Using HadISST SST Data to Replace SST_CPL in sst_HadOIBl_bc_1x1_2000climo_c180511.nc: Handling Land Values?

[jentus]

Hello everyone,

I’m currently working on replacing the sst_cpl data in sst_HadOIBl_bc_1x1_2000climo_c180511.nc with sea surface temperature (SST) data from HadISST. However, I’ve noticed that the HadISST dataset has missing values over land, while sst_cpl in sst_HadOIBl_bc_1x1_2000climo_c180511.nc includes varying SST values over land areas.

My questions are:

1. Is it appropriate to use HadISST data to replace sst_cpl, considering the missing values over land in HadISST?
2. How should I handle SST values over land? Would it be reasonable to set these land values to the sst_cpl dataset’s missing value, which is -1.8°C?

Any advice or suggestions on how to approach this would be greatly appreciated!

Thank you in advance!

 

[islas]

The model won't use the values over land but you do need to have non-NaN values in the file. Although, you might find that there are some grid points that you have that are missing that are considered to be land or partial land in the model e.g., around the coast lines. It is appropriate to use HadISST but you'll just need to fill in the missing values with something. I'd recommend not doing -1.8C in case there are some locations that are missing but are considered to be land by the model. It might be better to use some other in-filling procedure e.g., NCL's poisson_grid_fill poisson_grid_fill or setting the missing values around the coast lines to their nearest neighbor something equivalent.

 

[jentus]

The model won't use the values over land but you do need to have non-NaN values in the file. Although, you might find that there are some grid points that you have that are missing that are considered to be land or partial land in the model e.g., around the coast lines. It is appropriate to use HadISST but you'll just need to fill in the missing values with something. I'd recommend not doing -1.8C in case there are some locations that are missing but are considered to be land by the model. It might be better to use some other in-filling procedure e.g., NCL's poisson_grid_fill poisson_grid_fill or setting the missing values around the coast lines to their nearest neighbor something equivalent.

Hi islas，
Thank you for your reply, I will keep in mind your suggestions regarding filling in the missing values.

Additionally, I have a question: when I tried to replace cov with the HadISST_ice data, I found that sea ice also has values over land. Moreover, the cov_cpl values for April in the file sst_HadOIBl_bc_1.9x2.5_1850_2017_c180507.nc are even higher than the sic values for winter in HadISST. Is it reasonable if I only modify the winter sea ice for sensitivity testing, since the sea ice in March, April, and May is more extensive than that in DJF (winter)? Or do I need to replace the sea ice data for the entire year?

Thanks again for your help!

 

[jentus]

Hi islas，
Thank you for your reply, I will keep in mind your suggestions regarding filling in the missing values.

Additionally, I have a question: when I tried to replace cov with the HadISST_ice data, I found that sea ice also has values over land. Moreover, the cov_cpl values for April in the file sst_HadOIBl_bc_1.9x2.5_1850_2017_c180507.nc are even higher than the sic values for winter in HadISST. Is it reasonable if I only modify the winter sea ice for sensitivity testing, since the sea ice in March, April, and May is more extensive than that in DJF (winter)? Or do I need to replace the sea ice data for the entire year?

Thanks again for your help!

The cov_cpl values for May and June are also higher than the values in HadISST. I want to find the Fortran file where CESM reads the sea ice concentration (sic) to confirm whether sea ice over land is being read, but I haven't found it yet.