hammerdirt-analyst/landuse

Issue on page /hex-3000-m.html

louiseschreyers opened this issue · 2 comments

louiseschreyers commented

Very interesting new page! So this was done for all locations except river locations, right? A few remarks here:

  1. Regarding the list of the key differences, I would add something about using hex now, instead of circles around the survey locations of interest. I read "The hexagon allows for better scaling. More of the shoreline conditions can be accounted for when hexagons are used': could you specify this a bit more (1-2 sentences?). I can also suggest sentences based on your explanation of this.
  2. I do not think it's needed to give the total area of hex, or if you do I would include the unit.
  3. Could you include the % of unlabelled surface?
  4. Do the unlabelled surface end up in the 'Undefined class'. Then maybe good to specify somewhere if that is not the case.
  5. "The buffer is a hexagon of 3’000 m" --> maybe specify "of 3'000 m of diameter"
  6. Fig. 5.1: for readability, the line color and width of the hex could contrast more with the background. Now it's not very clearly visible.
  7. I see in the text that the radius of the hex is 1,500, but on the figure legend, I read 2,000. Which one is it?
  8. "The old map layers used grids of 100 m². The new map layers uses polygons.": you could change polygons to vector data maybe. So the resolution is much increased, that's why it's better overall.
  9. In the legend, might be good to add the 'undefined surface', as it seems it makes up for quite a large part of the land-use for the example given. I would also re-label some legend items, 'Buildings' is enough for instance. And instead of 'Intersection length', something like 'River network'. The length is the variable, but it's not represented on the map.
  10. Small things to add on fig. 5.1: a scale, north arrow. And also removing the legend subtitles.
  11. 5.3.1: maybe replace "objects of interest" with "top items found".
  12. Check consistencies in the terms, for instance I see 'buildings', 'built', 'settlements' for the same term. And also 'wood'/'forests'.
  13. I am not sure I understood this sentence: "The removal of rivers is tantamount to removing zeroes given the consistently low survey results in the class. ". Could you specify somewhere how many survey locations you considered for this analysis?
  14. From the co-variance table, I see that the variable 'River network length' and 'Number of river intersections' seem to be closer to variables typical of rural locations (with a lot of land dedicated to agriculture and forestes). Whereas the distance to river intersection to more typically urban variables.
  15. What is the 'city center' variable'?
hammerdirt-analyst commented

Hi Louise,

There are alot of converging topics. I will get them grouped according to topic first. These are notes that are being used to build a piece of software that considers the environmental characteristics and the existing survey data to respond to the question:

What will I find at the beach today?

For the moment, unless the intention is to develop a particular figure or defintion for the manuscript it is not a priority. But that will change in the next couple of days. The use case of the geo data is to label arrays that store magnitude and location data for a variety of environmental data. I will group these and assign a priority. Some can be resolved here, if not it is best to open an issue and solve it.

  1. Punctuation, content, definition, figure styles
    1. Questions 1, 6, 8, 9, 10 <= Low priority in the short term.
      best solution: provide alternate text or figures <= submit a pull request
    2. other best solution: wait till that book is finished then we can see what is appropriate or needs to be developed for the manuscript.

For questions 2, 3, 4 ,5 7:
2: that is the constant that is being used to make the calculation. it can not be omitted.
3: why? does that mean everything should be reported that way? I have not been made aware of that.
4: yes: the answer is given in the document already important changes <= but not clearly maybe.
5: yes: it is given already link
7. The radius of the hex is 1´500 m and the intersection of rivers and canals is 2'000 meters. Yeah, nothing says that they have to be the same.
11. Objects of interest is the term I use. It is not the top items, it is what ever you happen to be looking for. This can easily be changed for manuscript figures.
12. that happens automatically when you label the arrays. consistency is obligatory.
15. City center is the historical city center.

13

Rivers have much lower survey values and a lower variety of items (in our survey results). Therefore, when the rivers are removed so are alot of zeroes. Ther are 211 possible selections in a survey. If only one object type is found at a survey, that means there are 210 object types that were not identified. This goes to values such as the average or median or the fail rate. The sampling conditions are completely different than lakes. Swiss rivers move pretty fast and consistently.

14

I interpret that as vineyards, woods, wetland, orchards and undefined land use are located closer to river intersections than, buildings, city centers and schools.

hammerdirt-analyst commented

UPDATE

The buffer for river intersections has been changed to 1500 m for all calculations