IMPORTANT

THE GVI R PACKAGE IS NO LONGER UNDER MAINTANACE. Please visit https://github.com/STBrinkmann/CGEI

GVI: Greenness Visibility Index

The GVI R package helps researchers compute the Greenness Visibility Index (GVI) presented by Labib, Huck and Lindley (2021) and includes optimisation of the underlying algortihm proposed by Brinkmann et al. (2022). The GVI is calculated using a Digital Surface Model (DSM), Digital Terrain Model (DTM) and Greenness Raster. GVI is written in C++ to provide fast and light weighted functionality.

Functions
Installation
Methods
Examples
Citation

GVI can also be computed in Python, by Jonny Huck

Functions

Function	Description
viewshed	Computes the viewshed of a single point on a Digital Surface Model
visualizeWeights	Helper function, to adjust spatial weight parameters in the vgvi and vgvi_from_sf functions
vgvi_from_sf	Computes the viewshed and calculates the proportion of visible greenspace; Supports multiple points, lines or polygons
sf_to_rast	Function for computing a continuous raster map from a sf object by interpolating the sf values using Inverse Distance Weighting (IDW).

Installation

Install R
- (Recommended) Install RStudio
Install terra
Install GVI:
- Install remotes
  
  install.packages("remotes")
- Install GVI
  
  GVI is still in very active development. Therefore, the package is also not on CRAN yet. You can install the latest version of GVI from GitHub with:
  remotes::install_git("https://github.com/STBrinkmann/GVI")

Methods

Viewshed

The viewshed function computes a binary viewshed of a single point on a Digital Surface Model (DSM) raster. A radial buffer is applied on the observer position, and visibility is beeing calculated usig a C++ implementation of Bresenham’s line algorithm [Bresenham 1965, Bresenham 1977] and simple geometry. The result of the viewshed function is a radial raster where 0 = no-visible and 1 = visible area.

Viewshed Greenness Visibility Index (VGVI)

The VGVI expresses the proportion of visible greenness to the total visible area based on a viewshed. The estimated VGVI values range between 0 and 1, where 0 = no green cells are visible, and 1 = all of the visible cells are green.

Based on a viewshed and a binary greenspace raster, all visible points are classified as visible green and visible no-green. All values are summarized using a decay function, to account for the reducing visual prominence of an object in space with increasing distance from the observer. Currently two options are supported, a logistic and an exponential function.

$\begin{align\*} f(x) = \cfrac{1}{1 + e^{b(x-m)}} && \text{(logistic)}\\\\ f(x) = \cfrac{1}{1 + (bx^{m})} && \text{(exponential)} \end{align\*}$

The full algorithm has been described in Brinkmann et al. (2022).

Visualize Weights

The visualizeWeights function helps to adjust spatial weight parameters m and b used in the vgvi and vgvi_from_sf functions.

Example output of the visualizeWeight function to compare and parameterize the decay weights of a logistic (left) and an exponential (right) function.

Viewshed Greenness Visibility Index (VGVI) from sf

The vgvi_from_sf function combines the viewshed function to compute a raster of visible cells, and calculates the VGVI using a binary GreenSpace map. The input of the observer location is an sf object (supports points, lines and polygons). If this object is a line or polygon feature, it will be converted to points. For every point, first the viewshed, than the VGVI is being calculated. The output of the vgvi_from_sf function is a single sf point object containing the VGVI for every point. This function supports parallel computation on Windows, Linux and MacOS, however it is highly recommended to use Linux or MacOS.

Examples

Data

For the first two examples we will use a Digital Elevation Model (DEM), a binary Greenspace Mask based on a land cover classification and a Digital Surface Model (DSM). The DSM is generated from LiDAR data collected in 2013. A detailed explanation on how the DSM has been generated using R is provided in this tutorial. To reduce the size of the R package, the sample data has been uploaded to a separate GitHub repository and needs to be downloaded first.

# Download DEM
DEM_tmp <- tempfile(fileext = ".tif")
download.file(url = "https://github.com/STBrinkmann/data/raw/main/GVI_Data/GVI_DEM.tif",
              destfile = DEM_tmp, mode="wb")

# Download DSM
DSM_tmp <- tempfile(fileext = ".tif")
download.file(url = "https://github.com/STBrinkmann/data/raw/main/GVI_Data/GVI_DSM.tif",
              destfile = DSM_tmp, mode="wb")

# Download GreenSpace
GS_tmp <- tempfile(fileext = ".tif")
download.file(url = "https://github.com/STBrinkmann/data/raw/main/GVI_Data/GVI_GreenSpace.tif",
              destfile = GS_tmp, mode="wb")

Load DSM, DEM and Greenspace Mask, and generate the observer location as a sf POINT feature.

# Load libraries. if one is not installed, use the "install.packages()" function
library(terra)
library(sf)
library(sfheaders)

# Load raster objects
DEM <- rast(DEM_tmp)
DSM <- rast(DSM_tmp)
GreenSpace <- rast(GS_tmp)

# Generate single observer point
observer <- st_sf(sf_point(c(492243.3, 5454231.4)), crs = st_crs(26910))

1. Single Point

Calculate the viewshed for a 200 meters radius around the observers position at 1.7 meters height (eye level).

library(GVI)
vs <- viewshed(observer = observer, dsm_rast = DSM, dtm_rast = DEM,
               max_distance = 200, observer_height = 1.7, plot = TRUE)

Left: Digital Surface Model (DSM); Right: Viewshed, where green = visible and yellow = no-visible area.

The Viewshed Greenness Visibility Index (VGVI) can now be calculated using the vgvi function.

vgvi_from_sf(observer = observer,
             dsm_rast = DSM, dtm_rast = DEM, greenspace_rast = GreenSpace,
             max_distance = 200, observer_height = 1.7,
             m = 0.5, b = 8, mode = "logit")$VGVI
#> [1] 0.5255536

The output of \~0.525 indicates, that \~53.5% of the visible area, within a 200 meters radius, is green.

2. Road Network

We also provide sample data of a sf line feature to demonstrate the vgvi_from_sf function. This feature represents roads and paths, that can be reached by walking within 5 minutes from our observer location.

# Download line feature and load as sf object
isodistance <- read_sf("https://github.com/STBrinkmann/data/raw/main/GVI_Data/isoline.gpkg")

Compute the VGVI along the line feature. Since the resolution is 5 meters, points along the line feature will be generated every 5 meters.

vgvi_sf <- vgvi_from_sf(observer = isodistance,
                        dsm_rast = DSM, dtm_rast = DEM, greenspace_rast = GreenSpace,
                        max_distance = 200, observer_height = 1.7,
                        m = 1, b = 3, mode = "exponential", cores = 12)

Output of the vgvi_from_sf function. Red points indicate low, green points indicate high values of VGVI. Paths and roads along the park tend to have high visible greenness, the road in the east has low visible greenness.

3. Large Area of Interest

Often researchers need to compute the VGVI for a large study area, therefore we have provided a sample workflow for the City of Vancouver. Since this workflow is extensive, we have uploaded it on an external website. Sample Datasets for Greater Manchester and Vancouver, BC can be downloaded from Zenodo.

About

Citation

Run this command to get info on how to cite this package.

citation("GVI")
#> 
#> To cite GVI in publications use:
#> 
#>   Brinkmann, S.T. and Labib, S.M. (2022). GVI: R package for computing
#>   VGVI for Spatial Raster (v1.1). doi:
#>   https://doi.org/10.5281/zenodo.7057132.
#> 
#> Ein BibTeX-Eintrag für LaTeX-Benutzer ist
#> 
#>   @Manual{,
#>     title = {GVI: R package for computing VGVI for Spatial Raster (v1.1)},
#>     author = {Brinkmann S.T. and Labib S.M.},
#>     year = {2022},
#>     url = {https://doi.org/10.5281/zenodo.7057132},
#>   }

Package contributors

Brinkmann, Sebastian (Creator and author) e-mail: sebastian.brinkmann@fau.de

Dr. S.M. Labib (Author) e-mail: sml80@medschl.cam.ac.uk

Paper authors

S.M. Labib (1, 2*)
Jonny J. Huck (1)
Sarah Lindley (1)

1: Department of Geography, School of Environment, Education and Development (SEED), University of Manchester, Arthur Lewis building (1st Floor), Oxford Road, Manchester M13 9PL, United Kingdom.

2: Centre for Diet and Activity Research (CEDAR), MRC Epidemiology Unit, University of Cambridge, Clifford Allbutt Building, CB2 0AH, Cambridge, United Kingdom.

*corresponding author

Bibliography

Bresenham, J.E. (1965): Algorithm for computer control of a digital plotter. IBM Systems Journal, vol. 4, no. 1, pp. 25-30, doi: 10.1147/sj.41.0025.

Bresenham, Jack (1977): A linear algorithm for incremental digital display of circular arcs. Commun. ACM 20, 2 (Feb. 1977), 100–106. doi: 10.1145/359423.359432.

Brinkmann, S. T., Kremer, D., and Walker, B. B. (2022): Modelling eye-level visibility of urban green space: Optimising city-wide point-based viewshed computations through prototyping, AGILE GIScience Ser., 3, 27, https://doi.org/10.5194/agile-giss-3-27-2022.

Labib, S.M., Jonny J. Huck, and Sarah Lindley (2021): Modelling and Mapping Eye-Level Greenness Visibility Exposure Using Multi-Source Data at High Spatial Resolutions. Science of The Total Environment 755 (February): 143050. doi: j.scitotenv.2020.143050.

STBrinkmann / GVI

readme