Long-term hydrological budget over urban areas: approaches and challenges

Urbanisation substantially alters the permeability of the land surface and modifies the vegetation cover extent and type. These landcover changes profoundly affect the hydrological and energy budget. While the role of urbanization on the local flood response has been extensively studied, its effect on the long-term hydrological budget is much less known as computing the latent heat flux (evapotranspiration) in urban areas is still challenging. However, knowledge on the urban hydrological budget is important as many cities are planning water sensitive urban designs and water harvesting could be key to support an increasing amount of urban vegetation.

This study quantifies how urbanization changed the long-term hydrological budget of the island city state Singapore for the period between 1982 to 2021. To do so, we use two state-of-the-art mechanistic models which include all the major hydrological components such as runoff, soil and interception water storage, transpiration, and evaporation. The first is Tethys-Chloris (T&C), which is a mechanistic ecohydrological model that can resolve the water, carbon and energy budgets at high spatio-temporal resolutions but considers the urban effects in a simplistic way by only modifying the impervious fraction of the land surface and its roughness. The second is its urban counterpart, Urban Tethys-Chloris (UT&C), which explicitly resolves shading and radiation reflection within an urban canyon accounting for different urban vegetation types and configurations and explicitly resolves the local urban climate and hydrology. UT&C is however too computationally costly to simulate an entire city at high spatial resolution. Hence, a machine learning approach, where a multimodal neural network comprising a Conv1D layer followed by LSTM layers for dynamic meteorological inputs and Dense layers for static inputs such as urban properties, is used to re-map UT&C outputs over the entire city based on a few selected covariates.

By cross comparing the different approaches to compute the long-term water budget over urban areas, we highlight the involved uncertainties, and we can gain insights into the impact of urban development on modifying the water availability in Singapore. The island city-state relies on water harvesting as one source of water supply for households and knowledge of water availability is extremely important for long-term water management purposes.