Alternative download method for Brazilian data

[kratzert]

Apparently, data from Brazil can also be retrieved from https://www.snirh.gov.br/hidroweb/download, see Inventario20_10_2025.zip (note the date, this file seems to get updated frequently), which is a single database file that includes the data from all stations. Unclear yet which data and how many stations, but worth checking since the API is terribly slow.

[thiagovmdon]

Apparently, data from Brazil can also be retrieved from https://www.snirh.gov.br/hidroweb/download, see Inventario20_10_2025.zip (note the date, this file seems to get updated frequently), which is a single database file that includes the data from all stations. Unclear yet which data and how many stations, but worth checking since the API is terribly slow.

I will investigate a bit. Although I am Brazilian, I am not very well versed in working with our data. Just as a context, it seems that we had a very nice API some time ago, developed by a MSc student, however the Federal Water Agency had some issues, and in the end blocked all (including this good one). Now we have this that was implemented. I will investigate with friends whether is there a better alternative currently.

[kratzert]

Cool, thanks. Also, if you can understand the column names in the metadata csv and figure out if any of the column gives us information about a station having streamflow/stage data or not, that would be great. I already tried to filter the full list by what I thought means that river/water data is available. This reduced the number of stations in the metadata table from ~37k to ~4.5k. I am running a job since yesterday to query the data for these 4.5k stations but the vast majority of the station doesn't have streamflow records.

The homepage from above was shared by Henning Plessow from GRDC, who told me that ANA told GRDC to get the streamflow data from there.

[kratzert]

I actually had a colleague looking into the mdb files and aparently there is no timeseries data in them. Just the station list and otherwise empty tables...