A couple of population improvements

[koldunovn]

In the current main, there is a warning:

DtypeWarning: Columns (2) have mixed types. Specify dtype option on import or set low_memory=False.
  pop_dat = pd.read_csv(pop_path)

Can we specify dtype when open the data? Probably will five us some speedup.

Here https://github.com/koldunovn/climsight/blob/e4bf6d07f9f7b0cfbb60d9c123bc33b6665e22de/climsight.py#L498 We load population data once again. Instead I would load it once and pass to function as argument, it will save us one load.

Here https://github.com/koldunovn/climsight/blob/e4bf6d07f9f7b0cfbb60d9c123bc33b6665e22de/climsight.py#L791 we might check if data are not None and save a function call.

[AntoniaJost]

Totally agree that loading the data twice is not the efficiency we are looking for. I will fix this soon.

[AntoniaJost]

With respect to the warning: I am aware of it and have tried many ways to remove it but didn't manage to do so. Would appreciate some help from you @koldunovn .

[AntoniaJost]

Thoughts on new / different data sets

Gridded population of the world (GPW):

• data shouldn't be used for time-series analysis
• population estimates for all five years are derived from same input population (2010)
• same growth rate is used to extrapolate to all years --> overall doesn't reflect a true time series
• no prediction

WorldPop:

• no prediciton
• unconstrained data only available for individual countries, not global mosaic (way too many files to download)
• and for that "only" in 1km resolution instead of 100m
• only population count (population density only for individual countries)
• population for 2020

Global Human Settlement Layer:

• projections for 2025 to 2030 (derived from GPW)
• 100m resolution
• only population count
• GHS population grid, 1km
• GHS-BUILT-C (Mollweide projection), 10m
• GHS-DUC (Degree of Urbanisation)
• https://ghsl.jrc.ec.europa.eu/download.php?ds=pop --> best one so far, however, still no information about population growth rate, life expectancy, and migrations