Coverage using network of streets

Tracing a path through a streets’ network from a starting point to calculate the coverage along the network, it’s likely that this distance does not coincide with a network’s node. It will end at an intermediate point of the final section.

This article presents a way to get coverage from a starting point until it reaches a maximum distance along the streets’ network. The main contribution is the calculation of the last partial segment of every path.

Specifically, this article focuses on how to resolve the problem in obtaining full and partial sections that conform the traveled distance along the network, using the ‘pgr_withPointsDD’ function from the ‘pgrouting’ library. Starting from a node that is outside the graph and is not any of the network nodes.

In order to use the code in this article, you should install the extension ‘pgrouting’ of PostgresSQL, that allows you to analyze networks. More information on (http: //docs.pgrouting. org) and (

ATTENTION: Next steps will modify the PostgresSQL database. Be careful, otherwise you may harm existing databases.

  • Download database here.
  • Create new database and name it ‘db_temp’.
  • Restore downloaded database.

Once done, you must have 3 entities:

  • “orig_network” (Streets’ network)
  • “origin” (Starting point)
  • “orig_network_vertices_pgr” (network nodes)

In order to understand this process, below is an example based on an orthogonal and symmetrical streets’ network. But can be used in any network you have.

Below is network ‘pep’, every section measures 100 meters, from node to node and the starting point ‘origin’ is painted in red, as you can see in Figure 1.


Fig 1. Orthogonal network, 100 meters from node to node.

Procedure of calculating coverage using ‘pgrouting’.

In order to calculate a 500 meters coverage from red dot, you must follow the instructions:

  • Duplicate the network and rename it as ‘network’..
  • Create a table named ‘poi_tmp’, which will contain the geometries of the starting points.
  • Modify ‘poi_tmp’ table, adding the fields(x,y,edge_id,side,fraction, newPoint).
DO $$
#variable_conflict use_variable
DECLARE dist_var INT;
dist_var:=500;DROP TABLE IF EXISTS "network";
CREATE TABLE IF NOT EXISTS "network" as (SELECT * FROM "orig_network");
drop table if exists poi_tmp;
CREATE TABLE if not exists poi_tmp as (SELECT ST_Centroid(tmp."geom") the_geom,tmp."id" as pid from (SELECT * FROM "origin") tmp);
ALTER TABLE poi_tmp ADD COLUMN     edge_id BIGINT;
ALTER TABLE poi_tmp ADD COLUMN     fraction FLOAT;
ALTER TABLE poi_tmp ADD COLUMN     newPoint geometry;
  • Update the ‘edge_id’ field with the nearest section identifier.
  • Update the ‘fraction’ field with the nearest section fraction.
UPDATE "poi_tmp" set "edge_id"=nearest_section."tram_id" from (SELECT distinct on( As Punt_id, as Tram_id, ST_Distance(Sg.the_geom,Poi.the_geom) as dist FROM "network" as Sg,"poi_tmp" AS Poi ORDER BY, ST_Distance(Sg.the_geom,Poi.the_geom), nearest_section where "poi_tmp"."pid"=nearest_section."punt_id";
UPDATE "poi_tmp" SET fraction = ST_LineLocatePoint(e.the_geom, "poi_tmp".the_geom),newPoint = ST_LineInterpolatePoint(e.the_geom, ST_LineLocatePoint(e.the_geom, "poi_tmp".the_geom)) FROM "network" AS e WHERE "poi_tmp"."edge_id" =;
  • Create ‘tbl_final_nodes_tmp’ table that will contain the ‘id’, aggregate costs and the starting node.
  • Create ‘geo_final_nodes_tmp’ table that will contain the ‘tbl_final_nodes_tmp’ and their geometries.
DROP TABLE IF EXISTS tbl_final_nodes_tmp;
CREATE TABLE IF NOT EXISTS tbl_final_nodes_tmp AS(SELECT node,agg_cost,start_vid FROM pgr_withPointsDD('SELECT id, source, target, cost, reverse_cost FROM "network" ORDER BY "network".id','SELECT pid, edge_id, fraction, side from "poi_tmp"',array(select "pid"*(-1) from "poi_tmp"),dist_var,driving_side := 'b',details := false));
DROP table if exists geo_final_nodes_tmp;
CREATE TABLE IF NOT EXISTS geo_final_nodes_tmp as (select "orig_network_vertices_pgr".*,"tbl_final_nodes_tmp"."agg_cost", "tbl_final_nodes_tmp"."start_vid", dist_var from "orig_network_vertices_pgr","tbl_final_nodes_tmp" where "orig_network_vertices_pgr"."id" ="tbl_final_nodes_tmp"."node" order by "tbl_final_nodes_tmp"."start_vid" desc,"tbl_final_nodes_tmp"."agg_cost");

After executing the previous code, you will obtaing the picture showed in Figure 2. The blue dot shows the projection of the starting point upon the nearest section, and this will be the starting point of every path calculated.


Fig 2. Nodes conforming the coverage.

The ‘pgr_withPointsDD’ function, returns the nodes that conform the 500m coverage, but if you want to know where exactly finishes the coverage, you can’t.

In order to resolve this problem, continue reading:

  • Create ‘final_sections_tmp’ table that will contain every section that will conform the coverage, including the final sections, but, as you may imagine, only a fraction of the every final section is valid.
DROP table IF EXISTS final_sections_tmp;
CREATE TABLE IF NOT EXISTS final_sections_tmp as (select "network"."id","network"."the_geom","geo_final_nodes_tmp"."id" as node,"geo_final_nodes_tmp"."agg_cost" as coste,(dist_var  "geo_final_nodes_tmp"."agg_cost") as falta,"geo_final_nodes_tmp"."start_vid" as id_punt,  (select case when (dist_var-"geo_final_nodes_tmp"."agg_cost") / ST_Length("network"."the_geom")<=1 then (dist_var-"geo_final_nodes_tmp"."agg_cost") / ST_Length("network"."the_geom") when (dist_var-      "geo_final_nodes_tmp"."agg_cost") / ST_Length("network"."the_geom")>1 then (1) end) as      fraccio from "network","geo_final_nodes_tmp" where ST_DWithin("geo_final_nodes_tmp"."the_geom", "network"."the_geom",1)=TRUE);

You can see the result of executing the previous code in Figure 3.


Fig 3. Section conforming part of the 500m coverage.

In order to obtain the fractions of the final sections that complete the 500m coverage, below is a function written in ‘PL/pgSQL’, that does the necessary modifications.

  • Create ‘Coverage’ function.
  • Create ‘fraction_sections_raw’ table, that will contain complete and fractioned sections that conform the 500m coverage.
r final_sections_tmp%rowtype;
m final_sections_tmp%rowtype;
DROP TABLE IF EXISTS fraction_sections_raw;
CREATE TABLE fraction_sections_raw (the_geom geometry, punt_id bigint,id_tram bigint,fraccio FLOAT,node bigint,fraccio_inicial FLOAT,cost_invers FLOAT,cost_directe FLOAT,target bigint,radi_inic FLOAT);
FOR r IN SELECT "final_sections_tmp".* FROM "final_sections_tmp" WHERE "final_sections_tmp"."id" not in (select "edge_id" from "poi_tmp")
insert into fraction_sections_raw VALUES(ST_Line_Substring((r."the_geom"),case when (select ST_Line_Locate_Point((r."the_geom"),(select "geo_final_nodes_tmp"."the_geom" from "geo_final_nodes_tmp" where "geo_final_nodes_tmp"."id"=r."node" and "geo_final_nodes_tmp"."start_vid"=r."id_punt")))<0.001 then 0 else 1-r."fraccio"
case when (select ST_Line_Locate_Point((r."the_geom"),(select "geo_final_nodes_tmp"."the_geom" from "geo_final_nodes_tmp" where "geo_final_nodes_tmp"."id"=r."node" and "geo_final_nodes_tmp"."start_vid"=r."id_punt")))<0.001 then r."fraccio" else 1
FOR m IN SELECT "final_sections_tmp".* FROM "final_sections_tmp" WHERE "final_sections_tmp"."id" in (select "edge_id" from "poi_tmp")
insert into fraction_sections_raw VALUES(m."the_geom",m."id_punt"*(-1),m."id",0,m."node",0,0,0);
LANGUAGE 'plpgsql' ;
PERFORM "the_geom" FROM coverage();

After that, is necessary to modify the ‘fraction_sections_raw’ table, in order to obtain the complete 500m coverage.

  • Update ‘fraccio_inicial’ table with the calculated value.
  • Update ‘cost_directe’ and ‘cost_invers’ fields with corresponding values.
  • Update ‘fraccio’ field with the calculated values.
  • Modify the geometry using the calulated ‘fraccio’ field.
update "fraction_sections_raw" set "fraccio_inicial"="poi_tmp"."fraction" from "poi_tmp" where "id_tram"="edge_id";
update "fraction_sections_raw" set "cost_directe"="network"."cost", "target"="network"."target", "cost_invers"="network"."reverse_cost" from "network" where "id_tram"="id";
UPDATE "fraction_sections_raw" SET "fraccio"=((case when ("fraction_sections_raw"."fraccio_inicial" * ST_Length("fraction_sections_raw"."the_geom"))>dist_var then (dist_var/ST_Length("fraction_sections_raw"."the_geom")) else "fraction_sections_raw"."fraccio_inicial" end)+(case when ((1-"fraction_sections_raw"."fraccio_inicial") * ST_Length("fraction_sections_raw"."the_geom"))>dist_var then (dist_var/ST_Length("fraction_sections_raw"."the_geom")) else (1-"fraction_sections_raw"."fraccio_inicial") end));
update "fraction_sections_raw" set "the_geom"=final."the_geom"from(select distinct(ST_Line_Substring((m."the_geom"),(case when (select ST_Line_Locate_Point((m."the_geom"), (select "the_geom" from "geo_final_nodes_tmp" where "geo_final_nodes_tmp"."id"=m."node" and "geo_final_nodes_tmp"."start_vid" = m."punt_id"*-1)))<0.01 then 0 else 1-m."fraccio" END),(case when (select ST_Line_Locate_Point((m."the_geom") , (select "the_geom" from "geo_final_nodes_tmp" where "geo_final_nodes_tmp"."id"=m."node" and "geo_final_nodes_tmp"."start_vid" = m."punt_id"*-1)))<0.01 then m."fraccio" else 1 END)))  the_geom,m."id_tram"from "fraction_sections_raw" m where m."id_tram" in (select "edge_id" from "poi_tmp")) final where final."id_tram" ="fraction_sections_raw"."id_tram";
insert into "fraction_sections_raw" (select SX."the_geom",PI."pid" as punt_id,SX."id"as id_tram,999 as fraccio,SX."source" as node,PI."fraction" as fraccio_inicial , SX."cost",SX."reverse_cost" from "network" SX inner join (Select "edge_id","pid","fraction" from "poi_tmp") PI on SX."id"=PI."edge_id");
UPDATE "fraction_sections_raw" set "the_geom"=final."the_geom" from (select ST_Line_Substring((SXI."the_geom"),(case when (FT."fraccio_inicial" - (dist_var/ST_Length(SXI."the_geom")))>0 then (FT."fraccio_inicial"-(dist_var/ST_Length(SXI."the_geom"))) else 0 end),(case when (FT."fraccio_inicial" + (dist_var/ST_Length(SXI."the_geom")))<1 then (FT."fraccio_inicial"+(dist_var/ST_Length(SXI."the_geom"))) else 1 end)) as the_geom, FT."punt_id", FT."id_tram", FT."fraccio" from "fraction_sections_raw"FT inner join (select SX."the_geom" as the_geom, SX."id" as tram_xarxa from "orig_network" SX, "poi_tmp" PI where SX."id"=PI."edge_id") SXI on FT."id_tram"=SXI.tram_xarxa where FT."fraccio"=999) final where "fraction_sections_raw"."punt_id"=final."punt_id" and "fraction_sections_raw"."fraccio"=999;
  • Create ‘fractions_sections_tmp’ table that will contain every section that conform the 500m coverage.
  • Create ‘driving_distance’ table that will contain the union of every section.
DROP TABLE IF EXISTS fraction_sections_tmp;
CREATE TABLE fraction_sections_tmp AS (select distinct(the_geom),punt_id,radi_inic from fraction_sections_raw);
DROP TABLE IF EXISTS driving_distance;
CREATE TABLE IF NOT EXISTS driving_distance AS (Select ST_Union(TOT.the_geom) the_geom, TOT."punt_id" from (select the_geom,punt_id,radi_inic from fraction_sections_tmp) TOT group by TOT."punt_id");
END $$;

Once executed the previous code, you should have the same as you can see in the Figure 4.


Fig 4.  Final 500m coverage.

In Figure 4 you can see the final sections cutted where the 500m path finishes. You can compare with Figure 3 and see the differences.


As a final conclusion, we created a function that obtains the coverage through a network , from a given starting point and a given distance or cost function. Improving the result obtained using the functions of the ‘pgrouting’ library.


The designed function is applicable to any routing function in which starts from initial points, that can be outside of the network, to obtain the desired coverage.

The code can be downloaded here.

Josep López Xarbau

Publicat per Josep López Xarbau

Enginyer en Electrónica i Automàtica Industrial (ETSEIAT-UPC). Durant els últims catorze anys (1998-2012) he realitzat tasques de R+D+I, en el camp de l’acústica mediambiental i de l’anàlisi territorial urbà. Durant tres anys (2001-2004) he estat el responsable del departament de R+D+I de l’Agencia d’Ecologia Urbana de Barcelona (BCNECOLOGIA). He col·laborat amb l’Àrea d’Acústica de l’Escola Universitària Politècnica de Mataró (EUPMt) en l’anàlisi i realització de mapes acústics de diversos municipis del Maresme, Vallès Oriental i l’Anoia. Actualment soc tècnic de Recerca, Desenvolupament e Innovació al NETLAB i al Centre de Coneixement Urbà (CCU) de la Fundació TecnoCampus Mataró-Maresme.

Deixa un comentari

L'adreça electrònica no es publicarà Els camps necessaris estan marcats amb *