Earth-2 Model Intercomparison Project (MIP) is a python framework that enables climate researchers and scientists to inter-compare AI models for weather and climate.
I've been testing out ensemble runs on recent days (latest available data) with various datasource as initial conditions. The GFS one works with fcnv2_sm without any issues. For example, today is July 02, 2024, I was able to run it with today's conditions with GFS. However, I noticed that the IFS conditions are done using " root = "https://ecmwf-forecasts.s3.eu-central-1.amazonaws.com/" which throws an error of file not found. My guess is that this AWS does not have any recent data, only data from 20230118
I was able to make adjustments with this source instead: https://data.ecmwf.int/forecasts/20240702/00z/ifs/0p4-beta/oper/. I used requests.get instead of filesystem (it was causing some issues). The run was successful this morning, but now it looks like I'm missing some variables. I checked from the cache and I was missing u10, v10, no match for t2m and maybe some other slight mismatch (there's no tciwv).
Has anyone else been able to use recent IFS data?
Environment details
+ Followed the installation guide (git cloned, then used conda environment and built from source)
Version
source -main
On which installation method(s) does this occur?
Source
Describe the issue
I've been testing out ensemble runs on recent days (latest available data) with various datasource as initial conditions. The GFS one works with fcnv2_sm without any issues. For example, today is July 02, 2024, I was able to run it with today's conditions with GFS. However, I noticed that the IFS conditions are done using "
root = "https://ecmwf-forecasts.s3.eu-central-1.amazonaws.com/"
which throws an error of file not found. My guess is that this AWS does not have any recent data, only data from 20230118I was able to make adjustments with this source instead: https://data.ecmwf.int/forecasts/20240702/00z/ifs/0p4-beta/oper/. I used requests.get instead of filesystem (it was causing some issues). The run was successful this morning, but now it looks like I'm missing some variables. I checked from the cache and I was missing u10, v10, no match for t2m and maybe some other slight mismatch (there's no tciwv).
Has anyone else been able to use recent IFS data?
Environment details