Research

Hydroclim group is addressing problems from the broad domains of hydroclimatology and biometeorology. The group’s research centres on streamflow prediction at multiple spatiotemporal scales, the impact of weather on human well-being, unprecedented rainfall events, and climate change projections. We develop novel, data-driven statistical approaches and employ physical/conceptual models to predict hydroclimate variables. The goal of our research is to understand the relationship between Climate, Hydrology, and Biometeorology. A few current research interests are as follows.

Climate change projections

Our climate is changing as a result of human activities. General Circulation Models (GCM) are essential to understand the likely changes in climate variables in near future. GCMs project climate variables under different future scenarios. To estimate the future changes in hydrologic variables, climate change projections are forced into hydrologic models. However, raw GCM projections are not suitable for hydrologic analysis as raw GCM projections may have structural bias, scenario uncertainties, and a different spatiotemporal resolution as compared to the hydrologic models.

Therefore, post-processing approaches need to be applied to raw GCM projections. Hydroclim group is developing univariate to multivariate statistical methods for post-processing that can preserve inter-variable dependencies in the post-processed outputs. One of our climate change studies has been funded by the INSPIRE project, Department of Science and Technology, Government of India.

Hydroclimate uncertainties

Uncertainties in hydroclimate variables may arise from different sources: model structures, parameters, and inputs. For example, one potential source of uncertainty in measured streamflow comes from an inaccurate representation of the stage-discharge relationship. Hydrologists develop a mathematical relationship between the river discharge and river stage at a particular river gauge site known as the rating curve. Typically, subsequent measurements of river discharge are indirectly estimated by providing the gauging information in the pre-calibrated rating curve. Large uncertainty in rating curve parameters and inaccurate rating relationship induces uncertainty in the estimated streamflow. Additionally, seasonal vegetation growth, sedimentation, and faulty instruments can also be sources of streamflow uncertainty. Under a changing climate, rating curve uncertainty is reported to be as large as 40% of the measured streamflow. However, estimation of measurement error is often ignored in traditional hydrologic modeling considering the time and the money involved in the estimation process.

Our group is engaged in developing a framework to estimate the measurement uncertainty at stream-gauge stations that are minimally influenced by anthropogenic climate change. Once estimated, measurement uncertainty is included in a Bayesian framework to issue real-time streamflow forecasts. Our objective is to explore the tradeoff between the effort involved in developing a robust rating curve and the resulting benefit achieved from including it in streamflow forecasts.

Urban rainfall

Urban areas have been experiencing frequent heavy flooding resulting from extreme rainfall events attributed to climate change. In the recent past, several metro cities in India have witnessed high-intensity rainfall and subsequent floods. Urban floods are considered a great disruption to human life and property. Reliable urban flood prediction critically depends on the rainfall measurements, since it can impart significant uncertainty to the prediction.

Therefore, the development and maintenance of an urban rain-gauge network is essential. A recent study from our group (in collaboration with Prof PP Mujumdar) has shown that significant spatiotemporal variation in extreme rainfall distribution exists within Bangalore (India) city (Joseph et al., 2022, under-review). However, the density of urban rain gauges is typically a few which can not capture the spatial variation in urban rainfall. Our group is investigating the potential of crowd-sourced rainfall estimates for Bangalore.