# QGIS Heatmap by value (NOT by point-densty)

I need a heat map generated by points value irrespective of how close or far apart the points may be in QGIS 2.6.1 So far I'm using the raster/heatmap/heatmap function. I need the same colour to run through all the points of the same value.

From the heatmap dialog, i'm using "use weight from field" which points to the right attribute, but I'm not getting the right result. As you can see this image shows that the heatmap that is generated has the same colour that runs through different values so it can't be right.

How do i get a heat map that is NOT controlled by point-density

• Do you aim at interpolation instead of point density? Mar 30, 2016 at 4:43
• +1 to previous comment. "Heatmap" should only be used for point density, because it originally mapped "hotspots" of accident or crime or fire locations etc. You just want to interpolate - and interpolate exactly - a surface through your data points. I say "exactly" because some interpolation routines can do a degree of smoothing which means the output surface doesn't go through your points. Lookup "inverse distance weighting" and "kriging"... Mar 30, 2016 at 7:48

What you actually want to perform is an interpolation. Heat maps are used to estimate point density between observed points. When you check "Use weights from field", you are expressing that each point has a count other than 1, but the method is still estimating a point density.

Interpolation, on the other hand, estimates values at unknown points between measured points. It can be used, for example, to estimate temperatures between weather stations. Interpolations may be exact, meaning the estimate matches the observed value at known locations, or inexact, which, like a regression line, may produce an estimate which does not match the observed value at known locations. There are three widely used exact interpolation methods: Voronoi Polygons (AKA Thiessen Polygons), Inverse Distance-Weighted (IDW), and Spline.

Rather than produce a tutorial on all three methods, I'll just say that IDW is a widely used method that you should probably start with, unless you have a good reason not to. It requires the Interpolation plugin, which I believe is installed by default, but may not be enabled. If (or when) it is enabled, it can be accessed from the Raster→Interpolation→Interpolation menu. Choose "Inverse Distance Weighting (IDW)" as the Interpolation method, and set the Interpolation attribute to the point value (the field you show in the labels in your map).

Try adjusting the symbology rather than running an analysis - graduated symbols?

• That's a great idea - in fact its such a good idea i thought the same thing! and i DID try that. I could get different colour symbols but visually it was very unclear. when you say graduated, what specifically do you mean?
– sixD
Mar 30, 2016 at 3:07
• Glad you tried that, be sure to describe what you have already tried in your original question. Graduated symbols example: mappingignorance.org/fx/media/2013/12/Figure-11.jpg You can symbolize it with no outline and a faded spectrum to give the visual effect you seek. The values you want to portray are already clustered together by your points with the quantity value. Mar 30, 2016 at 16:59
• Fair enough, though would have been a long and unclear question if i had done that. I can't see how to get the faded spectrum you describe, is that a style setting? if so i can't find it. [Note for those following, to achieve what @Stella described: from the layer/properties/style tab choose graduate, then choose "advanced" then "size scale field".]
– sixD
Mar 30, 2016 at 22:13

You might try creating an unweighted heat map raster then use the raster calculator to divide the weighted raster by the unweighted one and see if that gives you what you're looking for