create configuration files #1
Conversation
There was a problem hiding this comment.
Hi Aki, good to quickly ask for feedback this way indeed! I have quite some comments, which are mostly questions. Let's see what's handy in terms of discussing. Maybe after reading through all of them, and giving a first reply, it's handier to shortly pick up the phone to discuss then to continue conversation in chat.
|
|
||
|
|
||
| auxiliary: | ||
| worldpop_url: https://data.worldpop.org/GIS/Population/Global_2021_2022_1km_UNadj/constrained/2022 |
There was a problem hiding this comment.
is this indeed the population source that is used? We should align it what is used in IBF portal, which is not necessarily the same at the moment. How do we best go about this? Should we start an overview somewhere with sources used per country, and agree on a process that IBF portal and IBF pipeline both draw from/link to this overview?
There was a problem hiding this comment.
this is used in the riverine flood pipeline I assume this descion wwas already made by DS. I think you have to updat ethe data inthe portal side
There was a problem hiding this comment.
We want to constantly improve our product in every iteration. So referring to "this is how it was done before" is not necessarily a good argument (although I admit we always have to find a balance, we cannot improve all at once).
@jmargutt do you maybe ideas on how to go about this?
There was a problem hiding this comment.
@jannisvisser @p-phung already set as objective the standardization of population layers across IBF (portal and pipelines), for every country. This specific source (WorldPop) is what we should be using, so we should update the portal accordingly instead of vice-versa, if you know what I mean.
| model: seasonal_rainfall_forecast | ||
| trigger_level: 50 | ||
| trigger_probability: 40 | ||
| area_threshold: 0.5 |
There was a problem hiding this comment.
what do all these properties mean here?
|
|
||
| countries: | ||
| - name: KEN | ||
| admin-levels: 2 |
There was a problem hiding this comment.
In Portal configuration, KEN drought is set to adminLevel 1, not 2. Is that not correct?
There was a problem hiding this comment.
wil change this back to 1, i was not sure
| Nov: | ||
| - OND: "0-month" | ||
| Dec: | ||
| - MAM: "2-month" |
There was a problem hiding this comment.
In December it is 3 months to the start of the season in March, not 2 months. I believe this was intentionally set up like this at the time, whereby Kenya was configured to be run "at the end of the month", whereas other pipelines were run at the beginning of the month. I would like to get rid of this exception, now that we unify drought across countries. So just make it generic. The pipeline can run at any time of the month, and if that is somewhere in December that implies a 3-month leadTime, not 2-month. Are you OK with changing that as well?
| - name: National | ||
| leadtime: | ||
| Jan: | ||
| - WET_SEASON: "0-month" |
There was a problem hiding this comment.
do you know more about the name of the season we should use in the case of LSO already?
| Feb: | ||
| - Belg_NAM: "0-month" | ||
| Mar: | ||
| - Belg_JAS: "3-month" |
There was a problem hiding this comment.
How is it reflected here that in March, you could both upload a 0-month ongoing event for Belg_MAM, and a 3-month upcoming event for Belg_JAS?
| for lvl in [1, 2, 3, 4]: | ||
| print(f"starting level {lvl}") | ||
| gdf = gpd.read_file( | ||
| rf"C:\Users\JMargutti\OneDrive - Rode Kruis\Rode Kruis\shapefiles\adm{lvl}_polygons.gpkg\adm{lvl}_polygons.gpkg" |
There was a problem hiding this comment.
same questions as above about the data source overview/alignment.
Also, put this filepath in a config file somewhere. And do not use path from Jacopo's personal onedrive.
| postgresql_port: 5432 | ||
| postgresql_database: global510 | ||
|
|
||
| countries: |
There was a problem hiding this comment.
Can you please already add Zimbabwe/ZWE to this list as well (even if commented out), to highlight that we want to include it as well.
And should ZMB also be in this list already? (Fine with me, if not)
| - name: Meher | ||
| leadtime: | ||
| Jan: | ||
| - meher: "0-month" |
There was a problem hiding this comment.
how do the names (e.g. 'meher' here) relate to the eventName you eventually use in the POST API-calls? In the current drought pipelines we agreed on "_" right? Where seasonName and regionName are again defined as in here, so e.g. "Belg MAM_Belg", where the part before the undescore ("Belg MAM") is then used as the eventName to display in portal/emails.
Developed a generic configuration file that defines the drought trigger model, threshold values, climate regions, and lead times for each area on a per-country basis. The CSV files in the config folder will link the climate regions with the admin level