Learn R Programming

fields (version 16.3)

NorthAmericanRainfall: Observed North American summer precipitation from the historical climate network.

Description

Average rainfall in tenths of millimeters for the months of June, July and August (JJA) for the period 1950-2010 (old version, 1700+ locations) or 1971-2023 (new version, 4800+ locations) and for North America. The old version uses monthly data from the Global Historical Climate Network (GHCN) version 2 and the new version uses Version 4.

Arguments

Format

For both versions the format is a list with components: "longitude" "latitude" "precip" "elevation" "precipSE" "trend" "trendSE" "type" "x.s" "sProjection" with elevation in meters, longitude as (-180,180), latitude as (-90, 90) and precipitaion in 1/10 mm ( precip/254 converts to inches of rainfall). Note that the usual convention is to report a total amount of rainfall over the season. Divide by 3 to get the mean.

precip is the intercept for 1980.5 (1997 for new version) when a straight line least squares regression is fit to each station's record. SE is the companion standard error from the least squares fit. If the station is complete, then precip and precipSE will just be the mean and standard deviation adjusted for a linear trend. The estimated trend trend and and its standard error trendSE are also included.

Also due to the centering, for complete data the intercept and trend estimate will be uncorrelated. The component type indicates whether the station has been "adjusted" (see below) or is still in "unadjusted" form.

x.s is a useful transformation of locations into stereographic coordinates that reduces the inflation of North Canada due to the usual lon/lat coordinates. Specifically it is found by:


  library(mapproj)
  xStereo<- mapproject( NorthAmericanRainfall$lon,NorthAmericanRainfall$lat, projection="stereographic")
  NorthAmericanRainfall$x.s<- cbind( xStereo$x, xStereo$y)
  hold<- .Last.projection()
   NorthAmericanRainfall$orientation<- hold$orientation
   NorthAmericanRainfall$projection <- hold$projection

Examples

Run this code
data(NorthAmericanRainfall)
x<- cbind(NorthAmericanRainfall$longitude,  NorthAmericanRainfall$latitude)
y<- NorthAmericanRainfall$precip
quilt.plot( x,y)
world( add=TRUE)

Zstat<- NorthAmericanRainfall$trend / NorthAmericanRainfall$trendSE
quilt.plot( x, Zstat)

if (FALSE) {
# with a better projection
library( maps)
library( mapproj)
bubblePlot( NorthAmericanRainfall$x.s, 
NorthAmericanRainfall$precip,
size=.3, col=rev(tim.colors()) )
map("world", 
projection=NorthAmericanRainfall$projection,
orientation= NorthAmericanRainfall$orientation,
add=TRUE, col="grey30")
}
# note station sparsity outside CONUS and lower CA.

# comparing older and newer versions 

data(NorthAmericanRainfall)
bubblePlot( NorthAmericanRainfall$lon, NorthAmericanRainfall$lat,
NorthAmericanRainfall$precip,
size=.3, col=rev(tim.colors()) )
world( add=TRUE, col="grey30")

data(NorthAmericanRainfall2)
bubblePlot( NorthAmericanRainfall2$lon,NorthAmericanRainfall2$lat,
NorthAmericanRainfall2$precip,
size=.3, col=rev(tim.colors()), xlim= c(-130,-50))
world( add=TRUE, col="grey30")

Run the code above in your browser using DataLab