Raster from point data in R different from point data in QGIS

Question

I have some point data, with longitude, latitude and a value, in .csv file format.

This is how the file looks in QGIS:

I wanted to create a raster of this map, with each of the points representing the center of a cell, in R, as practice. I first created an empty raster in R, and then added the values to the raster, like so:

##This is the datafile with the lon, lat and values:
        moisture_data <- read.csv("Suf_tab_lon_lat.csv", sep = ",", header = T)

        soil_raster <- extent(c(min(moisture_data$lon) - 0.0625,
                                max(moisture_data$lon) + 0.0625,
                              min(moisture_data$lat) - 0.0625,
                              max(moisture_data$lat) + 0.0625))
        soil_raster <- raster(soil_raster)
        res(soil_raster) <- 0.125
        projection(soil_raster) <- "+proj=longlat +datum=WGS84 +no_defs +ellps=WGS84 +towgs84=0,0,0"
        values(soil_raster) <- moisture_data$soil_moisture

However, when I plot the raster, this is how it looks:

I am unable to figure out why the data is getting flipped, and how to set this right. I am not sure how to post my original data, which might be more helpful, apart from simply emailing the CSV file to anyone who requests it.

Try using the last few lines of Simbamangu's advice here: https://gis.stackexchange.com/a/24602/67264 (starting with "From a csv") - you'll need to change the ncol and nrow accordingly but it might provide some insight on what's going on. — JepsonNomad, Apr 05 '19 at 21:51
If your points are on a nicely spaced grid in lat and long (and they look it) then using rasterFromXYZ will go from a data frame or matrix of rows with (long, lat, value) and build a raster over the bounds with NA in the missing pixels. Try it? — Spacedman, Apr 05 '19 at 23:41
Thanks a lot, @Spacedman, that did the trick, and was much more easier to boot! I have posted my code as an answer. — KVemuri, Apr 05 '19 at 23:56

KVemuri · Accepted Answer · 2019-04-06T00:30:36.660

Thanks a lot, @Spacedman and @JepsonNomad for your comments. The comment by @Spacedman was very helpful. This line of code worked for me:

    soil_raster2 <- rasterFromXYZ(moisture_data,res = 0.125,
                              crs = "+proj=longlat +datum=WGS84 +no_defs +ellps=WGS84 +towgs84=0,0,0",
                              digits = 3)
plot(soil_raster2)

And here is my plot (different dimensions, since I am working on a different machine, but you get the general idea):

Raster from point data in R different from point data in QGIS

1 Answers1