CDO

Hi Diego,

I am not sure if they are developing new remap tools. If you were aware of any good remap algorithm, you could post requesting the implementation in future versions of CDO.

Kind regards,

Estanislao

hi!

I believe that conservative remapping should be able to work with strong gradients. but there are corner cases, esp. when source and target cells are very close to each other. For further investigation the input and output data would be very helpful for debugging and verification.

thx in advance + merry Christmas
ralf

Hi Estanislao and Ralf,

thanks for your answers. I have right now no specific example to send but the main problem for anthropogenic emissions is that one usually has a discontinuous field that should be non-negative, but after the remapping one obtains new extrema outside of the original range (due to the Gibbs phenomenon), leading to negative emissions or increased emissions and, therefore, no mass flux conservation between the original and the remapped field. I have issued a ticket, giving some more explanation and two references in the literature, one showing the problem and evaluating a solution and another describing the algorithm for the solution. Here, I give the references:

Explanation of the problem and evaluation of a solution: Marsico and Ullrich (2022)

The algorithm of the solution: Bradley et al. (2019)

Cheers,

Diego

PS. This problem is quite important in my project: We would like to verify emission estimates by taking the existing emission datasets as priors in a simulation.

hi Diego,

I have experienced some of those problems while remapping river runoff. Are those negative points random pixels? If that is the case you could filter that data with python, and then linearly interpolate as I did.

Kind regards,

Estanislao