Ocean data missing values · python-questions

Stream: python-questions

Topic: Ocean data missing values

Haiying Xu (Dec 07 2020 at 16:15):

I am dealing with ocean data missing values, I have to replace missing values to some values, then I can compress them. Afterwards I have to convert them back to missing values after I decompress them. How do ocean scientists usually process missing values, using which masks or combination of mask values to process different level of data? For example, I am looking at /glade/p/cisl/asap/abaker/compression_samples/pop-lens/orig/b.e11.B20TRC5CNBDRD.f09_g16.030.pop.h.WVEL.192001-200512.nc, it has different count of missing values on different z_w_top level. How do scientists process them case by case?

Michael Levy (Dec 07 2020 at 16:29):

@Haiying Xu z_w_top is the dimension in the vertical direction; as it increases, we expect more missing values in WVEL because the variable value should be missing under the bottom of the ocean. E.g. if one cell is 50 m deep and another is 100 m deep, then the first cell will have missing values below the 5th level and the second will have missing values below the 10th level (the first 15 levels are all 10 m thick)

Michael Levy (Dec 07 2020 at 16:30):

for variables on the t-grid (TLAT and TLONG), the variable KMT tells you how many levels have non-missing values. So WVEL(i,j,k) is a missing value if k > KMT(i,j) [using fortran counting, i.e. the top level is k=1]

Michael Levy (Dec 07 2020 at 16:31):

variables on the u-grid (ULAT and ULONG), use the variable KMU instead

Michael Levy (Dec 07 2020 at 16:31):

note that KMT(i,j) = 0 if (i,j) refers to a land cell so you don't need to have separate logic for land / ocean cells

Haiying Xu (Dec 07 2020 at 16:54):

Great, thank you Michael for your explanation. So do scientists check on missing values directly such as "if v == nan" or not? If not, is it matter if the value is not nan?

Last updated: Jan 30 2022 at 12:01 UTC