Introducing New Postdoc: Meet Tejasvi Ashish Chauhan

Tejasvi recently joined our group as a Postdoc in October last year. He got a Ph.D. in civil engineering from the Indian Institute of Technology (IIT) Bombay, Mumbai, India, working on land surface-atmosphere dynamics and information theory. Welcome, Tejasvi!

Research Background

During my Ph.D. I worked on analyzing hydro-meteorological processes by representing them in the form of causal networks using information theory. Here is a summary of my work:

To find the strength of association between any two variables, e.g. precipitation and temperature, conventionally, the most commonly used metric is correlation, which tells us about the variability explained by first order (linear) estimates. But are all processes in the Earth system linear? No. Especially not at finer time scales. In those cases, correlation would turn out to be very low even if there is an underlying strong (non-linear) association. So how do you get a full picture of association between any two variables that considers non-linearity?

As it turns out, the information theory has a solution! The concept goes back to 1948 with seminal works of Claude Shannon — known as the father of information theory — who proposed a quantity called the Shannon’s entropy, which measures the variability or uncertainty of a distribution (we use the probability density function of a variable). The strength of association between any two time series can then be found by finding overlaps of their entropies from their distributions — called Mutual Information. In addition to only being linear, correlation can also generate misleading results if two variables are independent but driven by a common third variable. How do we avoid the effects of common drivers (also known as confounders)? The solution is to use conditional probability density functions (conditioned on the common driver) to generate a quantity called ‘Conditional Mutual Information’. Using these metrics, one can find directional associations between different components of earth system which can be represented in the form of a network.

You can read more about my work from following papers:

1. Chauhan, T., & Ghosh, S. (2020). Partitioning of memory and real-time connections between variables in Himalayan ecohydrological process networks. Journal of Hydrology, 590, 125434.
2. Chauhan, T., Devanand, A., Roxy, M. K., Ashok, K., & Ghosh, S. (2023). River interlinking alters land-atmosphere feedback and changes the Indian summer monsoon. Nature Communications, 14(1), 5928.
3. Chauhan, T., Chandel, V., & Ghosh, S. (2024). Global land drought hubs confounded by teleconnection hotspots in equatorial oceans. npj Climate and Atmospheric Science, 7(1), 15.
4. Chauhan, T., Bhadury, P., Rodda, S., Thumaty, K., Jha, C., & Ghosh, S. (2023). Resilience of South Asian mangroves to weather extremes and anthropogenic water pollution.

Research Interests and Plan

I am passionate about improving our understanding of Earth system processes using physically based approaches like thermodynamics, or data driven approaches like information theory, complex networks, etc.

As a part of this group, I plan to work on estimating the sensitivity of different components of the hydrological cycle to global warming using thermodynamic principles. As the planet warms up, the moisture holding capacity of air increases by 7% per degree rise in temperature. Since the evaporation doesn’t increase proportionally, and out of the increased evaporation, only a fraction comes to land, it makes land regions more arid in a warmer world. Still, we don’t observe all land regions getting dry at the same rate. What generates this spatial variability? Since, the ocean and land regions are warming at different rates, how does that affect the precipitation that we receive on land regions?

Interestingly, the Shannon’s entropy from information theory is pretty similar to Boltzmann’s entropy from thermodynamics! But do they have anything to do with each other in the Earth system? We don’t know yet. These are some of the questions that excite me and I look forward to exploring them as a part of the Biosphere Theory and Modeling group at the Max Planck Institute for Biogeochemistry.

Contact

If you want to know more about my work or want to discuss any idea that you are working on, I am always happy to discuss. Please feel free to write to me at tchauhan [AT] bgc-jena.mpg.de. You can also follow me on X (@tejasvichauhan) or LinkedIn.