15 March 2022
The SEADRIF company (https://seadrif.org/) provides parametric insurance of flood damage to the governments of Laos and Myanmar. Flood events and their consequences are identified through model simulations, not through in situ assessment of damage. These model simulations are done in near real-time and originate from various Deltares-hosted forecasting and simulation systems. The SEADRIF Simulation Engine is one of these systems. It provides estimates of various hydrological parameters including river runoff discharge.
down-scaled model estimate of 'snow water equivalent'
The SEADRIF Simulation Engine is hosted in the Deltares Private Cloud environment. The end users do not interact with the system itself. Rather, its outputs are operationally shared with the SEADRIF platform (http://seadrif.cimafoundation.org/) which is hosted by CIMA (www.cimafoundation.org). Within the platform, the model simulations are combined with various other sources of information (including, for example, satellite based flood estimates). The resulting estimates are shared with the SEADRIF company.
The SEADRIF Simulation Engine comprises various models that cover the Mekong and Ayeyarwady rivers in South East Asia. The river basins cover various types of terrain and climate ranging from the Southers rim of the Hymalayas to the Mekong delta. This means that the models must be able to adequately model various hydrological processes including, for example, the build-up and melting of snow packs and the relatively high levels of actual evapotranspiration in the much warmer climate downstream.
screenshot of the SEADRIF application
The models contained in the SEADRIF Simulation Engine were developed by Deltares in a World Bank financed project. These were subsequently included in a Delft-FEWS application for the purposes of routinely providing model simulation results. This application is hosted by and maintained by Deltares.
The SEADRIF Simulation Engine contains various rainfall-to-runoff and streamflow propagation models. These models are all wflow based. The wflow_sbm rainfall-to-runoff models are forced by online reanalyses based on ECMWF NWP forecasts. From every new weather forecast, the first 24-hours are ‘set aside’ and blended together. This creates a continuous timeseries referred to as ‘online reanalysis’. While imperfect, the data is available in near real-time which allows any insurance fee to be paid out very quickly. These quick payouts are a tried-and-tested means to reduce intangible flood damage as much as possible.
Precipitation, temperature, snow water equivalent inputs and streamflow estimates as model output
An important feature of the SEADRIF Simulation Engine is that the resulting timeseries should be as homogeneous as possible. This means that models and routines are purposely NOT updated and improved. For reinsurance purposes, a good understanding of model uncertainties is deemed more important than reducing uncertainties at the cost of having a lesser understanding thereof.
At the time of writing, the SEADRIF Simulation Engine has just entered its second year of operations. We expect the system to, at least, run for a total of three years, after which the insurance scheme will be re-evaluated.
Real-time hydrological forecasting expert