Agree with your suggestion. I can share how much the compression helps and how much grid interval approach works. With all due respect, it was fun reading your second reply. :-).
Regards Ghansham On Mon, Jan 6, 2020, 19:31 Maarten Sneep <[email protected]> wrote: > On 06/01/2020 04:27, Ghansham Sangar wrote: > > R/All > > > > Do we have some conventions for storing 2D lat/lon coordinates at an > interval. > > Say our data variable "band1" has dimensions (scans =24000, pixels= > 6000) and 2D > > lat/lon arrays at every 10th scan and 10th pixel, thus making dimensions > of lat/lon > > arrays (scans=2400, pixels=600). > > Modis generally provides lat/lon information for ocean color products > like this. Do > > we have something similar? This may be useful for radiometric swath > products specially. > > Before you put your users through the pain of dealing with incomplete > data, and data > where dimensions do not provide the required connections, make sure that > the > compression that is available will not solve the data volume issues. These > interpolated data are probably smooth enough that turning on shuffle and > compression > will reduce the data volume enough to allow for storing all geolocation > information. > You users will be grateful for that. > > Kind regards, > > Maarten Sneep > > -- > *KNMI* > *T*: +31 30 2206747 > *E*: [email protected] > *R*: A2.14 >
