Correctly geocoding lat and long in R

Question

I am trying to geocode a dataset using the given latitude and longitude. I am then going to use these points and join polygon features to them. However, the projection I am using for the points is incorrect. How do I go about this in a better more efficient way?

I begin by uploading the files and setting the projection. The polygon file I have already has a projection so I used that projection for the points.

fema <- readOGR(paste(fema.path, "/2010_DFIRM.shp", sep=""))
fema@proj4string
#CRS arguments:
 #+proj=lcc +lat_1=30.28333333333333 +lat_2=28.38333333333333 
 #+lat_0=27.83333333333333
 #+lon_0=-99 +x_0=600000 +y_0=3999999.9998984 +ellps=GRS80 +towgs84=0,0,0,0,0,0,0
 #+units=us-ft +no_defs 

data <- read.table(paste(inf.path, "/samp_HT_ANALYSIS", sep = ""), sep=",", fill=T)
coords <- data[,c(22:23)]
data.sp <- SpatialPointsDataFrame(data = data, coords = coords, proj4string =
            CRS("+proj=lcc +lat_1=30.28333333333333 +lat_2=28.38333333333333 +lat_0=27.83333333333333 +lon_0=-99 +x_0=600000 +y_0=3999999.9998984 +ellps=GRS80 +towgs84=0,0,0,0,0,0,0 +units=us-ft +no_defs"))

However, when I do the spatial join using the following code...

  test <- over(data.sp, fema, returnList = T)

No attributes overlap with my points.

I went on to map the points in ArcMap with a base layer map and they aren't mapped at all on the plane. Can someone please explain how I can solve this? Also, general information on projections in R would be great. I thought I could just set the CRS equal of one file equal to the other but I may have to reproject all of my files.

UPDATE: I have rerun my code using suggestions below. However, they are still not plotting correctly. I checked some of the indivudal lat long coordinates in my data and they are within my study region. Here is the updated code.

# Data temp for map
data.use <- read.csv(paste(inf.path, "/Houston_sample_subset.csv", sep = ""))
coords <- as.data.frame(matrix(0,nrow=nrow(data.use),2))
coords[,1] <- data.use$PROP_LATITUDE
coords[,2] <- data.use$PROP_LONGITUDE
data.sp <- SpatialPointsDataFrame(data = data.use, coords = coords, proj4string =
                                    CRS("+proj=longlat +datum=WGS84 +no_defs"))

# Study area
study.area <- readOGR("./FEMA_FLOOD/tl_2017_us_county/tl_2017_us_county.shp")
study.area@proj4string
study.area <- spTransform(study.area, crs(data.sp))
study.area <- subset(study.area, study.area$CBSAFP == 26420)

tiff("./PLOTS/test.tiff", width = 5, height = 5, units = "in", res = 800)
plot(study.area)
plot(data.sp, add =T)
dev.off()

The resulting plot is

Only the study region is plotted. I switched to plotting study region vs. fema maps because the fema maps take too long to generate.

Answer 1

I believe this is happening due to your SpatialPointsDataFrame being put into a wrong CRS. Your coordinates are in lat long correct? If so they should correspond to the WGS 1984 prj4, which is:

+proj=longlat +datum=WGS84 +no_defs

What you are doing is assigning lat long coordinates to a different projection. Unlike ArcMap, R does not do on the fly conversions. After assigning the proper coordinates to your spatial points you will have to convert one of your object's CRS to the other using the following function:

Here I am converting the fema shapefile's CRS to WGS 1984 (same projection as your data.sp)

fema <- spTransform(fema, crs(data.sp))

After both your spatial objects are in the same CRS, you can retry your Spatial Join.

Let me know if that makes sense!

Answer 2

I determined the solution. Lat and long were mixed up. Longitude and lattitude is the correct order (xy).

coords[,2] <- data.use$PROP_LATITUDE
coords[,1] <- data.use$PROP_LONGITUDE

is the correct way to set up the data.