[Q] What is the best resolution for Spatial CIMIS

[qjhart]

There is a question on final pixel size of Spatial CIMIS. We are debating to keep the 2k pixel size instead of going to 500m. See the message below with our pros and cons. We are thinking that 2km ETo is enough for the ETo variation. We are worried about storage and processing time related to 500 m. What is your opinion about it?

Notes from DWR 2024/09/15

I agree with you that ETo does not vary enough to justify going from 2 km to 500m pixel size. Below is the list of some Pros and Cons:

Pros:

• Using 500 m red band -> Better cloud detection -> Better K factor -> Slightly better Rs
• Smaller pixel during interpolation -> Better application of lapse rate -> Slightly better Tmin, Tmax Tdew (only in mountains, but anyway we are not sure on the lapse rate value, so this is not clear)

Cons:

• 16 times more data. one 2km pixel will convert into sixteen (16) 0.5 km pixel.
• More data -> 16X more storage
• More data -> more time to process Spatial Statistics

[qjhart]

In general, I would run this at the higher 500m resolution. There are a couple of reasons for this:

• Unless DWR made some drastic changes, I'm pretty confident the current setup runs at 500m and only downscales for the final output. So you would be changing even more processing

• If you run it at 2km, then you'll have less of a good method for testing new interpolations using old data.

• If later on you decide to run at 500m, then you'd need to recalculate everything, including the Amazon downloads.

• I made a real effort to create smaller per-day mapsets, and they are much smaller than what DWR is currenlty saving.

• I don't think the processing time changes that much.

[qjhart]

Our discussion with symsoft on 2024-10-04 indicated that internally, the process will be run at the 500m increment, while some of the output available to the public will be kept at the 2km resolultion.