"degree\",0.0174532925199433,\nID[\"EPSG\",9122]]],\nAXIS[\"latitude\",north,\nORDER[2],\nANGLEUNIT[\"degree\",0.0174532925199433,\nID[\"EPSG\",9122]]],\nUSAGE[\nSCOPE[\"unknown\"],\nAREA[\"World\"],\nBBOX[-90,-180,90,180]]] > library(sp)\n> cat(wkt(CRS(\"EPSG:4326\")))\nGEOGCRS[\"WGS 84\",\nDATUM[\"World Geodetic System 1984\",\nELLIPSOID[\"WGS 84\",6378137,298.257223563,\nLENGTHUNIT[\"metre\",1]],\nID[\"EPSG\",6326]],\nPRIMEM[\"Greenwich\",0,\nANGLEUNIT[\"degree\",0.0174532925199433],\nID[\"EPSG\",8901]],\nCS[ellipsoidal,2],\nAXIS[\"longitude\",east,\nORDER[1],\nANGLEUNIT[\"degree\",0.0174532925199433,\nID[\"EPSG\",9122]]],\nAXIS[\"latitude\",north,\nORDER[2],\nANGLEUNIT[\"degree\",0.0174532925199433,\nID[\"EPSG\",9122]]],\nUSAGE[\nSCOPE[\"unknown\"],\nAREA[\"World\"],\nBBOX[-90,-180,90,180]]] Progress in the R ecosystem for representing and handling spatial… 539 9  nl is used to represent the verbatim new line character in the processed article. USAGE[ USAGE[\nSCOPE[\"unknown\"],\nAREA[\"World\"],\nBBOX[-90,-180,90,180]]] The probability of confusion increases when coercing from sf to sp and vice \nversa, with the representations most often remaining unchanged.8 > sf_from_sp <- st_crs(CRS(\"EPSG:4326\"))\n> o <- strsplit(sf_from_sp$wkt, nl)[[1]]\n> cat(paste(o[grep(\"CS|AXIS|ORDER\", o)], collapse=nl))\nCS[ellipsoidal,2],\nAXIS[\"longitude\",east,\nORDER[1],\nAXIS[\"latitude\",north,\nORDER[2], > sf_from_sp <- st_crs(CRS(\"EPSG:4326\"))\n> o <- strsplit(sf_from_sp$wkt, nl)[[1]]\n> cat(paste(o[grep(\"CS|AXIS|ORDER\", o)], collapse=nl))\nCS[ellipsoidal,2],\nAXIS[\"longitude\",east,\nORDER[1],\nAXIS[\"latitude\",north,\nORDER[2], > sf_from_sp <- st_crs(CRS(\"EPSG:4326\")) o <- strsplit(sf_from_sp$wkt, nl)[[1]] p\np\n> cat(paste(o[grep(\"CS|AXIS|ORDER\", o)], collapse=nl)) (p\n( [g\np(\n|\n|\nCS[ellipsoidal,2],\nAXIS[\"longitude\",east,\nORDER[1],\nAXIS[\"latitude\",north,\nORDER[2], > sp_from_sf <- as(st_crs(\"EPSG:4326\"), \"CRS\")\n> o <- strsplit(wkt(sp_from_sf), nl)[[1]]\n> cat(paste(o[grep(\"CS|AXIS|ORDER\", o)], collapse=nl))\nCS[ellipsoidal,2],\nAXIS[\"geodetic latitude (Lat)\",north,\nORDER[1],\nAXIS[\"geodetic longitude (Lon)\",east,\nORDER[2], > sp_from_sf <- as(st_crs(\"EPSG:4326\"), \"CRS\") > sp_from_sf <- as(st_crs(\"EPSG:4326\"), \"CRS\")\n> o <- strsplit(wkt(sp_from_sf), nl)[[1]]\n> cat(paste(o[grep(\"CS|AXIS|ORDER\", o)], collapse=nl)\nCS[ellipsoidal,2],\nAXIS[\"geodetic latitude (Lat)\",north,\nORDER[1],\nAXIS[\"geodetic longitude (Lon)\",east,\nORDER[2], > o <- strsplit(wkt(sp_from_sf), nl)[[1]] p\np\n> cat(paste(o[grep(\"CS|AXIS|ORDER\", o)], collapse=nl)) Both of these coercions are using the same underlying PROJ and GDAL versions, \nand the same PROJ metadata. Once work in progress is completed, coercions should \nrespect the setting of sf::st_axis_order(). 1 3 540 R. S. Bivand It may useful for users to know of other differences between sf and sp/rgdal. Transformation in sf uses code in GDAL, which in turn uses functions in PROJ; in \nsp/rgdal, PROJ is used directly for transformation. In order to demonstrate more of \nwhat is happening, let us coerce these sf objects to sp (they are both planar with an \nx–y axis order): > b_pump_sp <- as(b_pump_sf, \"Spatial\")\n> b_pump_sp1 <- as(b_pump_sf1, \"Spatial\") > b_pump_sp <- as(b_pump_sf, \"Spatial\")\n> b_pump_sp1 <- as(b_pump_sf1, \"Spatial\") We will also set up a temporary directory for use with the on-demand grid down-\nload functionality in PROJ 7; this must be done before rgdal is loaded: > td <- tempfile()\n> dir.create(td)\n> Sys.setenv(\"PROJ_USER_WRITABLE_DIRECTORY\"=td)\n> library(rgdal) > td <- tempfile()\n> dir.create(td)\n> Sys.setenv(\"PROJ_USER_WRITABLE_DIRECTORY\"=td)\n> library(rgdal) > td <- tempfile()\n> dir.create(td)\n> Sys.setenv(\"PROJ_USER_WRITABLE_DIRECTORY\"=td)\n> library(rgdal) > td <- tempfile() > td <- tempfile()\n> dir.create(td)\n( > dir.create(td) > library(rgdal) In sf, areas of interest need to be given by the users, while in transformation and \nprojection in rgdal, these are calculated from the object being projected or trans-\nformed. The provision of areas of interest is intended to reduce the number of candi-\ndate coordinate operations found by PROJ. USAGE[ > WKT <- wkt(b_pump_sp)\n> o <- list_coordOps(WKT, \"EPSG:4326\")\n> c(nrow(o), nrow(o[o$instantiable,]), sum(o$number_grids))\n[1] 8 7 1\n> aoi0 <- project(t(unclass(bbox(b_pump_sp))), WKT, inv=TRUE)\n> aoi <- c(t(aoi0 + c(-0.1, +0.1)))\n> o_aoi <- list_coordOps(WKT, \"EPSG:4326\", area_of_interest=aoi)\n> c(nrow(o_aoi), nrow(o_aoi[o_aoi$instantiable,]), sum(o_aoi$number_grids))\n[1] 5 4 1 rgdal::list_coordOps() accesses the PROJ metadata database to \nsearch through candidate coordinate operations, ranking them by accuracy, \nreturning a data frame of operations. When an area of interest is provided, candi-\ndates falling outside it are dropped. Coordinate operations that cannot be instanti-\nated because of missing grids are also listed. We can see here that when an area \nof interest is not given, 8 candidate operations are found when the WKT string \ncontains datum information. Of these, 7 may be instantiated, with 1 needing a \ngrid (here, the operation that cannot be instantiated). Three operations cease to be \ncandidates if we use an area of interest. In sp/rgdal, the coordinate operation last used is returned and can be retrieved \nusing rgdal::get_last_coordOp(); coordinate operations are repre-\nsented as pipelines (Knudsen and Evers 2017; Evers and Knudsen 2017), intro-\nduced in PROJ 5 and using the PROJ key–value pair notation:9 1 3 1 3 Progress in the R ecosystem for representing and handling spatial… 541 > b_pump_sp_ll <- spTransform(b_pump_sp, \"EPSG:4326\")\n> cat(strwrap(get_last_coordOp()), sep=nl)\n+proj=pipeline +step +inv +proj=tmerc +lat_0=49 +lon_0=-2 +k=0.999601\n+x_0=400000 +y_0=-100000 +ellps=airy +step +proj=push +v_3 +step +proj=cart\n+ellps=airy +step +proj=helmert +x=446.448 +y=-125.157 +z=542.06 +rx=0.15\n+ry=0.247 +rz=0.842 +s=-20.489 +convention=position_vector +step +inv\n+proj=cart +ellps=WGS84 +step +proj=pop +v_3 +step +proj=unitconvert\n+xy_in=rad +xy_out=deg > b_pump_sp_ll <- spTransform(b_pump_sp, \"EPSG:4326\")\n> cat(strwrap(get_last_coordOp()), sep=nl)\n+proj=pipeline +step +inv +proj=tmerc +lat_0=49 +lon_0=-2 +k=0.999601\n+x_0=400000 +y_0=-100000 +ellps=airy +step +proj=push +v_3 +step +proj=cart\n+ellps=airy +step +proj=helmert +x=446.448 +y=-125.157 +z=542.06 +rx=0.15\n+ry=0.247 +rz=0.842 +s=-20.489 +convention=position_vector +step +inv\n+proj=cart +ellps=WGS84 +step +proj=pop +v_3 +step +proj=unitconvert\n+xy_in=rad +xy_out=deg > b_pump_sp_ll <- spTransform(b_pump_sp, \"EPSG:4326\") > b_pump_sp_ll <- spTransform(b_pump_sp, \"EPSG:4326\") > b_pump_sp_ll <- spTransform(b_pump_sp, \"EP\n> cat(strwrap(get_last_coordOp()), sep=nl) p\np\np\np\np\np\np\n> cat(strwrap(get_last_coordOp()), sep=nl) +proj=pipeline +step +inv +proj=tmerc +lat_0=49 +lon_0=-2 +k=0.999601\n+x_0=400000 +y_0=-100000 +ellps=airy +step +proj=push +v_3 +step +proj=cart\n+ellps=airy +step +proj=helmert +x=446.448 +y=-125.157 +z=542.06 +rx=0.15\n+ry=0.247 +rz=0.842 +s=-20.489 +convention=position_vector +step +inv\n+proj=cart +ellps=WGS84 +step +proj=pop +v_3 +step +proj=unitconvert\n+xy_in=rad +xy_out=deg Here, we can see that an inverse projection from the specified Transverse Mer-\ncator projection is made to geographical coordinates, followed by a seven-param-\neter Helmert transformation to WGS84 ellipsoid and datum. 10  https​://cdn.proj.org. > enable_proj_CDN()\n[1] \"Using: /tmp/Rtmp5keMtw/file1c4592b3b37c2\"\n> list.files(td)\ncharacter(0) > o <- list_coordOps(WKT, \"EPSG:4326\", area_of_interest=aoi)\n> c(nrow(o), nrow(o[o$instantiable,]), sum(o$number_grids))\n[1] 5 5 1 > o <- list_coordOps(WKT, \"EPSG:4326\", area_of_interest=aoi)\n> c(nrow(o), nrow(o[o$instantiable,]), sum(o$number_grids))\n[1] 5 5 1 > o <- list_coordOps(WKT, \"EPSG:4326\", area_of_interest=aoi)\n> c(nrow(o), nrow(o[o$instantiable,]), sum(o$number_grids))\n[1] 5 5 1 > b_pump_sp_llg <- spTransform(b_pump_sp, \"EPSG:4326\") > cat(strwrap(get_last_coordOp()), sep=nl) +proj=pipeline +step +inv +proj=tmerc +lat_0=49 +lon_0=-2 +k=0.999601 +x_0=400000 +y_0=-100000 +ellps=airy +step +proj=hgridshift +grids=uk_os_OSTN15_NTv2_OSGBtoETRS.tif +step +proj=unitconvert +xy_in=rad\n+xy_out=deg Now the downloaded grid is cached in the database in the designated CDN direc-\ntory and may be used for other transformations using the same operation. > list.files(td)\n[1] \"cache.db\"\n> file.size(file.path(td, list.files(td)[1]))\n[1] 319488\n> disable_proj_CDN() > list.files(td)\n[1] \"cache.db\"\n> file.size(file.path(td, list.files(td)[1]))\n[1] 319488\n> disable_proj_CDN() Once again, the distance between the point transformed from the sf object as read \nfrom file and the point with the degraded coordinate reference system emulating the \neffect of the change in behaviour of GDAL’s exportToProj4() in GDAL 6 and \nlater is about 125 m. Using the CDN shifts the output point by 1.7 m. For confirma-\ntion, the output transformed coordinates for the sp and sf objects using the Helmert \ntransformation are the same. > c(spDists(b_pump_sp1_ll, b_pump_sp_ll),\n+\nspDists(b_pump_sp_llg, b_pump_sp_ll))*1000\n[1] 125.057683\n1.751474\n> all.equal(unname(coordinates(b_pump_sp_ll)),\n+\nunname(st_coordinates(st_geometry(b_pump_sf_ll))))\n[1] TRUE > c(spDists(b_pump_sp1_ll, b_pump_sp_ll),\n+\nspDists(b_pump_sp_llg, b_pump_sp_ll))*1000\n[1] 125.057683\n1.751474\n> all.equal(unname(coordinates(b_pump_sp_ll)),\n+\nunname(st_coordinates(st_geometry(b_pump_sf_ll))))\n[1] TRUE > c(spDists(b_pump_sp1_ll, b_pump_sp_ll),\n+\nspDists(b_pump_sp_llg, b_pump_sp_ll))*1000\n[1] 125.057683\n1.751474\n> all.equal(unname(coordinates(b_pump_sp_ll)),\n+\nunname(st_coordinates(st_geometry(b_pump_sf_ll))))\n[1] TRUE Although it appears that most of the consequences of the change in representa-\ntion of coordinate reference systems from Proj4 to WKT2 strings have now been \naddressed, we still see signs on the mailing list and on Twitter that users, natu-\nrally directing their attention to their analytical or visualization work, may still \nbe confused. The extent of the spatial cluster of R packages is so great that it will \nundoubtedly take time before the dust settles. However, we trust that the operation \nof upgrading representations is now largely complete. Multiple warnings issued in \nsp workflows, now noisily drawing attention to possible degradations in workflows, \nwill by default be muted when sp 1.5 and rgdal 1.6 are released. USAGE[ The parameters are \ncontained in the best instantiable coordinate operation retrieved from the PROJ \ndatabase. > o <- list_coordOps(wkt(b_pump_sp1), \"EPSG:4326\", area_of_interest=aoi)\n> cat(nrow(o), o$ballpark, nl)\n1 TRUE > b_pump_sp1_ll <- spTransform(b_pump_sp1, \"EPSG:4326\") > cat(strwrap(get_last_coordOp()), sep=nl) Going on to the case of the degraded representation, only 1 operation is found, \nwith only ballpark accuracy. With our emulation of the dropping of +datum= sup-\nport in GDAL’s exportToProj4(), we see that the coordinate operation pipe-\nline only contains the inverse projection step, accounting for the observed shift of \nthe Broad Street pump to Ingestre Place. Finally, sp/rgdal may use the provision of on-demand downloading of transfor-\nmation grids to provide more accuracy (CDN, from PROJ 7).10 Before finding and \nchoosing to use a coordinate operation using an on-demand downloaded grid, the \ndesignated directory is empty: > enable_proj_CDN()\n[1] \"Using: /tmp/Rtmp5keMtw/file1c4592b3b37c2\"\n> list.files(td)\ncharacter(0) Using the CDN, all the candidate operations are instantiable, and the pipeline \nnow shows a horizontal grid transformation rather than a Helmert transformation. 1 3 R. S. Bivand 542 5  \u0007Outlook As has already been mentioned, either most sp-based workflows continue to func-\ntion using sf objects or changes have been made in packages like spdep and its mod-\nelling counterpart spatialreg to permit scripts and packages using these packages \nto continue to function. Some workflows require more attention than others, but 1 3 Progress in the R ecosystem for representing and handling spatial… 543 the transition to sf from sp, rgdal and rgeos should be unproblematic. Shifting to \nnew visualization packages like tmap, cartography and mapview should also be \nrelatively easy, and use of sf and the new visualization packages should certainly \nbecome standard for new research and teaching. It may take a little longer for stars \nto find its place, but work here is continuing, and will stabilize before long. Similar \nremarks apply to transition from raster to terra. The key challenges for handling spatial data using R concern the upstream \nsoftware libraries in the open-source geospatial software stack, not just PROJ as \ndescribed above or changes in validity requirements in GEOS and other libraries, \nbut also the opportunities opened up by access to cloud-based earth observation data \nstreams. Because incoming data may take forms as yet not provided for, even where \nGDAL drivers become available, changes in object representations may become \nnecessary. In particular, this relates to spatio-temporal data, where trajectory data \nare especially demanding. These data representation challenges are actually oppor-\ntunities for users and developers to continue cooperating to contribute to making the \nR ecosystem for handling and analysing spatial and spatio-temporal data even more \ncapable and performant. Finally, please note that a CODECHECK certificate for this \npaper is available at https​://doi.org/10.5281/zenod​o.40038​48. CODECHECK is an \nopen-science initiative to facilitate sharing of computer programs and results pre-\nsented in scientific publications (see https​://codec​heck.org.uk/) Funding  Open Access funding provided by Norwegian School of Economics. Funding  Open Access funding provided by Norwegian School of Economics. Open Access  This article is licensed under a Creative Commons Attribution 4.0 International License, \nwhich permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as \nyou give appropriate credit to the original author(s) and the source, provide a link to the Creative Com-\nmons licence, and indicate if changes were made. The images or other third party material in this article \nare included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the \nmaterial. 5  \u0007Outlook If material is not included in the article’s Creative Commons licence and your intended use is \nnot permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission \ndirectly from the copyright holder. To view a copy of this licence, visit http://creat​iveco​mmons​.org/licen​\nses/by/4.0/. nding  Open Access funding provided by Norwegian School of Economics. References https\nCRAN.R-proje​ct.org/packa​ge=rgeos​, R package version 0.5-3 Bivand R, Pebesma E, Gomez-Rubio V (2008) Applied spatial data analysis with R. Springer, New Y Bivand R, Pebesma E, Gomez-Rubio V (2008) Applied spatial data analysis with R. Springer, New Yor\nBivand R, Pebesma E, Gomez-Rubio V (2013) Applied spatial data analysis with R, 2nd edn. Springe\nk Bivand R, Pebesma E, Gomez-Rubio V (2008) Applied spatial data analysis with R. Springer, New York\nBivand R Pebesma E Gomez Rubio V (2013) Applied spatial data analysis with R 2nd edn Springer Bivand R, Pebesma E, Gomez-Rubio V (2013) Applied spatial data analysis with R, 2nd edn. Springe\nNew York Bivand R, Keitt T, Rowlingson B (2020) rgdal: bindings for the ’geospatial’ data abstraction library. http\n://CRAN.R-proje​ct.org/packa​ge=rgdal​, R package version 1.5-16 Brody H, Rip MR, Vinten-Johansen P, Paneth N, Rachman S (2000) Map-making and myth-making in \nBroad Street: the London cholera epidemic, 1854. Lancet 356:64–68 de Vries A (2014) Finding clusters of CRAN packages using igraph. https​://blog.revol​ution​analy​tics. com/2014/12/findi​ng-clust​ers-of-cran-packa​ges-using​-igrap​h.html, Revolutions blog. Accessed 25 \nAug 2020 Dunnington D, Pebesma E, Rubak E (2020) s2: spherical geometry operators using the S2 geomet\nlibrary. https​://CRAN.R-proje​ct.org/packa​ge=s2, r package version 1.0.2 Eddelbuettel D (2013) Seamless R and C++ integration with Rcpp. Springer, New York Eddelbuettel D, Balamuta JJ (2018) Extending R with C++: a brief introduction to Rcpp. Am Stat \n72(1):28–36. https​://doi.org/10.1080/00031​305.2017.13759​90 Eddelbuettel D, Balamuta JJ (2018) Extending R with C++: a brie\n72(1):28–36. https​://doi.org/10.1080/00031​305.2017.13759​90 6. https​://doi.org/10.1080/00031​305.2017.13759​90 ( )\np\ng\nEddelbuettel D, François R, Allaire J, Ushey K, Kou Q, Russel N, Chambers J, Bates D (2011) Rcp\nseamless R and C++ integration. J Stat Softw 40(8):1–18 Eddelbuettel D, François R, Allaire J, Ushey K, Kou Q, Russel N, Chambers J, Bates D (2011) Rcpp: \nseamless R and C++ integration. J Stat Softw 40(8):1–18\nEdd lb\nl D F\ni R All i\nJ U h\nK K\nQ R\nll N B\nD Ch\nb\nJ (2020) R Eddelbuettel D, Francois R, Allaire J, Ushey K, Kou Q, Russell N, Bates D, Chambers J (2020) Rcpp: \nseamless R and C++ integration. https​://CRAN.R-proje​ct.org/packa​ge=Rcpp, R package version \n1.0.5 Evangelista PF, Beskow D (2018) Geospatial point density. R J 10(2):347–356. https​://doi. org/10.32614​/RJ-2018-061i Evers K, Knudsen T (2017) Transformation pipelines for PROJ.4. https​://www.fig.net/resou​rces/proc\neding​s/fig_proce​eding​s/fig20​17/paper​s/iss6b​/ISS6B​_evers​_knuds​en_9156.pdf, \nfIG \nworkin Evers K, Knudsen T (2017) Transformation pipelines for PROJ.4. https​://www.fig.net/resou​rces/proce​\neding​s/fig_proce​eding​s/fig20​17/paper​s/iss6b​/ISS6B​_evers​_knuds​en_9156.pdf, \nfIG \nworking \nweek 2017 proceedings. References Akima H, Gebhardt A (2020) akima: interpolation of irregularly and regularly spaced data. https\nCRAN.R-proje​ct.org/packa​ge=akima​, R package version 0.6-2.1 Appelhans T, Detsch F, Reudenbach C, Woellauer S (2020) mapview: interactive viewing of spatial da\nin R. https​://CRAN.R-proje​ct.org/packa​ge=mapvi​ew, R package version 2.9.0 Baddeley A, Turner R (2005) spatstat: an R package for analyzing spatial point patterns. J Stat Softw \n12(6):1–42 Baddeley A, Rubak E, Turner R (2015) Spatial point patterns: methodology and applications with R. Chapman and Hall, Londoni Baddeley A, Turner R, Rubak E (2020) spatstat: spatial point pattern analysis, model-fitting, simulatio\ntests. https​://CRAN.R-proje​ct.org/packa​ge=spats​tat, R package version 1.64-1 Bivand R (1998) Software and software design issues in the exploration of local dependence. The Statis-\ntician 47:499–508 1 3 R. S. Bivand 544 Bivand R (2000) Using the R statistical data analysis language on GRASS 5.0 GIS database files. Com\nput Geosci 26(9):1043–1052 p\nBivand R (2002) Spatial econometrics functions in R: classes and methods. J Geogr Syst 4:405–421 Bivand R (2006) Implementing spatial data analysis software tools in R. Geogr Anal 38:23–40 Bivand R (2014) Geocomputation and open source software: components and software stacks. In: Abra mputation and open source software: components an Bivand R (2014) Geocomputation and open source software: components and software stacks\nhart RJ, See LM (eds) Geocomputation. CRC Press, Boca Raton, pp 329–355 (chap 14) hart RJ, See LM (eds) Geocomputation. CRC Press, Boca Raton, pp 329–355 (chap 14)\nBivand R (2020a) classInt: choose univariate class intervals. https​://CRAN.R-proje​ct.org/packa​ge=class​\nInt, R package version 0.4-3 Bivand R (2020b) spdep: spatial dependence: weighting schemes, statistics. https​://CRAN.R-proje​ct.or Bivand R (2020b) spdep: spatial dependence: weighting schemes, stati Bivand R (2020b) spdep: spatial dependence: weighting schemes, statistics. https​://CRAN.R-proje​ct.org/\npacka​ge=spdep​, R package version 1.1-5 Bivand R (2020b) spdep: spatial dependence: weighting schemes, statistics. https​://CRAN.R proje​ct.org/\npacka​ge=spdep​, R package version 1.1-5 packa​ge=spdep​, R package version 1.1-5 packa​ge=spdep​, R package version 1.1-5 Bivand R, Gebhardt A (2000) Implementing functions for spatial statistical analysis using the R language. J Geogr Syst 2(3):307–317 Bivand R, Lewin-Koh N (2020) maptools: tools for handling spatial objects. https​://CRAN.R-proje​ct.or Bivand R, Lewin-Koh N (2020) maptools: tools for handling spatial objects. https​://CRAN.R-proje​ct.or\npacka​ge=mapto​ols, R package version 1.0-1 packa​ge=mapto​ols, R package version 1.0-1 Bivand R, Piras G (2019) spatialreg: spatial regression analysis. https​://CRAN.R-proje​ct.org/pack\nge=spati​alreg​, R package version 1.1-5 ge=spati​alreg​, R package version 1.1-5 Bivand R, Rundel C (2020) rgeos: interface to geometry engine—open source (’GEOS’). References Accessed 25 Aug 2020l Giraud T, Lambert N (2016) cartography: create and integrate maps in your R workflow. J Open \nSource Softw. https​://doi.org/10.21105​/joss.00054​ p\ng\nj\nGiraud T, Lambert N (2017) Reproducible cartography. In: Peterson M (ed) Advances in cartography \nand GIScience. ICACI 2017. Lecture notes in geoinformation and cartography. Springer, Cham, \npp 173–183. https​://doi.org/10.1007/978-3-319-57336​-6_13 pp\np\ng\nGiraud T, Lambert N (2020) cartography: thematic cartography. https​://CRAN.R-proje​ct.org/packa​\nge=carto​graph​y, R package version 2.4.1 Gómez-Rubio V (2011) RArcInfo: functions to import data from Arc/Info V7.x binary coverages. https​://CRAN.R-proje​ct.org/packa​ge=RArcI​nfo, R package version 0.4-12 Gómez-Rubio V, López-Quílez A (2005) RArcInfo: using GIS data with R. Comput Geosci \n31(8):1000–1006 Gómez-Rubio V, Ferrándiz-Ferragud J, Lopez-Quólez A (2005) Detecting clusters of disease with R. J Geogr Syst 7(2):189–206 1 3 Progress in the R ecosystem for representing and handling spatial… 545 Gómez-Rubio V, Ferrándiz-Ferragud J, López-Quílez A (2015) DCluster: functions for the detection \nof spatial clusters of diseases. https​://CRAN.R-proje​ct.org/packa​ge=DClus​ter, R package version \n0.2-7 Gómez-Rubio V, Ferrándiz-Ferragud J, López-Quílez A (2015) DCluster: functions for the detection \nof spatial clusters of diseases. https​://CRAN.R-proje​ct.org/packa​ge=DClus​ter, R package version \n0.2-7 Herring JR (2011) Opengis implementation standard for geographic information-simple feature \naccess-part 1: common architecture. Open Geospatial Consortium Inc, Wayland, p 111 Herring JR (2011) Opengis implementation standard for geographic information-simple feature \naccess-part 1: common architecture. Open Geospatial Consortium Inc, Wayland, p 111 Hijmans RJ (2020a) terra: geographic data analysis and modeling. https​://CRAN.R-proje​ct.org/pack\nge=raste​r, R package version 3.3-13 Hijmans RJ (2020a) terra: geographic data analysis and modeling. https​://CRAN.R-proje​ct.org/packa​\nge=raste​r, R package version 3.3-13 Hijmans RJ (2020b) terra: spatial data analysis. https​://CRAN.R-proje​ct.org/packa​ge=terra​, R pac\nage version 0.8-6 Hornik K (2020) stringsAsFactors. https​://devel​oper.r-proje​ct.org/Blog/publi​c/2020/02/16/strin​gsasf​\nactor​s/index​.html. Accessed 25 Aug 2020 g\nISO (2004) ISO 19125-1:2004 geographic information—simple feature access—part 1: common \narchitecture. https​://www.iso.org/stand​ard/40114​.html, iSO 19125-1:2004. Accessed 25 Aug \n2020 ISO (2019) ISO 19111:2019 Geographic information—referencing by coordinates. https​://www.is\norg/stand​ard/74039​.html, ISO 19111:2019. Accessed 25 Aug 2020 Kalibera T (2019) Use of C++ in packages. https​://devel​oper.r-proje​ct.org/Blog/publi​c/2019/03/28/\nuse-of-c-in-packa​ges/index​.html, R blog. Accessed 25 Aug 2020 Kaluzny S, Vega S, Cardoso T, Shelly A (1998) S+SpatialStats. Springer, New York S, Cardoso T, Shelly A (1998) S+SpatialStats. Sprin Knudsen T, Evers K (2017) Transformation pipelines for PROJ.4. Geophysical research abstracts, vol \n19, \nEGU2017-8050. https​://meeti​ngorg​anize​r.coper​nicus​.org/EGU20​17/EGU20​17-8050.pdf. Accessed 25 Aug 2020 Kralidis AT (2008) Geospatial open source and open standards convergences. In: Hall GB, Leahy M \n(eds) Open source approaches in spatial data handling. References Springer, Berlin, pp 1–20 Lapa T, Ximenes R, Silva NN, Souza W, Albuquerque MdFM, Campozana G (2001) Vigilância da \nhanseníase em Olinda, Brasil, utilizando técnicas de análise espacial. Cadernos de Saúde Pública \n17:1153–1162. https​://doi.org/10.1590/S0102​-311X2​00100​05000​16 Lovelace R, Ellison R (2018) stplanr: a package for transport planning. R J 10(2):7–23. https​://do\norg/10.32614​/RJ-2018-053 g\nLovelace R, Nowosad J, Muenchow J (2019) Geocomputation with R. CRC, Boca Raton Lovelace R, Ellison R, Morgan M (2020) stplanr: sustainable transport planning. https​://CRAN.R-\nproje​ct.org/packa​ge=stpla​nr, r package version 0.6.2 p\nj\ng p\ng\np\np\ng\nMajure JJ, Gebhardt A (2016) sgeostat: an object-oriented framework for geostatistical modeling in \nS+. https​://CRAN.R-proje​ct.org/packa​ge=sgeos​tat, R package version 1.0-27 Meyer D, Dimitriadou E, Hornik K, Weingessel A, Leisch F (2019) e1071: misc functions of the Depart-\nment of Statistics, Probability Theory Group (Formerly: E1071), TU Wien. https​://CRAN.R-proje​\nct.org/packa​ge=e1071​, r package version 1.7-3 g p\ng\np\ng\nNeuwirth E (2014) RColorBrewer: ColorBrewer palettes. https​://CRAN.R-proje​ct.org/packa​ge=RColo​\nrBrew​er, R package version 1.1-2 Nowosad J (2019) ’CARTOColors’ palettes. https​://nowos​ad.githu​b.io/rcart​ocolo​r, R package version \n2.0.0 Pebesma E (2012) spacetime: spatio-temporal data in R. J Stat Softw 51(7):1–30 Pebesma E (2018) Simple features for R: standardized support for spatial vector data. R J 10(1):439–44 Pebesma E (2018) Simple features for R: standardized support for spatial vector data. R J 10(1):439 446\nPebesma E (2020a) sf: simple features for R. https​://CRAN.R-proje​ct.org/packa​ge=sf, R package version \n0.9-5 Pebesma E (2020b) spacetime: classes and methods for spatio-temporal data. https​://CRAN.R-proje​\nct.org/packa​ge=space​time, R package version 1.2-3 Pebesma E (2020c) stars: spatiotemporal arrays, raster and vector data cubes. https​://CRAN.R-pro\nct.org/packa​ge=stars​, R package version 0.4-3 g\ng\ng\nPebesma E, Bivand R (2005) Classes and methods for spatial data in R. R News 5(2):9–13 Pebesma E, Bivand R (2020) sp: classes and methods for spatial data. https​://CRAN.R-proje​ct.org/packa​\nge=sp, R package version 1.4-2l Pebesma E, Dunnington D (2020) In r-spatial, the Earth is no longer flat. https​://www.r-spati​\nal.org/r/2020/06/17/s2.html. Accessed 25 Aug 2020 Pebesma EJ, Wesseling CG (1998) Gstat, a program for geostatistical modelling, prediction and simul\ntion. Comput Geosci 24:17–31 p\nPebesma E, Bivand R, Ribeiro P (2015) Software for spatial statistics. J Stat Softw 63(1):1–8. https​://doi. org/10.18637​/jss.v063.i01 besma E, Bivand R, Ribeiro P (2015) Software for spatial statistics. J Stat Softw 63(1):1–8. https​://doi. org/10.18637​/jss.v063.i01 1 3 R. S. Bivand 546 Pebesma E, Mailund T, Hiebert J (2016) Measurement Units in R. R J 8(2):486–494. Publisher’s Note  Springer Nature remains neutral with regard to jurisdictional claims in publishe\nmaps and institutional affiliations. References https​://do\norg/10.32614​/RJ-2016-061 Pebesma E, Mailund T, Kalinowski T (2020) units: measurement units for R vectors. https​://CRAN.R-\nproje​ct.org/packa​ge=units​, r package version 0.6-7 p\nj\ng p\ng\np\ng\nRenka RJ, Gebhardt A (2020) tripack: triangulation of irregularly spaced data. https​://CRAN.R-proje​\n/\nk\ni\nk\nk\ni\n1 3 9 1 Renka RJ, Gebhardt A (2020) tripack: triangulation of ir\nct.org/packa​ge=tripa​ck, R package version 1.3-9.1 enka RJ, Gebhardt A (2020) tripack: triangulation of irregularly spaced data. https​://CRAN.R-proje​\nct.org/packa​ge=tripa​ck, R package version 1.3-9.1 (\n)\np\ng\nct.org/packa​ge=tripa​ck, R package version 1.3-9.1 ct.org/packa​ge=tripa​ck, R package version 1.3-9.1 Rowlingson B, Diggle PJ (1993) Splancs: spatial point pattern analysis code in S-Plus. Comput Geosci \n19:627–655 Rowlingson B, Diggle P (2017) splancs: spatial and space-time point pattern analysis. https​://CRAN.R-\nproje​ct.org/packa​ge=splan​cs, R package version 2.01-40 Sawicka K, Heuvelink GB, Walvoort DJ (2018) Spatial uncertainty propagation analysis with the spup R \npackage. R J 10(2):180–199. https​://doi.org/10.32614​/RJ-2018-047 Sawicka K, Heuvelink GB, Walvoort DJ (2018) Spatial uncertainty propagation analysis with the spup R \npackage. R J 10(2):180–199. https​://doi.org/10.32614​/RJ-2018-047\nTennekes M (2018) tmap: thematic maps in R. J Stat Softw 84(6):1–39 Tennekes M (2018) tmap: thematic maps in R. J Stat Softw 84(6):1–39 Tennekes M (2020) tmap: thematic maps. https​://CRAN.R-proje​ct.org/packa​ge=tmap, R package versi\n3.1 Venables WN, Ripley BD (2002) Modern applied statistics with S, 4th edn. Springer, New York Warmerdam F (2008) The geospatial data abstraction library. In: Hall GB, Leahy M (eds) Open \napproaches in spatial data handling. Springer, Berlin, pp 87–104 Wickham H (2014) Tidy data. J Stat Softw 59(10):1–23. https​://doi.org/10.18637​/jss.v059.i10 Wickham H, Chang W, Henry L, Pedersen TL, Takahashi K, Wilke C, Woo K, Yutani H (2020) ggplot2: \ncreate elegant data visualisations using the grammar of graphics. https​://CRAN.R-proje​ct.org/packa​\nge=ggplo​t2, R package version 3.3.2 Publisher’s Note  Springer Nature remains neutral with regard to jurisdictional claims in published \nmaps and institutional affiliations. Publisher’s Note  Springer Nature remains neutral with regard to jurisdictional claims in published \nmaps and institutional affiliations. 1 3 1 3 1 3"