Land behaves quite differently from oceans in the way it stores and exchanges heat and water. Unlike oceans, it dries out and is limited by water availability. These aspects of land are intricately linked to near surface atmospheric variables like 2m air-temperatures and humidity, making land–atmosphere exchange a key control on regional climates. 

Key questions that we work on are

Q. How can we improve the modelling of land-atmosphere exchange? 

Q. What role do they play in shaping variability in temperatures as well as extremes like heatwaves, droughts and floods? 

Q. Are there physical limits that govern these exchanges?

My work has shown that the heat and moisture exchange between land and atmosphere are subject to strong thermodynamic constraints (Ghausi et al., 2023, PNAS). These constraints primarily arise from how much work atmosphere can perform to generate motion. Building on this, I have developed energetically-constrained models of the atmospheric boundary layer that reproduce diurnal air temperatures and their variations across diverse climates (Ghausi et al., 2025, GRL).

Extreme temperature events can have severe impacts on terrestrial ecosystems. However, heatwaves can arise from diverse physical processes and can have distinct, sometimes contrasting, impacts and trends under warming. Understanding the mechanisms driving these events is essential for improving short-term forecasts, long-term projections, and assessments of their ecological impacts.

To understand this, we work on a range of statistical and analytical frameworks that uses the thermodynamic energy equations to understand temperature anomalies. We have already tested these approaches at regional and global scales. For instance, we showed that regional changes in solar and longwave radiation alone can drive both hot and cold extremes  (Tian et al., 2023, ERL). Extending this framework globally, we showed that global warming increases the proportion of sunny-dry and advective heatwaves, with important implications for terrestrial ecosystems (Tian et al., 2024, Nature Communications Earth & Environment).

We also work on developing analytical models that links radiative and surface evaporative conditions to changes in daily maximum temperatures and requiring no free parameters. These models use energetic constraints like surface-energy balance, heat storage variations in lower atmosphere and thermodynamic constraints on turbulent fluxes and provides strong physical frameworks to identify different drivers of temperature variability and extremes.

Our future work is centered around following questions:

1. How will the characteristics, drivers, and impacts of temperature extremes change under global warming?

2. How can we improve the physical understanding and prediction of temperature extremes across diverse climates?

Global warming intensifies the hydrological cycle by altering atmospheric moisture, evaporation, and precipitation. There exists strong physical basis to suggest that rainfall extremes should intensify with warming as the warmer atmosphere holds more moisture, yet the response of extreme rainfall also depends on atmospheric circulation, moist convection, and land–atmosphere interactions. These processes can produce a wide range of rainfall responses, including intensification beyond the expected Clausius–Clapeyron (CC) rate and, in some regions, surprisingly weak or negative relationships between rainfall extremes and temperature. Understanding these mechanisms is essential for reconciling observations with climate model projections and assessing future flood risks.

My research has focused on addressing this gap across regional to global scales. .Over South Asia, I showed that rainfall–temperature scaling becomes negative at high temperatures, in contrast to the positive streamflow–temperature scaling (Ghausi et al., 2020, GRL). I subsequently demonstrated that cloud-radiative effects help explain this apparent contradiction by influencing surface temperatures and their relationship with rainfall. Incorporating these effects into a thermodynamically constrained surface-energy-balance model revealed that extreme rainfall intensifies in line with climate model projections over India (Ghausi et al., 2022, HESS). Extending this framework globally, I showed that cloud radiative effects explain much of the negative scaling observed in the tropics and help reconcile the long-standing mismatch between observations and models (Ghausi et al., 2024, Nature Communications).

Our future research is now centered on the following questions: 

1. How global warming alters the dynamics of moist convection and vertical motion? 

2. How dynamic and thermodynamic drivers will interact to shape future rainfall extremes? 

3. How land–atmosphere feedbacks influence convective development and rainfall extremes?

Physics-based models provide mechanistic understanding of climate processes, are generalizable across regions but often rely on semi-empirical parameterizations, tunable parameters and miss local details. Recent advances in AI methods have shown emerging success in capturing climate patterns and predicting short-term weather but they lack interpretability and can often not be generalized to unseen data/region. Combining these two approaches offers a way to improve climate predictions while retaining the physical interpretation and understanding.

In our group, we are focused on developing hybrid approaches that embed machine learning within process-based models.  Our goal is to scale these hybrid frameworks to improve process representation, guide model improvements and provide hyper-local forecasts and early warning systems relevant for actionable climate information.

Our future research is now centered on the following questions: 

1. How can physical principles and machine learning be integrated to improve climate process representation and predictive skill?

2. How can hybrid models generalize across regions, climate regimes, and changing environmental conditions?

3. How can these frameworks enable localized climate predictions and reliable early-warning systems for extreme events?