Assign Closest Points to Members of Second Point Set

Download data and R Code for this example

Project Requirement:

  • Read two geocoded point sets from Comma Separated Value (CSV) files into R data objects.
  • Assign to each member of first point set the geographically closest point from second set.

Input Data - Measurements at two different point location sets:

File 1: Population density

File 2:Air Temperature



1) Read input files into R Spatial data objects.

2) Filter/remove invalid points from temperature measurements.

2) Assign closest temperature point to each population density point.

3) Write results - the point assignments file - as an ASCII .CSV file.

The R Script:

# Latest version: Assign closest points from a second point set

   temperaturePoints <- read.csv("temperaturePoints.csv")                                      
   densityPoints <- read.csv("densityPoints.csv")

# promote the input lists to SpatialPointsDataFrames

   coordinates(temperaturePoints) <- c("longitude", "latitude")
   coordinates(densityPoints) <- c("longitude", "latitude")             

# Remove temparature points with 'invalid' value of -199

   validTempPoints <- temperaturePoints[!temperaturePoints$temperature %in% c(-199),]

#  Define these vectors, used in the loop.

   closestSiteVec <- vector(mode = "numeric",length = nrow(densityPoints))
   minDistVec     <- vector(mode = "numeric",length = nrow(densityPoints))

# Get the vector index of the temperature station closest to each field station.
# Use the spDistsN1 function to compute the distance vector between each
# field station site and all of the temperature stations. Then, find and
# retain the actual temperature, and the index of the closest temperature
# to each transect station.
# spDistsN1 usage: spDistsN1(pointList, pointToMatch, longlat)
# where:
#         pointList   : List of candidate points.
#         pointToMatch: Single point for which we seek the closest point in pointList.
#         longlat     : TRUE  computes Great Circle distance in km,
#                       FALSE computes Euclidean distance in units of input geographic coordinates
# We use Great Circle distance to increase distance calculation accuracy at high latitudes
# See the discussion of distance units in the header portion of this file
# minDistVec stores distance from the closest temperature station to each density measurement point.
# closestSiteVec stores the index of the closest temperature station to each density measurement point.
   for (i in 1 : nrow(densityPoints))
      distVec <- spDistsN1(validTempPoints,densityPoints[i,],longlat = TRUE)
      minDistVec[i] <- min(distVec)
      closestSiteVec[i] <- which.min(distVec)
# Create the Temperature Assignment table: merge the temperature point list with the transect point list
# into a five-column table by merging the temperature point and transect point lists.
   PointAssignTemps <- as(validTempPoints[closestSiteVec,]$temperature,"numeric")
   FinalTable = data.frame(coordinates(densityPoints),densityPoints$density,
# Update the Temperature Assignment table column names 
   names(FinalTable) <- c("Long","Lat","Density","CloseTempIndex","Distance","Temperature")
# And, at the end, write the point assignment file.
  message("Write temperature/density assignment table to disk in .csv format")

Results - Temperature Assignment table with point assignments and support data:

File 1: Temperature Assignment table

Learning More:

The GIS Primer: The Nature of Spatial Information

The CRAN (R) Task View summarizes the R Analytical Environment's spatial data analysis and management tools.

Point of Contact for this Use Case: reeves [at] (Rick Reeves), NCEAS Scientific Programmer
This Use Case was updated June, 2010.