英文原地址：http://cran.r-project.org/web/views/Spatial.html

［本人亲自翻译，转载请注明：来自于OpenThings］

该英文的维护者Roger Bivand与人合著的《Applied Spatial Data Analysis with R》已翻译为中文《空间数据分析与R语言实践》，由清华大学出版社出版，感兴趣的可以买了来看，只需一顿饭钱，省了自己半年的摸索。［说句实在话，翻译的有点晦涩，不过国内搞这个的本来很少，不能苛求，无论如何咱们看中文版还是省点劲的。］

CRAN 任务概览: 空间数据的分析

Base R 包含很多函数用于空间数据的读取、可视化和焦点. 本概览的重点是"geographical"空间数据, 这些包括地理位置和相关的属性等。.Base R函数由贡献包组成,  位于CRAN, 其他的还在发展之中。一个活跃的地方是 R-Forge, 列出"Spatial Data and Statistics" 工程在 project tree。关于R-spatial包的信息, 尤其是sp 将被发送到R-Forge rspatial项目的 website , 包括一个可视化的gallery。

贡献包的目的领域主要有两个：将数据移入R或者移出和在R中分析空间数据或者可视化。

The                  R-SIG-Geo                mailing-list is a good place to begin for obtaining help and discussing    questions about both accessing data, and analysing it. The mailing list    is a good place to search for information about relevant courses, and a    list is hosted at the                  GeoDaCenter        .

The packages in this    view can be roughly structured into the following topics. If you think    that some package is missing from the list, please let me know.

•              Classes for spatial data            : Because many of the      packages importing and using spatial data have had to include      objects of storing data and functions for visualising it,      an initiative is in progress to construct shared classes      and plotting functions for spatial data.      Thesp            package is discussed in a note in                          R      News            .  Many other packages have become dependent on these      classes, includingrgdal            andmaptools.      Thergeos            package provides an interface to topology functions      forsp            objects using                          GEOS            .rgeos            is now available for Mac OSX on CRAN for pre-Mavericks R; for Mavericks+,       please see advice given on R-sig-mac:                          for a binary package            , and                          for source install using the Kyngchaos framework            . Also see the      README file in the package, for example at                          R-Forge            .      Theraster            package is a major extension of spatial data      classes to virtualise access to large rasters, permitting large      objects to be analysed, and extending the analytical tools available      for both raster and vector data. Used withrasterVis, it      can also provide enhanced visualisation and interaction. Thespatial.tools            package contains spatial functions meant to      enhance the core functionality of theraster            package,      including a parallel processing engine for use with rasters.      Themicromap            package provides linked micromaps using      ggplot2.      Thespacetime            package      extends the shared classes defined insp            for      spatio-temporal data (see                          Spatio-Temporal Data in R            ). TheGrid2Polygons            converts a      spatial object from class SpatialGridDataFrame to      SpatialPolygonsDataFrame.

  An      alternative approach to some of these issues is implemented in      thePBSmapping            package;PBSmodelling            provides      modelling support. In addition,GEOmap            provides mapping      facilities directed to meet the needs of geologists, and uses thegeomapdata            package.

•              Handling spatial data            : A number of packages have      been written using sp classes. Theraster            package      introduces many GIS methods that now permit much to be done with      spatial data without having to use GIS in addition to R. It may be      complemented bygdistance, which provided calculation      of distances and routes on geographic grids.geosphere            permits computations of distance and area      to be carried out on spatial data in geographical coordinates.      Thespsurvey            package provides a      range of sampling functions. Thetrip            package extends sp      classes to permit the accessing and manipulating of spatial data for      animal tracking. Thehdeco            package provides hierarchical      decomposition of entropy for categorical map comparisons. TheGeoXp            package permits interactive graphical exploratory      spatial data analysis.spcosa            provides spatial coverage      sampling and random sampling from compact geographical strata.

  The UScensus2000 suite of packages (UScensus2000cdp,UScensus2000tract) makes the      use of data from the 2000 US Census more convenient. An important      data set, Guerry's "Moral Statistics of France", has been made      available in theGuerry            package, which provides data      and maps and examples designed to contribute to the integration      of multivariate and spatial analysis.      Themarmap            package is designed for downloading, plotting      and manipulating bathymetric and topographic data in R.marmap            can query the ETOPO1 bathymetry and topography database hosted by the      NOAA, use simple latitude-longitude-depth data in ascii format, and take      advantage of the advanced plotting tools available in R to build      publication-quality bathymetric maps (see the                          PLOS                        paper). Modern country boundaries are      provided at 2 resolutions byrworldmap            along with      functions to join and map tabular data referenced by country      names or codes. Chloropleth and bubble maps are supported and general      functions to work on user supplied maps (see                          A New R package for Mapping Global Data            . Higher resolution country      borders are available from the linked packagerworldxtra.      Historical country boundaries (1946-2012) can be obtained from thecshapes            package along with functions for calculating      distance matrices (see                          Mapping and Measuring Country Shapes            ).

  Thelandsat            package with accompanying                          JSS paper                        provides      tools for exploring and developing correction tools for remote      sensing data.taRifx            is a collection of utility and      convenience functions, and some interesting spatial functions.

•              Reading and writing spatial data -rgdal            :      Maps may be      vector-based or raster-based. Thergdal            package provides      bindings to                          GDAL                        -supported raster      formats and                          OGR                        -supported      vector formats. It contains functions to write raster      files in supported formats. The package also provides                          PROJ.4                        projection      support for vector objects      (                          this site                        provides      searchable online PROJ.4 representations of projections).      Affine and similarity transformations on sp objects may be made      using functions in thevec2dtransf            package.      The Windows and Mac OSX CRAN binaries ofrgdal            include subsets of possible data source drivers; if others      are needed, use other conversion utilities, or install from source      against a version of GDAL with the required drivers. The CRAN OSX      binary is for pre-Mavericks R; for Mavericks+, please see advice      given on R-sig-mac:                          for a binary package            , and                          for source install using the Kyngchaos frameworks            . Also see the      README file in the package, for example at                          R-Forge            . Thergeos            package provides functions for      reading and writing well-known text (WKT) geometry, and thewkb            package provides functions for reading and writing      well-known binary (WKB) geometry.

•              Reading and writing spatial data - other packages            :      There are a      number of other packages for accessing vector data on CRAN:maps            (withmapdata            andmapproj)      provides access to the same kinds of geographical databases as S -RArcInfo            allows ArcInfo v.7 binary files and *.e00      files to be read, andmaptools            andshapefiles            read and write ArcGIS/ArcView shapefiles; for NetCDF files,ncdf            may be used. Themaptools            package also provides helper functions for writing map polygon      files to be read by WinBUGS, Mondrian, and the tmap command      in Stata. It also provides interface functions betweenPBSmapping            andspatstat            and sp classes,      in addition tomaps            databases and sp classes. There is      also an interface to GSHHS shoreline databases.      For visualisation, the colour palettes provided in      theRColorBrewer            package are very useful, and may      be modified or extended using thecolorRampPalette            function provided with R. TheclassInt            package provides      functions for choosing class intervals for thematic cartography.      Thegmt            package gives a simple interface between GMT      map-making software and R.geonames            is an interface to      the                          www.geonames.org                        service. If the      user wishes to place a map backdrop behind other displays, the theRgoogleMaps            package for accessing      Google Maps(TM) may be useful.ggmap            may be used for      spatial visualisation with Google Maps and OpenStreetMap.      TheplotGoogleMaps            package provides methods for the      visualisation of spatial and spatio-temporal objects in Google Maps in      a web browser.plotKML            is a package providing methods for      the visualisation of spatial and spatio-temporal objects in Google      Earth. A further option isleafletR, which provides basic      web-mapping functionality to combine vector data files and online      map tiles from different sources.OpenStreetMap            gives access to open street map raster images,      andosmar            provides infrastructure to access OpenStreetMap      data from different sources, to work with the data in common R manner,      and to convert data into available infrastructure provided by      existing R packages.RSurvey            may be used as a processing program for spatially      distributed data, and is capable of error corrections and data      visualisation.

  Integration with version 6.* and 7 (devel) of the leading open source      GIS, GRASS, is provided in CRAN packagespgrass6, usingrgdal            for exchanging data.RPyGeo            is a wrapper      for Python access to the ArcGIS GeoProcessor, andRSAGA            is a similar shell-based wrapper for SAGA commands.

•              Point pattern analysis            : Thespatial            package is a recommended package shipped with base R, and contains      several core functions, including an implementation of Khat      by its author, Prof. Ripley. In addition,spatstat            allows freedom in defining the region(s) of interest, and makes      extensions to marked processes and spatial covariates. Its      strengths are model-fitting and simulation, and it has a useful                          homepage            . It is the only      package that will enable the user to fit inhomogeneous point process      models with interpoint interactions.      Thespatgraphs            package provides graphs, graph visualisation and graph based      summaries to be used with spatial point pattern analysis. Thesplancs            package also allows point data to be analysed      within a polygonal region of interest, and covers many methods,      including 2D kernel densities.

  ecespa            provides      wrappers, functions and data for spatial point pattern analysis,      used in the book on Spatial Ecology of the ECESPA/AEET. The      functions for binning points on grids inash            may      also be of interest. Theads            package perform first-      and second-order multi-scale analyses derived from Ripley's      K-function. Theaspace            package is a collection of      functions for estimating centrographic statistcs and computational      geometries from spatial point patterns.spatialkernel            provides edge-corrected kernel density      estimation and binary kernel      regression estimation for multivariate spatial point process data.DSpat            contains functions for spatial modelling      for distance sampling data, andspatialsegregation            provides segregation measures for multitype spatial point      patterns.GriegSmith            uses the Grieg-Smith method on      2 dimensional spatial data. Thedbmss            package allows      simple computation of a full set of spatial statistic functions of      distance, including classical ones (Ripley's K and others) and more      recent ones used by spatial economists (Duranton and Overman's Kd,      Marcon and Puech's M). It relies on spatstat for core      calculation.latticeDensity            contains functions that compute      the lattice-based density estimator of Barry and McIntyre, which      accounts for point processes in two-dimensional regions with      irregular boundaries and holes.

•              Geostatistics            : Thegstat            package      provides a wide range of functions for univariate and multivariate      geostatistics, also for larger datasets, whilegeoR            andgeoRglm            contain functions for model-based      geostatistics. Variogram diagnostics may be carried out withvardiag. Automated interpolation usinggstat            is available inautomap. This family of packages is      supplemented byintamap            with procedures for automated      interpolation andpsgp, which implements projected      sparse Gaussian process kriging.  A similar wide range of      functions is to be found in thefields            package. Thespatial            package is shipped with base R, and contains      several core functions. ThespBayes            package fits Gaussian      univariate and multivariate models with MCMC.ramps            is a different Bayesian geostatistical modelling package.      Thegeospt            package contains some geostatistical and radial      basis functions, including prediction and cross validation. Besides,      it includes functions for the design of optimal spatial sampling      networks based on geostatistical modelling.

  TheRandomFields            package provides functions for      the simulation and analysis of random fields, and variogram      model descriptions can be passed betweengeoR,gstat            and this package.SpatialExtremes            proposes several approaches for spatial extremes modelling      usingRandomFields. In addition,CompRandFld,constrainedKriging            andgeospt            provide      alternative approaches to geostatistical modelling. ThespTimer            package  is able to fit, spatially predict and      temporally forecast large amounts of space-time data using [1]      Bayesian Gaussian Process (GP) Models, [2] Bayesian Auto-Regressive      (AR) Models, and [3] Bayesian Gaussian Predictive Processes (GPP)      based AR Models. Thertop            package provides functions for      the geostatistical interpolation of data with irregular spatial      support such as runoff related data or data from administrative      units. Thegeorob            package provides functions for fitting      linear models with spatially correlated errors by robust and Gaussian      Restricted Maximum Likelihood and for computing robust and customary      point and block kriging predictions, along with utility functions for      cross-validation and for unbiased back-transformation of kriging      predictions of log-transformed data. TheSpatialTools            package has an emphasis on kriging, and provides functions for      prediction and simulation.

  Thesgeostat            package      is also available. Within the same general topical area are      thedeldir            andtripack            packages for      triangulation and theakima            package for spline      interpolation; theMBA            package provides scattered      data interpolation with multilevel B-splines. In addition,      there are thespatialCovariance            package, which      supports the computation of spatial covariance matrices      for data on rectangles, theregress            package      building in part onspatialCovariance, and thetgp            package. TheStem            package provides      for the estimation of the parameters of a spatio-temporal      model using the EM algorithm, and the estimation of the      parameter standard errors using a spatio-temporal parametric      bootstrap.FieldSim            is another random fields      simulations package. TheSSN            is for geostatistical      modeling for data on stream networks, including models based on      in-stream distance. Models are created using moving average      constructions. Spatial linear models, including covariates, can be      fit with ML or REML. Mapping and other graphical functions are      included.

•              Disease mapping and areal data analysis            :DCluster            is a package for the detection of      spatial clusters of diseases. It extends and  depends on thespdep            package, which provides basic functions for      building neighbour lists and spatial weights, tests for spatial      autocorrelation for areal data like Moran's I, and functions for      fitting spatial regression models, such as SAR and CAR models. These      models assume that the spatial dependence can be described by known      weights. TheSpatialEpi            package provides implementations of      cluster detection and disease mapping functions, including Bayesian      cluster detection, and supports strata. Regionalisation of polygon      objects is provided byAMOEBA: a function to calculate      spatial clusters using the Getis-Ord local statistic. It searches      irregular clusters (ecotopes) on a map, and byskater            inspdep. Theseg            package provides functions      for measuring spatial segregation.      Thespgwr            package contains an implementation of      geographically weighted regression methods for exploring possible      non-stationarity. Thegwrr            package fits geographically      weighted regression (GWR) models and has tools to diagnose and      remediate collinearity in the GWR models. Also fits geographically      weighted ridge regression (GWRR) and geographically weighted lasso      (GWL) models. TheGWmodel            package contains functions for      computing geographically weighted models. Thesparr            package provides another      approach to relative risks. TheCARBayes            package      implements Bayesian hierarchical spatial areal unit models. In      such models the spatial correlation is modelled by a set of random      effects, which are assigned a conditional autoregressive (CAR) prior      distribution. Examples of the models included are the BYM model as      well as a recently developed localised spatial smoothing model.      TheglmmBUGS            package is a helpful way of passing out spatial      models to WinBUGS. ThespaMM            package fits spatial GLMMs,      using the Matern correlation function as the basic model for spatial      random effects. ThePReMiuM            package is for profile      regression, which is a Dirichlet process Bayesian clustering model;      it provides a spatial CAR term that can be included in the fixed      effects (which are global, ie. non cluster specific, parameters)      to account for any spatial correlation in the residuals.      Thespacom            package provides tools to      construct and exploit spatially weighted context data, and further      allows combining the resulting spatially weighted context data with      individual-level predictor and outcome variables, for the purposes      of multilevel modelling. Thegeospacom            package gnerates      distance matrices from shape files and represents spatially weighted      multilevel analysis results.

•              Spatial regression            :      The choice of function for spatial regression will depend on the      support available. If the data are characterised by point support      and the spatial process is continuous, geostatistical methods may be      used, or functions in thenlme            package. If the support      is areal, and the spatial process is not being treated as continuous,      functions provided in thespdep            package may be used.      This package can also be seen as providing spatial econometrics      functions, and, as noted above, provides basic functions for      building neighbour lists and spatial weights, tests for spatial      autocorrelation for areal data like Moran's I, and functions for      fitting spatial regression models. It provides the full range of      local indicators of spatial association, such as local Moran's I and      diagnostic tools for fitted linear models, including Lagrange      Multiplier tests. Spatial regression models that can be fitted using      maximum likelihood include spatial lag models, spatial error models,      and spatial Durbin models. For larger data sets, sparse matrix      techniques can be used for maximum likelihood fits, while spatial two      stage least squares and generalised method of moments estimators are      an alternative. When using GMM,sphet            can be used to      accommodate both autocorrelation and heteroskedasticity. Spatial      count regression is provided using custom MCMC      byspatcounts. TheMcSpatial            provides functions      for locally weighted regression, semiparametric and conditionally      parametric regression, fourier and cubic spline functions, GMM and      linearized spatial logit and probit, k-density functions and      counterfactuals, nonparametric quantile regression and conditional      density functions, Machado-Mata decomposition for quantile      regressions, spatial AR model, repeat sales models, and      conditionally parametric logit and probit.      Thesplm            package provides methods for      fitting spatial panel data by maximum likelihood and GM.spatialprobit            make possible Bayesian estimation of the      spatial autoregressive probit model (SAR probit model).

•              Ecological analysis            : There are many packages      for analysing ecological and environmental data. They includeade4            for exploratory and Euclidean methods in the      environmental sciences,  theadehabitat            family of packages      for the analysis of habitat selection by animals      (adehabitatHR,adehabitatHS,adehabitatLT, andadehabitatMA),pastecs            for the      regulation, decomposition and analysis of space-time series,vegan            for ordination methods and other useful      functions for community and vegetation ecologists, and many      other functions in other contributed packages. One such istripEstimation, basing on the classes provided bytrip.ncf            has entered CRAN recently, and      provides a range of spatial nonparametric covariance functions.rangeMapper            is a package to manipulate species range      (extent-of-occurrence) maps, mainly tools for easy generation of      biodiversity (species richness) or life-history traits maps. Thesiplab            package is a platform for experimenting with      spatially explicit individual-based vegetation models.ModelMap            builds on other packages to create models      using underlying GIS data.      An off-CRAN package      -                          Rcitrus                        - is for      the spatial analysis of plant disease incidence.      TheGeneland            package usesfields            andRandomFields            to make use of both geographic and genetic      informations to estimate the number of populations in a dataset and      delineate their spatial organisation. Thengspatial            package      provides tools for analyzing spatial data, especially non-Gaussian      areal data. It supports the sparse spatial generalized linear mixed      model of Hughes and Haran (2013) and the centered autologistic model      of Caragea and Kaiser (2009).      The                          Environmetrics                        Task      View contains a much more complete survey of relevant functions      and packages.

CRAN packages:

• ade4

• adehabitat

• adehabitatHR

• adehabitatHS

• adehabitatLT

• adehabitatMA

• ads

• akima

• AMOEBA

• ash

• aspace

• automap

• CARBayes

• classInt (core)

• CompRandFld

• constrainedKriging

• cshapes

• dbmss

• DCluster (core)

• deldir (core)

• DSpat

• ecespa

• fields

• FieldSim

• gdistance

• Geneland

• GEOmap

• geomapdata

• geonames

• geoR (core)

• geoRglm

• georob

• geospacom

• geosphere

• geospt

• GeoXp

• ggmap

• glmmBUGS

• gmt

• Grid2Polygons

• GriegSmith

• gstat (core)

• Guerry

• GWmodel

• gwrr

• hdeco

• intamap

• landsat

• latticeDensity

• leafletR

• mapdata

• mapproj

• maps

• maptools (core)

• marmap

• MBA

• McSpatial

• micromap

• ModelMap

• ncdf

• ncf

• ngspatial

• nlme

• OpenStreetMap

• osmar

• pastecs

• PBSmapping

• PBSmodelling

• plotGoogleMaps

• plotKML

• PReMiuM

• psgp

• ramps

• RandomFields (core)

• rangeMapper

• RArcInfo

• raster (core)

• rasterVis

• RColorBrewer (core)

• regress

• rgdal (core)

• rgeos (core)

• RgoogleMaps

• RPyGeo

• RSAGA

• RSurvey

• rtop

• rworldmap

• rworldxtra

• seg

• sgeostat

• shapefiles

• siplab

• sp (core)

• spacetime (core)

• spacom

• spaMM

• sparr

• spatcounts

• spatgraphs

• spatial

• spatial.tools

• spatialCovariance

• SpatialEpi

• SpatialExtremes

• spatialkernel

• spatialprobit

• spatialsegregation

• SpatialTools

• spatstat (core)

• spBayes

• spcosa

• spdep (core)

• spgrass6

• spgwr

• sphet

• splancs (core)

• splm

• spsurvey

• spTimer

• SSN

• Stem

• taRifx

• tgp

• trip

• tripack

• tripEstimation

• UScensus2000cdp

• UScensus2000tract

• vardiag

• vec2dtransf

• vegan

• wkb

Related links:

• CRAN Task View: Environmetrics

• Rgeo: Spatial Statistics with R

• R-SIG-Geo mailing list