In 1832, a cholera outbreak prompted the first use of geospatial analysis when Charles Piquet drew a map depicting cholera hot spots throughout 48 Paris dis-
tricts. Fast forward nearly 200 years to witness how technological breakthroughs are revolutionizing the geospatial business and giving interactive insights that were not feasible even a decade ago, as various companies track the spread of the COVID-19 pandemic.
We have seen data become enormously more prosperous over the last five years. Archaeology, disaster response, urban planning, infrastructure, logistics, retail, and transportation are just a few of the industries that have embraced advanced geospatial analytics.
The 2nd episode of the SpaceJam series, titled, ‘The Geospatial Way’, focused on the critical aspects of the geospatial industry and its evolution over the years.
Rachana: Krishna, at ISRO, you were working on launch vehicle development, which is one of the most incredible things to accomplish. How and why did you make the switch from machine learning to geospatial data analytics?
After graduating from IIST with a bachelor’s degree in avionics, I spent three fascinating years working at the Indian Space Research Organization. I’ve worked as a project engineer for the GSLV MARK III. After a few years, I decided to pursue higher education, which is how I moved to the United States to pursue my Masters from the University of Notre Dame.
I didn’t start with machine learning but the application of computational optimisation and other mathematical principles in machine learning and data science. When I was looking for employment, all I wanted was a data science position that indeed used data and machine learning rather than just using the words data and AI as a buzzword, which SatSure certainly did. That’s how I came to this organization.
Rachana: Shobitha, you used to be an SAP developer, so how did you make this transition?
Shobitha: My passion for programming led me to work as an SAP developer, but Earth observation was always something I was interested in. So, after a few years of gratifying work in SAP, I began to focus on earth observation, GIS remote sensing. At that time, I pursued a master’s degree in earth observation and GIS from IIRS and ITC, which was like the best of both worlds, combining teaching from both disciplines.
Rachana: Before commencing your entrepreneurial career with Numer8, Devleena , you had a lot of prior experience in data analytics and data intelligence in non-space sectors, but how did the space bug bite you?
Devleena: During my corporate days, I worked on the INMARSAT project, where I dealt with various data types, including satellite data. It’s always been about the kind of problem you’re working on, and dealing with spatial and non-spatial data is the last thing on your mind.
As a result, we are capable of slicing and dicing both types of data to solve the problem. So, we can’t only look at data through a geospatial lens; we also need to look at it through data analysis or data science lens.
Rachana: Krishna, you are working as a data science product manager at SatSure. Can you briefly explain the roles and responsibilities of this position?
Krishna: A product manager has specific roles and responsibilities, such as working with two teams on each project: a business team that knows the clients’ needs and a data science team that works on technical elements. As a product manager, I must be able to grasp both the technical and business aspects of the project and act as a liaison between the two teams to meet product delivery deadlines.
Geospatial Data Applications in Different Sectors
Rachana: Shobitha, you’ve worked with IIRS (Indian Institute of Remote Sensing) and now with the Norwegian institute; can you compare and contrast the geospatial research methodologies used in India and Europe?
Shobitha: I studied at IIRS and now I am pursuing PhD from a Norwegian institute. The significant differences are the focus areas of research using geospatial data, which are linked with localized challenges for each country, the number of open datasets available, and the encouragement to this sector. In nations like India, there are few possibilities to explore geospatial data technology, and companies like SatSure and Numer8 are having trouble getting their products on the market and convincing customers.
Rachana: Devleena, what are your opinions on the potential for downstream geospatial data applications in developing countries?
Devleena: We are still only doing groundwork, but agriculture is the first sector to implement when it comes to spatial data applications in developing nations. Other domains where geospatial data applications can be deployed include transportation, supply chain, government operations, waste management, etc.
For example, during the Covid scenario, several IIT professors worked on a model for providing overnight vaccines without considering the location factor. The justified question here is how to solve supply chain problems using geographic data or adding location to it. Firms such as Numer8 and SatSure enter the picture by integrating their ready-to-use APIs into the models.
Second, this is no longer a niche industry. We must educate people that these are readily available, easily integrated, and cost-effective solutions that can be used for free by leveraging open-source satellite data and applied to the entire socio-economic problems they attempt to solve.
Product Development Challenges in Using Geospatial Data
Rachana: A significant amount of AI and machine learning technology is involved, and certain biases will emerge throughout the preparation of these products, which is referred to as fair AI. So, Krishna, how are you ensuring that impartiality?
Krishna: Fairness in AI has well-known case studies, but satellite data goes beyond all of that and has its own set of biases. For example, models vary depending on geographical location, and when it comes to assessment metrics at the farm level, how to quantify performance characteristics because different areas have altered meteorological and cropping patterns.
We must examine the entire process to eliminate biases, beginning with the data source and ending with ground-level expert knowledge. In India, we can’t utilize the same data in different places; agricultural farm scores alter when you move from a rain-fed tropical farm to an irrigation-fed farm in a desert, like Rajasthan. If we don’t put this effort into data throughout the modelling process, we’ll just be repeating the bias.
Rachana: Shobitha, if we talk about your research on air pollution in Norway, can you share some insights on how it is done across the world, aligning with your research?
Shobitha: Normally, when air pollution is talked about, we assume traffic and industrial smoke, but there are a lot of localized sources of air pollution that are required to control as per WHO guidelines. Being less polluted, Norway still invests in air pollution monitoring because a lot of aerosol pollution is there during winters due to wood burning in every household.
There are other sources for air pollution, and these are not localized; we can see NASA images where the Sahara Desert storm carries dust from one place to another. So, it cannot be localized research; it’s a global effort to monitor the emissions and pollution levels in varying locations. Even after the lockdown in 2020, it was one of the warmest years due to past climate changes.
The per capita carbon footprint is usually higher in developing countries; We use a combination of different datasets for monitoring. Satellite data is one of them. As the altitude increases, there is a variation in pollution level, and we can’t monitor it with satellite data. So, my research is to monitor the pollution levels at different heights, using satellite data and ground monitoring stations in Europe.
Rachana: Let’s talk about business cases here; who will pay to get this kind of data or insights?
Shobitha: Primarily this data is used by Government bodies for policy planning. For example, If IPCC sets some limits on air pollution to monitor and prepare these reports, government bodies require these insights.
Understanding Sustainable Business Models based on Geospatial Data
Rachana: Devleena, Let’s speak about business models: with OFish and your other products serving fishers and retailers, do you see yourself as a B2C unicorn in the future, or do you foresee yourself as a government-centric company?
Devleena: In my opinion, the term Unicorn refers to a company that is performing well but does not represent the bottom line. But, in our situation, I believe the challenge is whether we can develop a viable, lucrative geospatial business aimed at customers rather than the government. Although the government is not yet a customer, engaging with them will provide us access to a vast pool across large geographic areas.
On the other hand, working with the government necessitates a large sum of money and a lot of patience due to the long sales cycles, which is not a viable choice for a startup. Fisherfolks are our primary beneficiaries, and we will have a successful model in the future that focuses on our customers’ needs.
There is usually a top-down model in the fisheries sector with various regulations across geographies, but let’s take the bottom-up approach. There are 4 million fishermen in India and many more across continents. We are familiar with supply chain issues on the ground and can offer solutions. So, you might not be a Unicorn, but you can indeed have a lucrative firm if you understand the problem and have the right metrics.
Rachana: There is a lot of vertical integration that can be done in the geospatial and upstream sectors. There are people who design, manufacture, operate, and launch satellites and create a variety of data products from them. So, do you think businesses like Numer8, who deal with geospatial data, will be able to exist in the future?
Devleena: We use various data sets; therefore, we aren’t strictly a geospatial company, but our solutions would be futile without geospatial data. Again, it all depends on the business you’re working. For example, soil quality may be determined by soil monitoring in a different vertical; similarly, our target area is fisheries, which has been neglected for a long time, allowing us to develop solutions.
When it comes to vertical integration, the European Commission initially provided 4-5 different geospatial datasets. Then AWS and GCP arrived and began supplying a plethora of other geographic datasets based on client demand.
As a result, from a business perspective, you must continuously evolve with the industry. The best part about the geospatial earth observation data Analytics Company is that it does not prevent us from supplementing with various fresh data sets to stay in the company and expand our horizons to other possibilities.
Entrepreneurial Challenges in Earth Observation Industry
Rachana: Many constellations are planned; Thus, there may be a considerable number of satellites orbiting around Earth in the next ten years. So, are the images generated currently sufficient in terms of frequency bands, quality, and location?
Devleena: Timely data availability, in my opinion, is the most severe issue we and other companies in this area confront. Data should be easily accessible, downloadable, and always available. When developing a model, we frequently encounter gaps in historical data availability, so these concerns should be rectified.
For example, suppose a non-GIS person wants to evaluate a given ocean metric, such as SST (Sea Surface Temperature). In that case, he will need to go through several datasets, which can be made more accessible by supplying data for commercial applications.
Rachana: You’ve given the earth observation industry a whole new viewpoint by educating the public and looking for possible clients and investors. So, as a bootstrapped entrepreneur, what challenges have you faced?
Devleena: Our job is highly unusual, and most people have no idea what we’re talking about when we try to correlate satellite data with fisheries. We need to educate them on how satellite data may add value to their lives by presenting the situational perspective while speaking with them. It’s an ever-evolving process, and the main problem for a bootstrapped business is that there are no ready-made models for reference.
Real-time data can be applied to various segments, but we must first comprehend the challenges at hand and have the vision to develop better solutions. In the long term, obstacles will become opportunities in this area; for example, while I work in fisheries, someone else will execute similar ideas in another field.
Future of Geospatial Industry
Rachana: Can geospatial data likely become an essential aspect of policymaking in future, and if yes, to what extent?
Devleena: Currently, no government policies use real-time data, but every socioeconomic problem has a space-time component, and in the long run, it could become a part of several government policies. Considering how the poverty rate has changed throughout time. What are the causes of people’s pain, and how may they be alleviated? As a business, we address food security issues on the high seas by considering the geographical component from an ocean standpoint.
But what about the 800 million people who rely on seafood as a source of income? As a result, the location parameter is an essential consideration in any policy formulation. Even if you want to develop regulations for the mining industry or miners, you need to know where they live and how connected they are to the rest of the world.
Rachana: How do you see the geospatial sector evolving in the next few decades? When it comes to real-time data, there is a lot of emphasis on data privacy, as well as several biases; how do you see all of this evolving in the next few decades?
Krishna: On the policymaking side, there is a lot of room for growth for geospatial data in the future, but we need to be more careful and aware of the parameters that must be in place while encompassing the geospatial domain’s future.
For example, in cyberspace, we have cyber security rules that aren’t up to par, resulting in crimes. Similarly, good guidelines would be required to guide us down this route. Here privacy is another crucial piece of information.
Consider two scenarios: the first is a disaster management policy in which it’s better to have more data available from all sources, allowing us to better aid people in mitigating the situation. On the other hand, we can’t make sensitive information accessible to everyone if we’re trying to figure out where some endangered species are based on weather and vegetation patterns.
Shobitha: The geospatial industry, in my opinion, will be the next IT industry. Geospatial data is extraordinarily vital and tough to obtain, but by incorporating communities, it is already happening that individuals are mapping a wide range of regions.
So, there’s fantastic community-based geospatial data on the way, as well as things like digitizing a property or a farm, which is a continuous process. As a result, having these will offer a great deal of transparency to numerous functions in the future. I believe geospatial data will have a role in the future.
Devleena: There are various ongoing disaster management applications, such as the Red Cross citizen awareness tool, which is available in the Australian community. They discuss the neighboring wildfires, validate using satellite data, and monitor where they should focus their attention for the day. The manner it may bring down to citizen ownership, in my opinion, can offer value to a variety of sectors.
This article was originally published in The SatSure Newsletter (TSNL)