Urban green mapping

Urban green mapping

Year:2022

QGIS

Python

Description

Little for fun project :)

Problem statment:

In 2019, The city of Prague (capital of the Czech Republic) set a goal of planting a million trees on its territory within eight years

Details here: zastromujprahu.cz and adaptacepraha.cz

Out of curiosity, I want to map how total % urban green has improved after this policy was implemented.

It's best to use satellite observations for such task as it's very similar to landuse change mapping.

Altought, in this case, the simpler method based on normalized difference vegetation index (NDVI) can be used. This method utilized the Sentinel-2 satellite imagery.

How I did it:

1.) Download:

Shapefile of municipality districts of Prague city.

Cloudless Sentinel-2 satellite imagery from 2020 to 2021. Probably the easiest option for this is to use Semi-Automatic Classification Plugin for QGIS by amazing Luca Congedo.

Note: Sentinel-2 images were downloaded for summer period (June to August) from 2015 to 2021. This mostly corresponded to only one cloudless image only for each year.

Figure 1: Sentinel-2 Natural Color (combination of (B4, B3, B2))
Figure 2: Zoomed to Prague city centre
Figure 3: Sentinel-2 Color Infrared (combination of B8, B4, B3)

2.) Calculate NDVI

Equation for NDVI goes as: NDVI = (NIR-RED) / (NIR+RED).

These refers to: NDVI = (B8 - B4) / (B8 + B4) of Sentinel-2.

In my case, I did this by raster calculator in QGIS.

Figure 4: Sentinel-2 NDVI
Figure 5: NDVI, Zoomed to Prague city centre

3.) Divide image to non-vegetation and vegetation

Again, I used raster calculator with conditional statments where NDVI>0.19 was vegetation and NDVI<0.19 was non-vegetation.

Figure 6: Vegetation and non-vegetation urban map

4.) Calculate percentage of urban green in each municipality district

There are many ways how to handle this task. I used zonal statistics where raster values were summed to each polygon.

Lastly, to obtain percentage of vegatation the sum was divided by count.

Result:

Figure 8: Animation of urban green percentage in each satellite image

Also, let's have a look how Prague city centre vegetation has changed since 2015.

There was a positive trend in development. Altought, this result also highligted a seasonality changes in summer during wet/dry periods. The drop in 2020 was probably caused by unusual drought summer, thus, low-vegetatation (grass+shrub) died or has too low NDVI.

Figure 9: Temporal change of vegetation in Prague 1, 2, and 3

There was a positive trend in development. Altought, this result also highligted a seasonality changes in summer during wet/dry periods. The drop in 2020 was probably caused by unusual drought summer, thus, low-vegetatation (grass+shrub) died or has too low NDVI.

---

Note:

The results of this project cannot be interpreted as the absolute truth. It is always necessary to do further analysis.

There is also possibility to detect only trees, so more accurate analysis can be done in future.

---

Data sources: