• Towards A Global Flood & Flash Flood Early Warning Early Action System Driven by NASA Earth Observations and Hydrologic Models
Program GEO-Global Flood Risk Mod
PI / Institution Andrew Kruczkiewicz / Columbia University
Start Date February 20, 2018
End Date February 20, 2021
  • Summary

    We propose to enhance disaster manager capacity to better prepare, respond and recover to floods. In the current state, disaster management organizations in developing countries are not preparing sufficiently for flash floods. The lack of available, accessible and usable information, combined with a lack of capacity to take action, has led to a state of insufficient preparedness. In an ideal state, flash flood decision making will be user driven and include the following components; user identification of actions that can be taken on various lead times, co-development of a flash flood Early Warning Early Action (EWEA) system on relevant lead times, setting of forecast thresholds which trigger a predefined set of ‘standard operating procedures’.  A diverse set of stakeholders such as forecast developers, disaster managers, government officials and boundary organizations will be invited to join a Technical Working Group (TWG) for EWEA on flash floods. The primary task of the TWG is to systematically identify a set of flash flood preparedness actions that can be taken if a suitable flash flood forecast existed. The set of actions will be incorporated into action-forecast matrix. EOS (MODIS, EO-1, Landsat,
    SAR/sentinel, TRMM/GPM) will be used to model flash flood risk and monitor flash flood events, leading to the development of a flash flood geohazard map which can be a useful tool in understanding shifts in risk, including on seasonal to subseasonal, seasonal, interannual and decadal timescales. While NASA EOS are integrated into all aspects of the proposal, they are the principal feature of the ‘User driven flash flood forecasting research’ work package. This section will aim for global forecasts for flash floods, which cover the lead times of interest to disaster managers. To achieve this, an intercomparison of existing flash flood forecasts will be conducted using a novel observational datasets and user defined evaluation metrics and properties. Second, the identified global hydrologic forecasts, from systems based on EF5 and LISFLOOD, will be compared and evaluated using a disaster manager’s definition of forecast utility. Next, the models based on EF5 will be refined, generating a global hydrologic forecasting ensemble, decreasing uncertainty. Finally, EF5 models will be rerun for the historical period of data with the resulting analyses being made available within a data commons architecture. This will enable broad open access of the data to the scientific community for further research on flooding. We will formally reconvene the TWG to present the results of the intercomparison for the tailored case studies to our end-user community, working with them to explore suitability for forecast based action and potentially develop SOPs. It is important to note that the goal of the project is not to develop SOPs, but rather, provide the tools and opportunity for stakeholders to co-develop SOPs for a NASA EOS driven flash flood EWEA if desired. We will prioritize interactive and iterative monitoring evaluation throughout the project, which will foster better communication between remote sensing experts and end-users. An assessment of forecast value, such as a cost benefit analysis, of a potential EWEA system is planned, which will include an analysis of averted losses and provide a lens into a system where the case of increasing averted loss ratio driven by an increase in lead time could drive the funding to increase forecast suitability to longer lead times. We feel this approach represents a paradigm shift in disaster management expanding on principals from operational EWEA frameworks such as the Red Cross Forecast-based Financing and World Food Programme’s FooDSECuRE. This would include moving from the ubiquitous ‘opt-in’ approach where reaching a threshold in a forecast could lead to action (but likely nothing will happen) to a system whereby standard operating procedures are agreed upon in advance, taking action is mandatory.