Atmospheric Toolbox

Issues with harpmerge CLI tool

HI there,

I’m trying to merge three NO2 S5P swaths into a gridded dataset over Australia. I’m using Harpmerge for it, but the merged netcdf file renders no data when it’s visualised in VISAN.

Command (inspired by the RUS training video)
harpmerge -ap ‘bin();squash(time, (latitude, longitude))’ -a ‘latitude > -54.750420830999985 [degrees_north];latitude < -9.240166924999869 [degrees_north];longitude > 112.91944420700005 [degrees_north];longitude < 159.10645592500003 [degrees_north];bin_spatial(2275,-54.750420830999985,0.02,2309,159.10645592500003,0.02);derive(latitude {latitude});derive(longitude {longitude})’ .\processed\ .\

The only difference in my setup is that HARP is now version 1.11. Has there been a change in the module since 1.9.2?

Any help would be greatly appreciated. Thanks.


You are filtering longitude between 112.9194 and 159.1065, but then use 159.1065 as the offset for the spatial regridding. This will, of course, not provide any valid data in your grid. :wink:

Be aware that HARP 1.12 is now the latest version, so I would recommend you use that.

Also, be aware that you should perform the QA filtering tropospheric_NO2_column_number_density_validity>70 filter to only include ‘good’ measurements.
Furthermore, I would advise you to add a keep() operation to only keep the variables that you need. This will greatly improve the performance of your harpmerge.

I would suggest you read some of the other topics on this forum regarding the S5P NO2 regridding. This should help you prevent making a lot of the initial mistakes.

Thanks for your reply.

Yes I see the issue, I had been a bit confused by the documentation and the ESA supported RUS training which says that the coordinates should be from lower right corner. But now that you’ve pointed out this issue it is clear in the documentation that longitude goes from low to high numbers. Thanks for pointing this out.

I will of course filter for QA, the documentation suggests > 75 for NO2, is >70 a new threshold?


Sorry, that was a typo. It should indeed be >75.