Personal information
Verified email domains
Biography
I am one of the active developers of the mesoscale hydrological model (mHM) and the lead developer of reservoir representation in mHM.
Since earning a Master's degree in 2013 (Water Resources Engineering), I have gained over a decade of valuable experience in the development and application of hydrological models. Currently, I hold the position of a PhD researcher in the lab of Luis Samaniego at the Helmholtz Centre for Environmental Research (UFZ).
My PhD aims to contribute towards locally relevant flood forecasting in managed river basins, at global scale. The first chapter of my PhD is an application paper (Najafi et al. 2024, Nature Communications) where we developed and tested a high-resolution (1 km) flood early warning system, FEWS, with mHM, on a real flood event (The 2021 summer flood in Germany), retrospectively. This proof of concept produced impact forecast as well as early notice time till 100 years return period water level at each model grid.
Integrating such local level FEWS in regional/continental domains in global scale poses challenges. Gridded hydrological models, such as mHM, incur simulation errors at local level using the existing (classic or state-of-the-art) stream network upscaling methods. While hyperresolution modeling theoretically addresses this issue, its high computational cost bars its use in large-scale modeling, prompting the search for alternatives. In the second chapter of my PhD (under review), we augment global hydrological models with the missing "eagle vision". We achieve this by developing a new stream network upscaling technique called subgrid catchment conservation. SCC offers three distinct advantages: 1) generates locally relevant streamflow i.e., ensures consistency of streamflow performance across catchment sizes (1 km2 to 4,680,000 km2), 2) improves consistency of streamflow across model resolution, and 3) resolves multiple gauges within a grid.
Reservoirs stand as the bastions of humanity's defence against floods. Preparing FEWS for managed basins in large-scale modeling was another challenge we tackled in the final chapter of my PhD. We developed a new reservoir module for mHM (Shrestha et al. 2024, WRR), an improvement over the state-of-the-art representation in large-scale modeling. The research delves into three key aspects: 1) employing machine learning methods to reverse estimate non-consumptive demands (e.g., hydropower), 2) sensitivity of simulations to reservoir bathymetry, and 3) possible thresholds to identify and exclude non-disruptive reservoirs from the model simulation.