It's been a while since I've written on Remote Sensing - my last post was on the Oil Spill in Mauritius (MV Wakashio running aground in August 2020). In this post, I will use Synthetic Aperture Radar (SAR) Imagery acquired by Sentinel-1 satellite to map the built-up area, technically known as Urban Footprint, for a couple of Indian cities across two points in time for change-detection purposes.
For context, not only does India occupy three spots in the Top 10 fastest growing cities of the world as of today, but it could also occupy all the 10 spots in the list henceforth until 2035! While urban centers are hotbeds of commercial activity, innovation and growth, rampant urbanization is fraught with risk for obvious reasons.
So how does the Imagery captured by Earth Observation satellites help to monitor Urbanization?
With Remote Sensing, one is able to extract information about the characteristics of surface features based on how it responds to radiation. The source of radiation could be passive (Sunlight typically) or active (the transmitter instrument onboard the satellite itself). Upon interacting with features on the surface of the Earth, the reflected radiation is captured by the Receiver instrument onboard the Earth Observation satellite itself. The extent of radiation energy that is reflected back towards the satellite depends predominantly on two broad factors - the structural and biogeochemical properties of the object and the physical and geometric characteristics of the energy wavelengths it is exposed to. The latter is already known while the reflected energy (final output) is acquired by the Receiver instrument - thus, one can derive the missing piece of information (surface type) by referring to the vast body of knowledge on reflectivity characteristics of different surfaces derived through scientific research over the years. This is the fundamental method involved in any Remote Sensing workflow - be it the detection of Urban Footprint, Water Bodies, Vegetation etc. (this helps to delineate too - distinguishing Land pixels from Water pixels, Healthy Vegetation from Distressed Vegetation etc).
SAR Imagery is acquired through active illumination - the satellite itself sends pulses of microwaves towards the earth (microwaves lie at the higher frequency-end of the radiowaves spectrum) and captures its reflectance which is technically known as backscatter as the energy signals return in the same direction from where they were issued) - and is preferred to Multispectral Imagery (acquired through passive illumination - solar radiation) for detecting Urban Footprint as building exteriors reflect a sizeable portion of microwaves unlike surfaces which typically surround an urban structure (such as water, vegetation or barren land) which reflect only a limited portion of the microwaves towards the satellite, thereby facilitating accurate delineation of urban features. Multispectral Imagery can also be utilized, however, the delineation of infrastructure is comparatively less straightforward. Besides, solar radiation cannot pierce atmospheric constituents such as clouds and aerosols as the wavelengths which make up sunlight (visible light, infrared and ultraviolet rays) are small - microwaves in comparison have much longer wavelengths and can bypass these with ease, thereby rendering it suitable for acquiring imagery at night, on cloudy/rainy days, and even in highly polluted areas. Thus, this consistency in data acquisitions over the same study area makes SAR imagery datasets comparable - an enormously valuable characteristic if one wants to do temporal (two points in time) or multi-temporal (more than two points in time) analysis for Change Detection purposes - do refer the technical explanation on the utility of this high Coherence characteristic of radio waves.
While the process of extracting Urban Footprint from SAR Imagery can be viewed here, let me show you some of my outputs-
The Slider is best viewed on PC
Slider 1: A RGB composite over a cross-section of New Delhi, India derived using the Mean, Difference and Coherence between two Sentinel-1 SAR images acquired in the month of October 2020 (left). For context, the underlying area can be seen in the natural-color Google Earth basemap on the right. The RGB composite of the entire study area can be viewed in Slider-mode here.
Red spots represent those pixels which have very high coherence and very low backscatter - these are typically bare soil and rocky areas which remain unchanged between the two points in time and intrinsically reflect a negligible portion of the microwaves back towards the satellite. The yellow spots represent those pixels which have very high coherence and high backscatter - these are the built-up areas which not only remain unchanged between the two points in time but also reflect a significant portion of energy back towards the satellite due to single and/or double-bounce scattering. The green spots represent those pixels which have low coherence and low backscatter - these are typically vegetated/forested - for example, wind changes the orientation of the leaves which result in low similarity between the two images and intrinsically, vegetation reflects only a small portion of microwaves back towards the satellite due to volume scattering. The blue spots represent those pixels which have very low coherence and very low backscatter - these are typically water bodies which change their structure every passing second or can even be agricultural land which has undergone ploughing between the two points in time.
Slider 2: Delineating the Built-up areas i.e. the Urban Footprint (white pixels) from the other surface features (black pixels) over a different cross-section of New Delhi as depicted in the Google Earth basemap on the right. This was done by masking out just the yellow pixels from the RGB composite of this cross-section.
Slider 3: RGB composite over the entire study area derived using the Mean, Difference and Coherence from two Sentinel-1 SAR images acquired in the month of January 2016 (left) and RGB composite over the same study area derived using the Mean, Difference and Coherence from two Sentinel-1 SAR images acquired in the month of October 2020 (right).
This near-five year composite comparison helps in Change Detection. Can you spot some areas where the Urban landscape has changed?
For example, while moving around Slider 3, notice the cluster of yellow pixels that emerge at the top-centre of the composite from 2020 (north-west to the Ghaziabad label). Upon inspecting this spot using the Historical Imagery feature on Google Earth, this is what I found-
Slider 4: Google Earth Basemap from 2016 (left) and current Google Earth Basemap in 2020 (right)
The cluster of new yellow pixels lie directly over the recently-built WUPPTCL Ataur Power Substation. Radar Remote Sensing is brilliant!
Download the two RGB composite KMZ files here which you can visualize using Google Earth PC software in case you would like to study the changes in New Delhi's landscape yourself.
Malappuram in Kerala, India was ranked as the fastest growing city in the world by The Economist based on Total % change in population forecasted between 2015-2020 (a whopping 44%). Since growth in Population would necessitate growth in Built-up Area, this presented an opportunity for me to extract and compare the change in Urban Footprint over the city around this timeline.
Slider 5: Urban Footprint over Malappuram and adjoining areas in October 2016 (left) and October 2020 (right) respectively. Download the KMZ files here in case you would like to view/compare the RGB Composite and Urban Footprint yourself.
As anticipated, the increase in Urban areas (white pixels) is very evident. Since we know the spatial resolution of the Imagery (each pixel in Sentinel-1 Interferometric Wide Swath mode covers 5 m by 20 m. i.e. an area of 100 sq. m), it is possible to compute the spatial extent of the increase in net Built-up Area over the four-year timeline-
The Urban Footprint image from 2016 has ~17,000 white pixels
whereas the Urban Footprint image from 2020 has ~20,000 white pixels. An increase of 3000 pixels represents an addition of 300,000 sq. m or 0.3 sq. km of Built-up area - equivalent to 21+ Eden Gardens cricket stadiums! While not directly equivalent, it wouldn't be too far-fetched to conclude that the 44% forecasted growth in the population of Malappuram corresponded with an increase of Built-up area by 18%.
Kindly note that Urban Footprint represents the horizontal-extent of Built-up area and not the vertical.
A single ~1 GB-sized Sentinel-1 SAR Imagery dataset contains plenty of information (Amplitude, Phase, Elevation, Orientation etc.) regarding the surface features on Earth as well as the Satellite's own position - the magic lies in manipulating the available data points. Isn't it fascinating that one extract so much useful information just by processing reflected energy values stored in a tiny pixel!
Explore my other workflows on Remote Sensing here.
ABOUT US
Intelloc Mapping Services, Kolkata | Mapmyops.com offers Mapping services that can be integrated with Operations Planning, Design and Audit workflows. These include but are not limited to Drone Services, Subsurface Mapping Services, Location Analytics & App Development, Supply Chain Services, Remote Sensing Services and Wastewater Treatment. The services can be rendered pan-India and will aid your organization to meet its stated objectives pertaining to Operational Excellence, Sustainability and Growth.
Broadly, the firm's area of expertise can be split into two categories - Geographic Mapping and Operations Mapping. The Infographic below highlights our capabilities-
Our Mapping for Operations-themed workflow demonstrations can be accessed from the firm's Website / YouTube Channel and an overview can be obtained from this brochure. Happy to address queries and respond to documented requirements. Custom Demonstration, Training & Trials are facilitated only on a paid-basis. Looking forward to being of service.
Regards,
Credits: EO College, ESA's SNAP