Continuous surface mapping

Continuous surface maps use a method to aggregate points within a specified search radius to create a smooth surface that represents the density of events across the area. The popular method for creating maps of this type is called kernel density estimation. This method is now becoming increasingly more available and easier to use within standard GIS software (see  Contacts).

Kernel density estimation is a particularly useful method as it helps meet a number of aims of creating hotspot maps. The method,

• helps to more precisely identifying the location, the spatial extent and intensity of crime hotspots

• is visually attractive, so helps to invoke further enquiry and exploring the reasoning behind why crime and disorder is concentrated in some areas.

The density surface that is created can reflect the distribution of incidents against the natural geography of the partnership area. This may include representing the distribution of crime and disorder that follows natural boundaries such as reservoirs and lakes, or an alignment that follows a particular street along which there is a high concentration of offending. The method also helps to less subjectively and more accurately separate areas that are ‘hotter’ than others. In this sense, the method meets the main aims of hotspot mapping – it provides an accurate and invoking method that helps identify hotspots of crime and disorder for exploring and understanding in a more focused manner how crime and disorder is generated in these areas of high incident activity.

Issues with kernel density estimation.

Kernel density estimation does have its problems;

• Often the first question that is asked of the map is, ‘how many crimes are there in the hotspot?’ The map that is generated is a density surface relating to the number of crimes within a user-defined area (e.g. crime events per square kilometre). The density surface acts as a visual tool to guide further enquiry. Selection routines in GIS products can provide a count of the number of crimes within the area defined as the limit of the hotspot.

• Many agencies using these methods are becoming easily caught in the ‘false lure’ of the sophisticated looking geo-graphic they have produced, being reluctant to question its validity, or their accuracy in representing the underlying crime point distribution. The kernel density estimation method can produce attractive mapping output. However, it is vital to ensure that the underlying point data is accurate otherwise it may lead to merely creating a good looking map of wrong hotspots. Issues that relate to data quality are described in the data quality, data precision, and data protection section of this toolkit.

• Problems that relate to the setting of range methods still remain. Where this is particularly the case is when little regard is given to the legend thresholds which help decide when a cluster of crimes can be defined as a hotspot. This visual definition of a hotspot being very much left to the ‘whims and fancies’ of the map designer. For example, a map showing the distribution of crime as a continuous surface can have as little or as many hotspots on it depending on the ranges selected by the map designer to show spatial concentrations of these point events. Approaches providing ways forward for defining hotspot thresholds are described in the defining hotspot thresholds section of this toolkit.