R miscalculating values when using gain()

Question

I'm downloading an average temperature from WorldClim for a set of specific locations (lat-long) using getData from raster package. The raw values are around 100. Then I apply gain() of 0.1 to all of them and R returns values of 1.0, while it should be 10.0. The problem applies only to WorldClim resolution 10 degrees (at 2.5 the temperature values are correct).

Here's what I'm doing:

library(raster)
library(sp)

Some random data

x <- runif(10, -72.85, -72.78)
y <- runif(10, 44.5, 44.6)
coords <- data.frame(x, y)

Downloading WorldClim data, applying gain of 0.1, binding lat-long and climate data in one data frame

r <- getData("worldclim",var="tmean",res=10)
gain(r) <- 0.1
points <- SpatialPoints(coords, proj4string = r@crs)
values <- extract(r,points)
df <- cbind.data.frame(coordinates(points),values)

Does anyone know why gain 0.1 changes 100 into 1.0 instead of 10.0?

Cheers

Welcome to SO! You will be most likely to get your question answered if you create a [reproducible example](https://stackoverflow.com/questions/5963269/how-to-make-a-great-r-reproducible-example). The example code in `raster::gain()` works as you suggest it should. — DaveArmstrong, Nov 24 '22 at 13:18
Your code isn't reproducible. You should include sample data rather than reading it from a file that we don't have. Also include `library()` calls for all the packages you use. — user2554330, Nov 24 '22 at 13:20

Robert Hijmans · Accepted Answer · 2022-11-26T18:07:27.553

You have uncovered a bug in the current version of "raster". This bug crept in because in that version methods from "terra" have replaced methods from "rgdal" (because that package will be retired). What happens in your example is that the scale and offset were applied twice. This problem goes away if you install terra version 1.6-46 and raster version 3.6-8. These are currently the development versions. On windows they can be installed with

install.packages('terra', repos='https://rspatial.r-universe.dev')
install.packages('raster', repos='https://rspatial.r-universe.dev')

With these versions, I see:

library(raster)
#Loading required package: sp

coords <- data.frame(5, 52)
r <- getData("worldclim",var="tmean",res=10)
#Warning message:
#In getData("worldclim", var = "tmean", res = 10) :
#  getData will be removed in a future version of raster
#. Please use the geodata package instead

extract(r, coords)
#     tmean1 tmean2 tmean3 tmean4 tmean5 tmean6 tmean7 tmean8 tmean9 tmean10 tmean11 tmean12
#[1,]     23     27     50     72    121    150    167    168    144     108      63      34
 
gain(r) <- 0.1
extract(r, coords)
#     tmean1 tmean2 tmean3 tmean4 tmean5 tmean6 tmean7 tmean8 tmean9 tmean10 tmean11 tmean12
#[1,]    2.3    2.7      5    7.2   12.1     15   16.7   16.8   14.4    10.8     6.3     3.4
 
sessionInfo()
#raster_3.6-8
#terra_1.6-46

Also, the suggestion about using the geodata package instead of getData is important. You are using obsolete data.

Thanks a lot. I see that in geodata there's no need to use gain(), nor spatialpoints(). Am I right? — curiousRguy, Nov 27 '22 at 17:31
Correct about gain. In neither case you need SpatialPoints. "geodata" is in the "terra" realm so SpatialPoints certainly do not apply. — Robert Hijmans, Nov 27 '22 at 17:39

R miscalculating values when using gain()

1 Answers1