Data Subset for OMHCHO: OMI/Aura Formaldehyde (HCHO) Total Column 1-orbit L2 Swath 13x24 km V003

[EarthdataForumContributor]

I want to provide feedback about subset data download for OMHCHO V003 data. In the data subset, almost all the variables are cropped, but the corner points of latitude and longitudes (e.g. PixelCornerLatitudes and PixelCornerLongitudes) are not cropped. Hence they have a different dimension size as compared to latitude and longitude.

It would be nice to have these two variables also cropped.

[GES DISC - jimacker]

The variables 'PixelCornerLatitudes' and 'PixelCornerLongitudes' contain the coordinates of the corners of each pixel. The corner locations fall on either side of the pixel centers which are defined in the geolocation variables 'Longitude' and 'Latitude' which have dimensions nTimes and nXtrack. Extra elements are needed to define all four corners for each pixel, so the dimension sizes of the corner variables are (nTimes+1) and (nXtrack+1). Variables in the subset output files are cropped according to the mask which is based on the nTimes and nXtrack dimension sizes. Because the dimension sizes of the PixelCorner variables do not match all the other variables in the file, they do not get cropped.

If you require cropped pixel corners there is another product OMPIXCOR (https://disc.gsfc.nasa.gov/datasets?key ... COR&page=1) that contains pixel corner information that will crop as you expect. Enter the same search criteria for OMPIXCOR that you entered for OMHCHO and the cropped data for the two will correspond to one another.

NOTE that OMI data at the GES DISC come from multiple providers, and this issue only applies to data from this provider -- not to all OMI products.

[EarthdataForumContributor]

Many thanks for the additional information. I understand why it was difficult to crop also the pixel corners due to dimension size restriction.

Generally, with the other dataset, I have worked with, an additional dimension of length 4 is created for pixel corners, and hence the size of pixel corner variable is 4X(size of lat/lon variable).

However, now with the additional dataset containing the pixel corners, I can proceed further.