augment historic cement data

[0UmfHxcvx5J7JoaOhFSs5mncnisTJJ6q]

List of things to do:

• review https://zenodo.org/record/3380081 for regional and temporal coverage
• if it adds information over readUSGS(), add it as a data source
• integrate readvanRuijven2016(), readUSGS(), and possibly readAndrew2019() into a calcCement() function that yields harmonised composite cement production data
• replace readvanRuijven2016() with calcCement() in industry projections

Tagging @fbenke-pik, who foolishly volunteered to work on this. I don't dare to assign you ;)

[fbenke-pik]

@0UmfHxcvx5J7JoaOhFSs5mncnisTJJ6q This link points to V4 from 2019, a newer version is available. Is this intentional?

[0UmfHxcvx5J7JoaOhFSs5mncnisTJJ6q]

I did not check thoroughly, but it seemed to me that the regional scope differs between versions of the data set.

[fbenke-pik]

Concerning task 1 on the task list above:

I checked V4 of the data from Andrew.

At first glance, it contains cement production data (tonnes) for the following countries:
Brazil, Chile, Indonesia, Lebanon, Mexico, Moldova, Mongolia, Namibia, South Africa, Togo, Jamaica, Korea, (Norway), Russia, Saudi Arabia, Ukraine

The other country sheets seem to only include emissions related to cement production or clinker production (clinker is not i scope for this, right?).

Reading in the data will give us a few additional / alternative data points for some of these countries. Sometimes as few as two or three, but in other cases a longer time series. E.g. for some countries (Ukraine, Chile, Mexico, Moldova), we get data starting from 1990 (USGS earliest year is 1998).

@0UmfHxcvx5J7JoaOhFSs5mncnisTJJ6q What do you think? Is it worth getting these additional data points?

[0UmfHxcvx5J7JoaOhFSs5mncnisTJJ6q]

I checked Armenia, which has clinker production, and clinker is an OK correlate to cement. We would have to figure out cement-to-clinker ratios, but those would be nice to have anyway. But the whole thing seems rather heterogeneous …

We do have long time series from 1970 on via readvanRuijven2016() that goes until 2011. So actually we are only looking for updates from 2012 on and maybe some overlap to gauge differences in the data.

https://zenodo.org/record/4738593/files/4.%20cement_production_5_final.csv seems to have collated data. Maybe we test reading that and comparing to van Ruijven manually first?

[fbenke-pik]

https://zenodo.org/record/4738593/files/4.%20cement_production_5_final.csv seems to have collated data. Maybe we test reading that and comparing to van Ruijven manually first?

I read in the data and plotted the cement production in comparison to Ruijven and USGS. You can find the html here

@0UmfHxcvx5J7JoaOhFSs5mncnisTJJ6q please take a look and let me know how to proceed.

[0UmfHxcvx5J7JoaOhFSs5mncnisTJJ6q]

First impression: not much new there compared to USGS. Can you send me your ~/dev/playground/readCement/4. cement_production_5_final.csv file so I can check some stuff on my end?

[fbenke-pik]

From the Zenodo description

"Cement production data by country are primarily derived from USGS statistics."

So your observation makes sense.

[fbenke-pik]

@0UmfHxcvx5J7JoaOhFSs5mncnisTJJ6q So looks like combining vanRuijven and USGS is sufficient. I assume you will also take care of the task item "replace readvanRuijven2016() with calcCement() in industry projections"?

Let me know if there is anything else I should do to finish this task.