> > You understand correctly: we have a sparse set of points and we want > to interpolate them into a raster. >
In that case, kriging is your best option. You need to get a good understanding of how kriging and variograms work in order to get optimal results. But if you get the basic parameters right, you should get at least satisfactory results. > We try IDW and Kriging but the result doesn’t fit. IDW is not a good choice for sparse and irregularly spaced data. > > Indeed, points represent bus stop and their attribute field is a > travel time: so, when a point/pixel is near 2 bus stops, the travel > time is equal to the shortest travel time of the 2 stops and not to > the sum. > With a spatial interpolation method, the interpolated value will always tend towards the travel time of the nearest bus stop (euclidean distance, not actual walking distance!). > > But we first want to use an easier method: we would like to create a > raster like a DEM: between 2 bus stops, the travel time increase or > decrease to reach the value of the travel time of the second stop. > Your result will be a weighted average between the two, with the nearer stop weighing more. Have you considered using a network analysis approach, instead? After all, your data is network data. And gvSIG has a very nice network extension. Cheers, Ben > > > I hope my explanations are clear enough. > > > > Thanks, > > > > Maude > > > > > > > From: Joachim Bergerhoff [mailto:agi...@gmail.com] > Sent: vendredi 19 août 2011 21:05 > To: Metais Maude > Subject: Fwd: [Sextante-users] make DEM from data points > > > > > > > ---------- Forwarded message ---------- > From: Benjamin Ducke < benjamin.du...@oxfordarch.co.uk > > Date: 2011/8/19 > Subject: Re: [Sextante-users] make DEM from data points > To: > Cc: sextante-users < sextante-users@lists.forge.osor.eu > > > > Hi Joachim > > Do I understand correctly: > > You have a sparse set > of points with recorded travel times? > > And you want to (a) interpolate them into a continuous > raster surface and (b) extract isochrones from the > result of (a)? > > The first step would be to choose an > interpolator function that can handle your sparse > points input. What have you tried? Kriging? IDW? > Splines? And using what settings? > > Cheers, > > Ben > > > > > > ----- Original Message ----- > > We are looking for an algorythm that can draw iso lines from a set > > of data points. Like drawing isometric lines or a DEM from a set of > > elevation data. (It is actually isochrone lines from a set of travel > > times). We've tried with a fillvoid function, but the voids between > > the points seemed to large to be filled. Any suggestion ? > > Thanks, Joachim > > > > > _______________________________________________ Sextante-users > > mailing list Sextante-users@lists.forge.osor.eu > > https://lists.forge.osor.eu/listinfo/sextante-users > > > ------ Files attached to this email may be in ISO 26300 format (OASIS > Open Document Format). If you have difficulty opening them, please > visit http://iso26300.info for more information. > > _______________________________________________ Sextante-users mailing > list Sextante-users@lists.forge.osor.eu > https://lists.forge.osor.eu/listinfo/sextante-users ------ Files attached to this email may be in ISO 26300 format (OASIS Open Document Format). If you have difficulty opening them, please visit http://iso26300.info for more information. _______________________________________________ Sextante-users mailing list Sextante-users@lists.forge.osor.eu https://lists.forge.osor.eu/listinfo/sextante-users
