‹ Back

Trial Methods #1

 

Mapping Observation Density: Parus major in Akershus [02-03-2026]

A small spatial experiment exploring how biodiversity observations can be transformed into simple density maps using open data and QGIS.

 

Purpose

 

The trial explores how public data can be transformed into quick spatial density map. The objective was to practice a simple workflow:

- Clean observation data

- Gecode locations

- Generate heatmap using kernel density estimations in QGIS

 

Species: Parus major (Great Tit (E), kjøttmeis (N))

Region: Akershus district

Date Filter: 1 january 2026

Sample size: ~50 observations

 

Data source

Observations were collected from Artsobservasjoner, Norway's citizen scince database. Collection was done manually, coordinates were not available in the interface, and therefore addresses and locality descriptions were extracted and geocoded. 

 

Work Flow

Step 1 Data compilation: OBservation data were copied to an Excel sheet and structured with fields for species, location description, municpilaity (in the district) and observation date

 

Step 2 Address standardization: Location strings were standardised to improve geocoding success (Place, Municipality, Akershus, Norway) SOme examples here:

 

Step 3 Geocoding: Addresses were converted to coordinates using a Python Script and the Open Street Map Nominatim geocoding service. 

• Vålerfeltet, Lierfoss, Aurskog-Høland, Akershus, Norway
• Gjønnesskogen, Bærum, Akershus, Norway
• Engene ind. omr., Sætre, Asker, Akershus, Norway
• Karusveien 8, Nittedal, Akershus, Norway
• Lurudveien 30, Lillestrøm, Akershus, Norway
• Engervannet, Bærum, Akershus, Norway
• Støtterudvegen, Rælingen, Akershus, Norway

 

Step 4 Import to QGIS: The geocoded dataset was imported as a point layer using latitude and longitude fields.

 

Step 5 Projection: The project was set to ETRS89/UTM Zone 33N (EPSG:25833) to ensure distance calculations use meters. 

 

Step 6 Density mapping: A kernel density estimation (heatmap) was generated with radius 1000 m and Pixel size 100 m. 

 

Interpretation

• Observation density reflects reporting activity, not population size

• Urban areas and accessible locations may have observer bias

• Some locations were approximate centroids due to vague locality descriptions

• One observation could not be geocoded and was excluded. 

 

What this method is useful for

The workflow can be useful for 

• rapid spatial screening of biodiversity observations
• identifying clusters of reporting activity
• preliminary spatial exploration before deeper ecological analysis

 

What's next?

• I will try to explore greater volume of observations of single species
• Contrast this amongst multiple species observations
• Seasonal observation density
• Comparison with protected areas or wetlands
• Integration with GBIF datasets