Redefining Risk Management Through Geotechnical Expertise and Technological Innovation: Dr. Vaka on Emerging Technologies, AI & Engineering Judgement

As Head of Innovation and Technology at Geofem, Dr. Divya Sekhar Vaka shares his insights on the emerging technologies transforming risk management, the role of AI and digital transformation, and how innovation is shaping the future of earth observation and satellite remote sensing. He also explains why, despite rapid technological advances, engineering judgement remains at the heart of effective decision-making.

What emerging technologies in the world of satellite remote sensing are you most excited about right now? How do you see them shaping the near future of engineering, data analysis and risk understanding?

For the first time, we are moving from a world where data was scarce to one where the challenge is understanding and integrating enormous volumes of information.

We now have access to radar satellites, optical satellites, LiDAR, GNSS measurements, weather data and many other sources. The challenge is no longer data availability; it is how we integrate all these datasets to extract meaningful information.

I am also excited about the upcoming NISAR mission and the opportunities it will create for vegetation monitoring, infrastructure monitoring, natural hazard assessment, and environmental applications.

NISAR's unique combination of L-band and S-band radar observations will provide greater sensitivity to vegetation structure, soil moisture variations, and ground deformation processes than many existing missions.

For infrastructure and geohazard monitoring, the longer L-band wavelength is particularly valuable because it can maintain coherence over vegetated and rapidly changing terrains where traditional C-band systems often struggle.

This will improve our ability to monitor slopes, landslides, subsidence, and critical infrastructure across larger areas and with greater reliability. For environmental applications, the mission will provide new insights into biomass changes, flooding, and land surface processes, supporting both scientific research and operational decision-making.

Combined with advances in cloud computing and AI, I think we are moving towards a future where satellite data becomes a routine part of engineering decision-making.

How have AI and automation changed the way you and the Geofem team approach innovation? What challenges can you solve today that would have been difficult or impossible to address just a few years ago?

AI and automation have significantly changed the way we work, especially when dealing with large volumes of geospatial and Earth Observation data. A few years ago, many workflows required substantial manual effort, from data preparation to analysis and interpretation. Today, much of that can be automated, allowing us to focus more on understanding the results and solving real-world problems.

One area where I see a major difference is in identifying patterns across large datasets. For example, when monitoring infrastructure networks that extend hundreds or thousands of kilometres, AI can help identify subtle relationships and areas of concern that would be difficult to detect manually. It is not replacing expert judgement, but it is allowing us to analyse data at scales that were simply not feasible before.

Identifying clusters with varying deformation mechanisms along a railway line.

Can you share an example of an idea at Geofem that started small but evolved into something with significant real-world impact for clients, infrastructure, or communities, and why it worked?

A good example is our work on landslide and slope stability monitoring using satellite-based InSAR technology.

The initial idea started with a relatively straightforward question: can we improve landslide susceptibility mapping by combining InSAR measurements with machine learning techniques?

One of the early outcomes of the GAIA project was the development of a new approach to landslide susceptibility mapping that integrated InSAR-derived deformation measurements with geological, topographic, and environmental datasets.

The key innovation was the inclusion of observed slope movement, enabling us to better distinguish between potentially unstable and actively deforming areas.

The methodology was later published in Remote Sensing and became the basis for our subsequent work on dynamic landslide hazard assessment and Geofem’s published journal paper: InSAR Integrated Machine Learning Approach for Landslide Susceptibility Mapping in California.  

As we continued developing the approach, we realised that susceptibility alone is not enough for decision-makers. They also need to understand how hazard levels change over time, particularly during periods of intense rainfall.

This led us to develop dynamic landslide hazard mapping methods that combine satellite-derived deformation measurements with rainfall information and other environmental factors to assess how slope conditions evolve. Recently, “Dynamic Daily Rainfall-Driven Temporal Landslide Hazard Mapping Utilising InSAR and Machine Learning” was presented at the LARGE conference in New Zealand.

What I find most rewarding is that the work has not stopped there. We are continuing to refine the algorithms and explore ways to better capture the relationship between rainfall, ground deformation, and slope failure processes. The goal is to move towards more proactive risk assessment and early warning capabilities that can help infrastructure owners, local authorities and communities better manage landslide risk.

I believe the idea gained traction because it addressed a real challenge. Landslides are complex and highly dynamic, and no single dataset provides all the answers. By combining Earth Observation, machine learning, and environmental data, we were able to develop a more practical and scalable approach to understanding slope stability over large areas.

Engineering teams need concrete, reliable outputs — while innovation always requires experimentation. How do you balance these two priorities in a high-stakes technical environment?

In engineering environments, reliability and trust are essential because decisions often have safety, operational, and financial implications.

New ideas are first tested through research, prototypes, and pilot studies. Once we understand their strengths and limitations, we gradually introduce them into operational workflows with proper validation and quality control.

Having a strong scientific foundation helps. If you understand the physics, the assumptions and the limitations behind a method, you can innovate confidently while still delivering outputs that clients can trust.

If you could solve one major industry or environmental challenge through technology immediately, what would it be and why?

I would choose the ability to reliably forecast infrastructure failures and natural hazards before they occur. Today, we are very good at monitoring and detecting changes. We can measure ground deformation, identify areas of instability, and observe environmental changes with increasing accuracy. However, forecasting exactly when and how a problem will develop remains much more difficult.

If we could possibly combine Earth Observation, environmental modelling, AI and, even, continuous monitoring to provide reliable early warning of failures, it would have enormous benefits. It could help protect communities, improve infrastructure resilience, reduce maintenance costs, and ultimately save lives.

Geofem places strong emphasis on the geotechnical interpretation of data. In what ways are evolving technologies and new analytical tools shaping and improving that engineering interpretation?

Advances in Earth Observation and AI are helping us identify patterns and relationships that would have been difficult to detect a few years ago. For example, we can now analyse long-term ground deformation trends across entire infrastructure networks, assess how slopes respond to rainfall events, or identify areas where changing environmental conditions may be increasing geotechnical risk.

The real value, however, comes from combining advanced analytical tools with geotechnical expertise. A deformation measurement on its own is just a number; understanding whether it represents seasonal movement, settlement, slope instability, or a potential failure mechanism requires engineering knowledge and experience.

The new technologies allow us to ask better questions, analyse larger datasets, and make more informed decisions, but engineering judgement remains at the centre of the process.

About Dr. Divya Sekhar Vaka

Dr. Divya Sekhar Vaka is Head of Innovation & Technology at Geofem, with ten years of experience in InSAR for infrastructure and environmental monitoring. He holds a PhD in SAR Interferometry from the Indian Institute of Technology Bombay.

His expertise lies in using InSAR to monitor and analyse urban subsidence, slope stability (including landslides, mines, and tailings dams), subsidence due to oil/natural gas extraction, groundwater exploitation, and earthquake deformations, with a specific focus on land and geohazards. He is a member of IEEE, IEEE-GRSS, ISPRS, and ISRS, and has participated in, published, and reviewed technical papers at reputable conferences and in journals.

Dr. Sekhar has contributed to numerous projects funded by the European Space Agency and other European funding bodies, while also successfully delivering a wide range of commercial initiatives.

His interests and skills encompass a wide range of areas, including machine learning, deep learning, geophysical modelling, GPS-GNSS, SAR-optical data fusion, change detection, sea level rise-vertical land motion, and geospatial analysis.

Speak to the Geofem team today to learn more about how we can deliver deeper insights into your assets through advanced monitoring technology and engineering expertise.