Tracking bug for Drought Event issues

[pradh]

1. Madagascar drought event shows up with name "N/A": https://screenshot.googleplex.com/3aeetVhdtV8PEVQ, event: https://autopush.datacommons.org/browser/droughtEvent/2022-06_grid_1/-21_43

2. Horn of Africa Drought in 2022 isn't seen on the dashboard:

   • See our dashboard in Jun 2022: https://screenshot.googleplex.com/QEwEkDU5PPsSFSo
   • See similar stat (SPI 9-month) in Jun 2022: https://screenshot.googleplex.com/Fhcd5ebLQxsHP42

[pradh]

Hi Alex, do you mind looking at issue #2.

Issue #1 is a frontend thing.

[Fructokinase]

Differences between East African Drought drought data vs GPCC monthly SPI

1. Different measurement methods. The east African drought dashboard uses chirps as a measurement method, which is sattelite image based and has a far better resolution (0.05 degress) vs GPCC SPI data, which is derived from a bunch of weather stations on land. It could be that there are less weather stations in Africa in general.

2. This dashbaord uses gamma distribution while we use pearson (effect is unknown but calculation method is different.

How accurate are the GPCC monthly SPIs?

I ran a few validation scripts that maps GPCC SPIs (the one that we use) for USA, France, Sudan, and Somalia. It seems like the SPIs generally capture droughts where they exist.

SPI 9 month

• USA 9 month SPI

Compared for example to US drought.gov, which is also 9 month SPI, the shapes look very similar. Sep 2022 from drought.gov for example.

Also do look at drought.gov 2020. We can see the shapes of wet areas earlier in the year into a distinct separation of dryer western USA especially in Oct. So I don't think the current data from GPCC is wrong in terms of SPI 9.

• Sudan 9 month SPI

This can be compared with the datasource above.

• Somalia 3 month SPI

• Somalia 6 month SPI

• Somalia 9 month SPI

• France 3 month SPI

• France 6 month SPI

• France 9 month SPI

Here's a reference for France for comparison. Inside that you can see the SPI 3 and 6 month maps.

Why didn't Somalia show up on GPCC SPI?

Here's a list of SPI values for Somalia in 2022. It seems like 1 degree data show that it is on the dryer end, but not severely.

Lessons learned here:

• For things involved with precipitation, prefer sattelite data (with finer granularity) over weather station data.
• SPI is more like a measurement method itself rather than a variable. Maybe our schema can change to reflect that?

Next steps: Add sattelite based data, now that we know we can get SPI at 0.05 degrees of granularity.