Guidance for Processing DSCOVR EPIC MAIAC AOD 443 nm Data
Posted: Tue Jan 21, 2025 12:21 pm America/New_York
I am trying to process DSCOVR EPIC MAIAC data using R and the terra library. The issue I am encountering is that the projection is not being applied correctly. I am studying the South American region, so I am using Tile 31, as it is described as suitable for this area (Source: https://asdc.larc.nasa.gov/documents/dscovr/DSCOVR_VESDR_SDRG_02.pdf).
Using R, for a specific file, I can use the following code:
library(terra)
library(rhdf5)
library(hdf5r)
aod <- h5read(h5_files[1], "tile31/AOD_443")
This gives me a 1000x1000 matrix. However, the HDF5 file does not include longitude or latitude variables. and the matrix in rows and columns does not include this information either.
To address this, I created an extent based on the product description for Tile 31 (following the information in the pdf described before):
extent <- c(-115, -25, -90, 0)
r_aod <- rast(aod, extent = extent)
Now, I am unsure how to define the projection for this raster. According to the metadata for Tile 31, the projection is described as: Projection = "10018.7541 m SIN centered at -70"
I attempted to assign a sinusoidal CRS (ESRI:53008) to the raster and then project it to WGS84 (EPSG:4326), as my main goal is to crop the raster to an area of interest projected in EPSG:4326. However, the resulting raster does not match the area I tried to crop. This leads me to believe there is likely a projection issue.
Additionally, is there a user guide available specifically for this product? It has been quite challenging to use this product without additional information on how to process the data. Would you recommend another product derived from DSCOVR EPIC that includes AOD 443 nm data but is more straightforward to process?
Using R, for a specific file, I can use the following code:
library(terra)
library(rhdf5)
library(hdf5r)
aod <- h5read(h5_files[1], "tile31/AOD_443")
This gives me a 1000x1000 matrix. However, the HDF5 file does not include longitude or latitude variables. and the matrix in rows and columns does not include this information either.
To address this, I created an extent based on the product description for Tile 31 (following the information in the pdf described before):
extent <- c(-115, -25, -90, 0)
r_aod <- rast(aod, extent = extent)
Now, I am unsure how to define the projection for this raster. According to the metadata for Tile 31, the projection is described as: Projection = "10018.7541 m SIN centered at -70"
I attempted to assign a sinusoidal CRS (ESRI:53008) to the raster and then project it to WGS84 (EPSG:4326), as my main goal is to crop the raster to an area of interest projected in EPSG:4326. However, the resulting raster does not match the area I tried to crop. This leads me to believe there is likely a projection issue.
Additionally, is there a user guide available specifically for this product? It has been quite challenging to use this product without additional information on how to process the data. Would you recommend another product derived from DSCOVR EPIC that includes AOD 443 nm data but is more straightforward to process?