loreabad6 / masters-thesis-geotech

BNA-EU: Advances of my master thesis for the Erasmus Mundus Masters of Geospatial Technologies

Hi there, I am Lorena! 👋

I am PhD researcher working with geospatial technologies.

• 🔎 I’m currently working as a researcher at the Risk, Hazard & Climate Lab and the EO-Analytics Lab of the Z_GIS department at the University of Salzburg
  • 🌍 My PhD research is focused on the use of big EO data and remote sensing methods to map and monitor geomorphological processes and landscape dynamcis.
• 🚲 I also enjoy sustainable transport as a research topic and a way of life
• 📊 I am interested in data visualization for scientific communication
• 💻 I really like programming in R. My motto: I bet you can do that in R ❗
  • I co-authored a 📦 for spatial networks analysis, check it out!

• 📦 Packages I use often for my research:

stars

• Want to support fairer academic evaluation? Try ProAc by Fabio Crameri. Here is mine:

• Want to know more? Check my CV!
• Not on my CV, my hobbies! Hiking ⛰️ | Reading 📚 | Cycling 🚴
• Find me at:

masters-thesis-geotech's Issues

Reproducibility issues: revisiting after 2 years

After having several issues with the dbConnect() function to reproduce the BNA-Europe.Rmd, I updated the code to change the dependency to Rpostgres instead of RPostgreSQL.

Also dbWriteTable() is has changed syntax, since schemas are not recognized as string vectors, so now:

dbWriteTable(
    conn,
    name = Id(schema = "received", table = "sa_boundary"),
    value = study_area_bb
  )

for example, should be the preferred syntax.

@elebrux the changes in the Rmd should fix the issue we were looking at:

dbWriteTable(conn, c("received","sa_boundary"), study_area_bb) 
Error in dbExistsTable(conn, name) : length(name) == 1 is not TRUE

Include a better way to extract the minimum area of analysis

There are issues in the country script for example in the UK in Corby, where an additional LSOA is added in spite of intersecting with the OSM boundary.

      DROP TABLE IF EXISTS generated.sa_pop_grid;
      CREATE TABLE generated.sa_pop_grid AS
      SELECT l.cell_id, l.lsoa11nm, l.population, l.jobs, l.geometry
      FROM received.lsoa l, received.sa_boundary b
      WHERE ST_Intersects(l.geometry,b.geometry)
      AND ST_Area(ST_Intersection(l.geometry, b.geometry))>100000;

The same goes for the Netherlands, in Delft, one buurt does not show because of the way I call the data:

      CREATE TABLE generated.sa_pop_grid AS
      SELECT l.cell_id, l.bu_naam, l.population, l.jobs, l.geometry
      FROM received.nl_buurt l, received.sa_boundary b
      WHERE ST_Intersects(l.geometry, b.geometry)
      AND (ST_Area(ST_Intersection(l.geometry, b.geometry)) > 500000 
            OR ST_Contains(b.geometry, l.geometry));

Besides, I am not using a buffer around the study area as it increases it in area considerably, and would mean that I should include them within the downloaded OSM data. I keep the boundary feature mainly for the GEOSTAT data set for general use in Europe, but will take a look at the results when I add the US to the possible analyses and compare results with PFB.

Boundary verification

Include a step where to confirm that the study area name that was provided corresponds to the actual boundary to be used. This is an issue, for example with Cardiff. If I call Cardiff, Wales it takes me to Australia. If I call Cardiff it takes me to the US, and only if I call Cardiff, Wales, UK it works.

Update SQL syntax

I am not sure when this changed, but currently running SQL chunks in R only allow one statement at a time. This is to avoid SQL injection.
MANY chunks in BNA-Europe.Rmd have really long multiple SQL lines, so an efficient way to migrate this should be thought of.