As part of work in statistical mapping for school, I need to use QGIS to find the spatial extent of commercial nodes. A node is defined for this exercise as a zone of 75m in diameter where the density of shops is greater than or equal to 5.

The problem is less one of finding the spatial extent of the nodes than to find and map the nodes themselves. To find the nodes, I have 5 shapefiles, each corresponding to a type of commerce in Villeurbanne (the study area for this exercise). I have merged these five shapefiles into a single one, in the hope of simplifying my life (but perhaps that's a mistake...). I therefore have available a vector file where 2018 points represent the various shops in Villeurbanne.

What I would like to do is to analyze a circle of 75m in diameter (or a radius of 37.5m) around each point representing a shop, in order to know how many shops are found in this circle. I don't know whether it's possible with vector software... I know that a similar operation exists in raster software like ILWIS for analyzing data in continuous grids, because I have done that before, but I have to work in QGIS for this excercise.

Once the shops have been found within a 75m diameter zone, I will have to be able to create the regions of intersection, which would give me the nodes.

I'm a beginner with QGIS, an application that I have used only once or twice last year by following instructions, but this year no instructions are indicated.

Dans le cadre d'un travail à réaliser en carto statistique pour la fac, je dois déterminer l'étendue spatiale de nodules commerciaux sur Q-Gis. Un nodule est défini dans l'exercice comme une zone de 75m de diamètre où la densité de commerces est supérieure ou égale à 5.

Mon problème n'est pas tant de déterminer l'étendue spatiale des nodules que de déterminer et cartographier les nodules eux-mêmes. Pour déterminer les nodules, je dispose de 5 shapefile correspondant chacun à un type de commerce à Villeurbanne (la zone d'étude de l'exercice). J'ai rassemblé ces cinq shapefile en un seul, dans l'espoir de me simplifier la vie (mais c'est peut-être une erreur...). J'ai donc comme base de travail un fichier vectoriel où 2018 points représentent les différents commerces de Villeurbanne.

Ce que je souhaiterais faire, c'est analyser un cercle de 75m de diamètre (ou 37,5m de rayon) autour de chaque point représentant un commerce, pour savoir combien de commerces se trouvent dans ce cercle. Je ne sais pas si c'est possible sur un logiciel vectoriel... Je sais qu'une opération similaire existe sur les logiciels raster comme Ilwis en analysant l'information contenue dans les mailles contigues pour l'avoir déjà fait, mais je dois travailler sur Q-Gis pour cet exercice.

Une fois les commerces trouvés dans une zone de 75 m de diamètre, il faudrait que je puisse regrouper les zones d'intersection, ce qui me donnerait mes nodules.

Je suis novice sur Q-Gis, logiciel que je n'ai utilisé qu'une ou deux fois l'an dernier en suivant des fiches des consignes, mais cette année, aucune procédure n'est indiquée...

J'espère que ma question est claire

Merci d'avance à tous.

  • Could you use the GRASS add-in to QGIS? By the way, I believe your nodes are where the total count of shops in a zone is 5 or greater, not where the "density" is 5 or greater. – whuber Dec 30 '12 at 16:29
  • I think you can use the buffer tool (vector/geoprocessing tools/buffer). For what you describe it can make anything you want. Just make sure that your points are in CSR whose measuring units are in meters. – Gerardo Jimenez Dec 30 '12 at 22:37
  • @whuber : you're right, "density" isn't the proper word. I don't know how to use GRASS. I can add a Grass shapefile in my Q-GIS (Wroclaw). – Cécile Dec 31 '12 at 9:03
  • @GerardoJimenez : I've already used the buffer, using the shapefile with all the points representing the shops. It gaves me a shapefile with a lot of circles, the ones on the others.... It's a big mess... How can I know how many shops there are in each of those circles? – Cécile Dec 31 '12 at 9:07
  • The CSR of my points is WGS84, is it in meters ? – Cécile Dec 31 '12 at 9:15

Here is one method

  • First make sure you re-project your shapefile into a projected CRS with meters as units. Let's say this file is named 'points'

enter image description here - Next use Vector -> Geoprocessing Tools -> Buffer. Use 37.5 as the buffer distance. Call the buffer layer as 'buffer'.

enter image description here

  • Then use Vector -> Analysis Tools -> Point in Polygon. Use 'point' as input point layer and 'buffer' as input polygon layer. This will count how many points from your original shapefile fall within each of the buffered polygon. PNTCNT field will have the counts. enter image description here
  • Now select the polygons where PNTCNT >= 5 enter image description here
  • Then use Vector -> Geometry tools -> Polygon Centroids on the selection to get the original point locations. You can also just use the Attribute Table to identify ids of the original points.
  • 2
    your geoprocessing model is a good solution, but i think it would be better if, after making buffer, make intersection between "buffer_layer" features , then select the polygons result of more than 5 intersecting features. then get centroids of those polygon intersections. – geogeek Dec 31 '12 at 10:59
  • Thanks a lot to spatialthoughts and geogeek !! :) I've done as you said, considering the intersections, and it works ! Have a happy new year. – Cécile Dec 31 '12 at 15:27

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.