Modify columns' values depending on other columns - r

Question

My dataset contains NDVI values and NDVI-QualityDescriptor values(PixelQa) for different areas in different dates. I basically want to erase (setting to NA) the NDVI values that are related to bad quality descriptor (PixelQa). The number suffix of the column names relates both data: PixelQa_1 is related to NDVI_1 and so on.
Therefore to "clean" my data I have to check PixelQa values in order to assess if I have to change its related NDVI value. There is 3 possible situations:
PixelQa is NA -> then NDVI should be also NA.
Pixel Qa is 66±0.5 OR 130±0.5 -> then NDVI remains the same value.
Pixel Qa is different to 66±0.5 OR 130±0.5 -> then NDVI value is set to NA (this is bad quality data which needs to be ignored).
My dataset could be:
DataNDVI_split <- data.frame("21feb1987_NDVI" = c(0.123, NA, 0.192, 0.234, NA), "21feb1987_PixelQa" = c(66.30, NA, 66.00, 79.87, NA), "18jul1987_NDVI" = c(0.223, NA, 0.230, 0.334, NA), "21feb1987_PixelQa" = c(66.30, NA, 66.00, 79.87, NA), stringsAsFactors = FALSE)
DataNDVI_split
X21feb1987_NDVI1 X21feb1987_PixelQa1 X18jul1987_NDVI2 X21feb1987_PixelQa2
1 0.123 66.30 0.223 66.30
2 NA NA NA NA
3 0.192 66.00 0.230 66.00
4 0.234 79.87 0.334 79.87
5 NA NA NA NA
And "clean" it should look like:
DataNDVI_split <- data.frame("21feb1987_NDVI" = c(0.123, NA, 0.192, 0.234, NA), "21feb1987_PixelQa" = c(66.30, NA, 66.00, 79.87, NA), "18jul1987_NDVI" = c(0.223, NA, 0.230, 0.334, NA), "21feb1987_PixelQa" = c(66.30, NA, 66.00, 79.87, NA), stringsAsFactors = FALSE)
DataNDVI_split
X21feb1987_NDVI1 X21feb1987_PixelQa1 X18jul1987_NDVI2 X21feb1987_PixelQa2
1 0.123 66.30 0.223 66.30
2 NA NA NA NA
3 0.192 66.00 0.230 66.00
4 NA 79.87 NA 79.87
5 NA NA NA NA

Answer 1

Here's a tentative solution.
First, I'd split up the data into two separate dataframes, thus:
df_ndvi <- DataNDVI[grepl("NDVI", DataNDVI$Data), ]
df_ndvi
Data X21feb1987 X18jul1987
1 NDVI1 0.123 0.223
2 NDVI2 NA NA
3 NDVI3 0.192 0.230
4 NDVI4 0.234 0.334
5 NDVI5 NA NA
df_pixel <- DataNDVI[!grepl("NDVI", DataNDVI$Data), ]
df_pixel
Data X21feb1987 X18jul1987
6 PixelQa1 66.30 66.00
7 PixelQa2 NA NA
8 PixelQa3 66.00 124.23
9 PixelQa4 79.87 86.00
10 PixelQa5 NA NA
To perform the desired changes, there are many possible ways. One way is by using a forloop through all the columns in df_ndvi (except the first!) and defining an ifelse statement to see whether or not the conditions hold true and to define actions to be taken in either case:
for(i in 2:3){
df_ndvi[,i] <- ifelse(df_pixel[,i] < 65.5 | df_pixel[,i] > 66.5, NA, df_ndvi[,i])
}
This results in these corrections in df_ndvi:
df_ndvi
Data X21feb1987 X18jul1987
1 NDVI1 0.123 0.223
2 NDVI2 NA NA
3 NDVI3 0.192 NA
4 NDVI4 NA NA
5 NDVI5 NA NA
EDIT:
If you prefer to split-up the data in this way:
DataNDVI_split <- data.frame("21feb1987_NDVI" = c(0.123, NA, 0.192, 0.234, NA), "21feb1987_PixelQa" = c(66.30, NA, 66.00, 79.87, NA), "18jul1987_NDVI" = c(0.223, NA, 0.230, 0.334, NA), "21feb1987_PixelQa" = c(66.30, NA, 66.00, 79.87, NA), stringsAsFactors = FALSE)
DataNDVI_split
X21feb1987_NDVI X21feb1987_PixelQa X18jul1987_NDVI X21feb1987_PixelQa.1
1 0.123 66.30 0.223 66.30
2 NA NA NA NA
3 0.192 66.00 0.230 66.00
4 0.234 79.87 0.334 79.87
5 NA NA NA NA
then the for loop could be adapted thus:
for(i in c(1,3)){
DataNDVI_split[,i] <- ifelse(DataNDVI_split[,i+1] < 65.5 | DataNDVI_split[,i+1] > 66.5, NA, DataNDVI_split[,i])
}
The result is this:
DataNDVI_split
X21feb1987_NDVI X21feb1987_PixelQa X18jul1987_NDVI X21feb1987_PixelQa.1
1 0.123 66.30 0.223 66.30
2 NA NA NA NA
3 0.192 66.00 0.230 66.00
4 NA 79.87 NA 79.87
5 NA NA NA NA