View from India: Using satellite imagery to glean location intelligence
Image credit: Scaliger | Dreamstime.com
Geospatial data and satellite images bring us close to real-time feeds and have diverse applications.
People with problems and people with expertise in geospatial technology could come together to address societal issues. The Government of India (GoI) has paved the way for this with an announcement made earlier this year. GoI has announced liberalised guidelines for geo-spatial data. Drafted by the Department of Science and Technology (DST) the guidelines bring sweeping changes to India’s mapping policy, specifically for Indian companies. It means that granular geospatial data and services, many of which were in the restricted zone, will be made available publicly.
As the government-led 'Smart City Mission' rolls out, satellite imagery could throw light on urban analytics. This includes buildings, farmlands, regional boundaries and land use planning. In all likelihood, these parameters could shed information on accessibility and live-ability. “Satellite imagery could give inputs on location intelligence, which includes the development of land over time. Apart from that, property tax, election campaigns, forest fires and diseases are other aspects that can be concurred from satellite imagery,” said Anand S, CEO, Gramener, speaking at the Nasscom 'XperienceAI Virtual Summit 2021'.
Satellite images bring us close to real-time feeds. The data is humongous and the idea is to put it to use to study economies. For instance, Stanford University has predicted poverty levels using night-time satellite imagery, as well as per capita economic expenditure across continents. Dark and light images denote poverty levels.
This is one example. What could be gleaned from this example is whether a similar exercise can be carried out in India. It also needs to be seen if satellite imagery can help determine the wealth levels of individuals. A beginning has been made. “Satellite imagery has revealed that residents belonging to certain parts of south and north India were financially better off than their counterparts in Bihar and north-east India. The house size and type of rooftops was a determining factor. Rooftops made of clay, concrete, tile and steel threw light on the social status of households,” added Anand.
Mapping technologies and geospatial services are among the upcoming opportunities for startups, entrepreneurs research institutes and companies. Scalable innovation could hopefully transform industries. Prof Ashutosh Sharma, secretary, Department of Science and Technology (DST), Government of India, has indicated to the press that Atmanirbharta (self-reliance) in geospatial products and solutions can generate business worth around Rs 1 lakh crore by 2030. It is projected to have an economic impact on Digital India, smart cities, e-commerce and autonomous drones.
As the industry matures, biodiversity, rivers and mountains can be mapped. Blue economy is another sunrise opportunity where geospatial data can be tapped for fisheries, deep sea mining and offshore oil and gas production.
There are other applications as well. Geospatial tech tools can be leveraged for culling out information derived from the geospatial data and satellite imagery. Political elections are one such application. Geospatial data may throw light on the population of the constituency, its people, their age, occupation and other demographics.
Geospatial brings to mind images of geographic spaces, their layout and sometimes throws light on the manner in which geographic spaces have spread. With geospatial, comes intelligence, in the form of artificial intelligence (AI) tools which convert the imagery into data.
Geospatial AI is a subset of data science. It goes beyond looking at things that happen to understand why they happen. Geospatial AI could be tapped to save lives through footprint data and population levels. It may also serve some unexpected purposes. A case in point is agriculture. For instance, crop types could be mapped for extracting its various features. Crop monitoring could shed light on damage to crops. Geospatial AI may be leveraged for precision farming and production control. It could also aid other dimensions of agriculture, such as crop insurance and management.
Geospatial data or geodata is available in formats such as raster or image and vector or line drawing. Tech tools enable the map shapes to function as numeric, which means one can add or subtract details on the maps. Raster data is usually converted into vector data as it is relatively easier to process. Geospatial data, especially vector data, can be stored in databases such as PostGIS, Oracle Spatial, MongoDB and Neo4j Spatial, among others. As enablers, these geospatial tools and platforms help visualise, analyse and interpret data, besides helping edit the data. Multilayer overlays, business intelligence and processing capabilities offer further advantages and there are also many Python programming libraries that help process geospatial data.
With this kind of a backgrounder, what comes to mind is how geospatial data can be used for tracking diseases and its spread. For instance, mosquitoes could be among the world’s most dangerous animals. “Global statistics indicate that every four minutes a case of chikungunya [severe fever caused by a mosquito bite] is confirmed somewhere in the world. Annually 400 million dengue infections happen,” reasoned Anand. World Mosquito Program (WMP), a not-for-profit initiative, decided to tackle this. With an aim to protect the global community from mosquito-borne diseases, WMP partnered with Gramener under a 'Microsoft AI for Good' grant.
Design-led data science company Gramener has applied computer vision models on high-resolution satellite images to determine population densities and vulnerability to mosquito-borne diseases at a sub-neighbourhood level. Based on the city, population and coverage expected, the AI solution works out a fine-grained release and monitoring plan. With this, the WMP team has acted quickly to maximise the impact of their solution. The model is over 70 per cent more accurate and saves over 99 per cent of manual effort. Gramener’s AI solution is built to scale and has shown the ability to be adapted to various locations around the world.
WMP was first executed in Kampala, the capital city of Uganda in Africa. Research revealed that wolbachia (a genus of intracellular bacteria) are a safe and natural bacteria found in several insects including mosquitoes. However, wolbachia was not present in the mosquitoes that transmit dengue and chikungunya. These mosquitoes were then infected with wolbachia so that they don’t transmit the mosquito-borne disease.
The solution helped cut down the time taken to identify and monitor release points from three weeks to a few hours. It has also enabled an accurate forecast release plan, while rationalising costs during the release planning process and cutting down the need for costly iterations by the field teams. WMP has protected five million people across 12 countries from diseases such as dengue and malaria. The goal is to protect 100 million people in the next five years, thereby accelerating the social impact for the community.
WMP is one example. Could we tweak the tech approach to examine the spread of Covid in the country, going region by region? Could the approach help in studying the spread of the virus and send out timely alerts about the same? Alternatively, could geospatial data and tech tools be deployed to identify clusters where people have not been vaccinated? Perhaps therein lies a possibility.
Sign up to the E&T News e-mail to get great stories like this delivered to your inbox every day.