BuiltUp Mapping Using Sentinel 2A and Terra Package

by : Defani Arman Alfitriansyah

Forestry Student Faculty of Forestry and Environmental Science University of Kuningan, West Java, Indonesia

NDBI (Normalize Difference Built-Up Index)

Built-up area extraction can be effectively conducted using the Normalized Difference Built-up Index (NDBI), which was introduced by Zha et al. (2003) to automate the mapping of built-up land. The model was first applied to urban areas in Nanjing, China, achieving an accuracy of 92.6%, and has since been widely adopted for land cover identification, particularly for distinguishing built-up areas including residential settlements, using optical remote sensing data.

The NDBI formula is defined as:

\[\text{NDBI} = \left(\frac{\text{SWIR1} - \text{NIR}}{\text{SWIR1} + \text{NIR}}\right)\]

Calculate NDBI Index

Load library

library(terra)

terra 1.8.42

load raster file

SWIR1 = rast("D:/SPATIAL DATA/RASTER/SWIR1.tif")
NIR = rast("D:/SPATIAL DATA/RASTER/NIR.tif")

print(NIR)

class       : SpatRaster 
dimensions  : 931, 868, 1  (nrow, ncol, nlyr)
resolution  : 10, 10  (x, y)
extent      : 880180, 888860, 9223500, 9232810  (xmin, xmax, ymin, ymax)
coord. ref. : WGS 84 / UTM zone 48S (EPSG:32748) 
source      : NIR.tif 
name        : Grayscale 
min value   :      1090 
max value   :     17041 

print(SWIR1)

class       : SpatRaster 
dimensions  : 466, 434, 1  (nrow, ncol, nlyr)
resolution  : 20, 20  (x, y)
extent      : 880180, 888860, 9223500, 9232820  (xmin, xmax, ymin, ymax)
coord. ref. : WGS 84 / UTM zone 48S (EPSG:32748) 
source      : SWIR1.tif 
name        : Band_1 
min value   :   1167 
max value   :  11593 

Resample data

Ressample_band = resample(SWIR1, NIR)
print(Ressample_band)

class       : SpatRaster 
dimensions  : 931, 868, 1  (nrow, ncol, nlyr)
resolution  : 10, 10  (x, y)
extent      : 880180, 888860, 9223500, 9232810  (xmin, xmax, ymin, ymax)
coord. ref. : WGS 84 / UTM zone 48S (EPSG:32748) 
source(s)   : memory
varname     : NIR 
name        :   Band_1 
min value   :  1257.00 
max value   : 10425.06 

Math opration band

NDBI = (Ressample_band-NIR)/(Ressample_band+NIR)

Plot NDBI

plot(NDBI, col= topo.colors(100),
     plg= list(title= "Value"),
     main= "Normalize Difference Built Up Index")
grid(lty= 2,
     lwd= 0.5,
     col= "gray50")

NDBI

class       : SpatRaster 
dimensions  : 931, 868, 1  (nrow, ncol, nlyr)
resolution  : 10, 10  (x, y)
extent      : 880180, 888860, 9223500, 9232810  (xmin, xmax, ymin, ymax)
coord. ref. : WGS 84 / UTM zone 48S (EPSG:32748) 
source(s)   : memory
varname     : NIR 
name        :     Band_1 
min value   : -0.4451231 
max value   :  0.4360467 

Threshold Classification

In this analysis, I applied a threshold value of 0.002 to separate built-up from non-built-up land cover. The workflow was implemented in R using Sentinel-2 spectral bands (NIR and SWIR1), followed by resampling to align the raster grids and threshold-based classification. This approach demonstrates how NDBI can be used to detect urban areas and monitor land-use changes.

Threshold Equestion :

\[ B(x,y;T) = \begin{cases} 1, & I(x,y) > T,\\[4pt] 0, & I(x,y) \le T, \end{cases} \]

Threshold Function Explanation:

The threshold function \(C(x,y)\) assigns each pixel to a class based on the computed index value. If \(\mathrm{NDBI}(x,y) > T\), the pixel is classified as built-up (\(C=1\)); otherwise, it is classified as non-built-up (\(C=0\)).

threshold = 0.01
Built_up = NDBI>threshold
plot(Built_up, 
     col= c('white', 'red'),
     legend = FALSE,
     main= "Built Up Area")

grid(lty= 2,
     lwd= 0.5,
     col= "gray50")

Built_up

class       : SpatRaster 
dimensions  : 931, 868, 1  (nrow, ncol, nlyr)
resolution  : 10, 10  (x, y)
extent      : 880180, 888860, 9223500, 9232810  (xmin, xmax, ymin, ymax)
coord. ref. : WGS 84 / UTM zone 48S (EPSG:32748) 
source(s)   : memory
varname     : NIR 
name        : Band_1 
min value   :  FALSE 
max value   :   TRUE 

References

1. Muhaimin, M., Fitriani, D., Adyatma, S., & Arisanty, D. (2022). Mapping build-up area density using normalized difference built-up index (NDBI) and urban index (UI) wetland in the city Banjarmasin. IOP Conference Series: Earth and Environmental Science, 1089(1), 012036. https://lnkd.in/gR749BEF

2. Zha, Y., Gao, J., & Ni, S. (2003). Use of normalized difference built-up index in automatically mapping urban areas from TM imagery. International Journal of Remote Sensing, 24(3), 583–594. https://lnkd.in/gxvMPmHF