Using GloFAS for Flood Monitoring: Website Overview

Flood Module
Using this document, you will get a general understanding of the many functions and datasets of the GloFAS map-viewer. Some especially useful aspects will be explained more in-depth in further documents. Additionally, GloFAS offers more advanced video tutorials and webinars, showcasing some aspects of the portal in greater detail. After learning the basics, the video tutorials are a good place for further information.
GloFAS (Global Flood Awareness System) is an online platform provided by Copernicus. According to Copernicus, the platform’s aim is “to support preparatory measures as well as an emergency response to predicted and ongoing major flood events at a global scale“. GloFAS includes a wide range of datasets that can be accessed on the fly using the map viewer and even downloaded. Use cases not only include ongoing floods but also forecasts using predictive datasets e.g., regarding rainfall. This tutorial is going to explain the basics of the map viewer.To access the GloFAS map viewer, enter https://globalfloods.eu, scroll down a bit and click the map viewer icon (Figure 1). You will be asked to register a free-of-charge account to make use of it.

Once you enter the page, get familiar with the interface. You can see the map view in the center, a layers-panel on the left side, and thematic sections at the top as well as further options on the right-hand side (Figure 2).

Within the top menus, the following options are available:

Initial Conditions

The initial conditions present data that are relevant as base data for flooding. The datasets included under this theme include:

Initial 3 day precipitation and anomaly: This displays the total precipitation (mm) brought on by meteorological forcing inputs 3 days prior to the predicted run date. The initial 3 day precipitation anomaly (unitless) is the total precipitation irregularity resulting from meteorological forcing input over the three days prior to the forecast run.
Initial 3 day Snow melt and anomaly: This shows the accumulated snow melt (mm) resulting from meteorological forcings inputs that occurred 3 days before the date of forecast run. The initial 3 day snow melt anomaly (unitless) is the accumulated snow melt irregularity resulting from meteorological forcing input over the three days prior to the forecast run.
Initial Soil moisture and anomaly: Initial Soil moisture (%), is the ratio between the soil water content in the top layer (currently with 7 cm depth) compared with the maximum water content that this soil layer can hold from meteorological forcing input at initial time (00 UTC) on the day of the forecast run. The soil moisture anomaly is expressed as a ratio between the soil water content in the top layer (currently with 7 cm depth) compared with the maximum water content that this soil layer can hold.
Initial Snow cover and anomaly: The initial snow cover is the mean snow cover (mm of water equivalent) from meteorological forcing input for the 24-hour period (00-00 UTC) before the date of the forecast run. Areas where the snow cover is 10 m or above are considered as glaciers. The initial snow cover anomaly is the daily mean snow cover irregularity (unitless) resulting from meteorological forcing input for the 24-hour period (00-00 UTC) before the date of the forecast run.
Initial 2 m temperature and anomaly: Daily mean 2m temperature (°C) from meteorological forcing input for the 24-hour period (00-00 UTC) before the date of the forecast run. Daily mean 2 m temperature irregularities (unitless) from meteorological forcing input for the 24-hour period (00-00 UTC) before the date of the forecast run.

Meteorological (Figure 3)

The data provided under the meteorological theme is information that is used to produce hydrological simulations. Under this theme, three categories of datasets are provided, historical hydro-meteorological time series which is used to derive thresholds used to derive some GloFAS products, real time hydro-meteorological observations which is necessary in defining the initial conditions for hydrological forecasts and meteorological forecasts which are used as future meteorological input for the hydrological models.

At the time of production of this manual, the following datasets are available under the meteorological tab;

Accumulated precipitation: This is the amount of accumulated rainfall over the forecast range of 10 days for the mean of ensemble ECMWF forecast.
Precipitation: Probabilities-this shows the percentage probability of rainfall exceeding a given threshold (current available thresholds 50,150,300) of accumulated rainfall over the forecast range of 10 days for the ensemble ECMWF forecast.
Rainfall animation: this is the animation of the daily (00-00 UTC) precipitation for the first 10 days of the 30-day forecast horizon as the mean of the ECMWF ensemble forecast.

For further reading of the details of the data provided in this category, refer to the GloFAS wiki page.

Hydrological (Figure 4)

Under the hydrological theme, you can access, the following datasets:

Return period exceedance: This shows the probability of predicted ensemble streamflow to exceed a specific return period (5 and 20 years currently available) discharge.
Flood summary: This shows a color categorized map of maximum of ensemble mean discharge (ENS-max) for the specific forecast period (1-3 days, 4-10 days, 11-30 days) which is computed based on a combination of 2, 5 and 20 year exceedance probabilities. For a detailed explanation of the associated legend, refer to GloFAS Flood Summary.
Seasonal outlook reporting points: This shows points where ensemble hydrographs and probability (persistence) tables for the high (>80th percentile) and low (<20th percentile) flow categories are available, displaying the river flow forecast out to 4 months. For details on how to interpret the symbols refer to GloFAS Seasonal Reporting Points.
Seasonal outlook basin overview: This shows the maximum area-averaged probability of unusually high (>80 percentile) or unusually low (<20 percent) weekly river flow for the four-month forecast period in 305 major world basins. For a detailed interpretation of the legend, refer to GloFAS Seasonal basin overview.
Seasonal outlook river network: This shows the maximum area-averaged probability of unusually high (>80 percentile) or unusually low (<20 percent) weekly river flow for the four-month forecast period. This data is only available for model river network for upstream areas greater than 1000 km2. The percentages are represented in color coded categories. For further details refer to GloFAS Seasonal river network overview.

Flood Risk (Figure 5)

Rapid Flood mapping: This shows estimated flood extent at 1km spatial resolution for basins that are larger than 5,000 km2 and the maximum return period is greater than 10 years in the 30 day forecast horizon.
Rapid Impact assessment: This shows the potential impact of floods, aggregated over administrative boundaries, on population, land use (agriculture and urban), schools, health and education based on their intersection with the rapid flood mapping layer. This is also available for basins greater than 5,000 km2 and the maximum return period is greater than 10 years in the 30-day forecast horizon.

Evaluation

Within the evaluation tab, you can access three datasets.

Forecast Skill: Maximum lead time (in days), up to
30-days ahead, when the GloFAS river discharge forecast skill (CRPSS; Continuous Ranked Probability Skill Score) is greater than 0.5, evaluated against both persistence and climatology benchmark forecasts with river discharge reanalysis as reference. The CRPSS ranges from -Inf to 1, with a no skill value of 0 and a perfect value of 1. Detailed results are shown when clicking on individual stations.
Hydrological Model Performance: Modified Kling-Gupta Efficiency (KGE) and the three decomposition scores (bias and variability ratios and correlation) with monthly mean and daily time series for stations with observations. In addition, the daily GloFAS simulation is also provided for the full reanalysis period together with the flood thresholds. Detailed results are shown when clicking on individual stations, which are colored according to the KGE values.
Seasonal Forecast Skill: Maximum lead time (in weeks), up to 16-weeks ahead, when the GloFAS Seasonal River discharge forecast skill (CRPSS; Continuous Ranked Probability Skill Score) is greater than 0.5, evaluated against a weekly climatology benchmark forecast with river discharge reanalysis as reference. The CRPSS ranges from -Inf to 1, with a no skill value of 0 and a perfect value of 1. Detailed results are shown when clicking on individual stations.

Static (Figure 6)

Under this theme are the base datasets that are related to flood monitoring this include:

Major rivers: This layer shows the major rivers.

Major river basins: Includes major river basins from the Global Run off Data center.

Lakes and reservoirs: Includes a subset of lakes and reservoirs from Global Lakes and Wetlands database and the Global Reservoirs and Dams Database based on the following conditions:

Lake surface area above 100 km2

Significant impact on river discharg

Reservoir capacity above 0.5 km3

Flood hazard 100 year return period: This includes areas that are inundated with flood events with a return period of over 100years based on GLOFAS climatology and permanent water bodies derived from the Global Lakes and Wetlands Database and the Natural Earth Lakes map.
Upstream area: GloFAS river network are drawn with river pixels above 100 km2, represented by the upstream area in km2.
Reservoir impact: Shows the potential impact of reservoirs on river discharge at a global scale.
Administrative regions: This shows merged administrative regions merged from NUTS-2 and NUTS-3 classification from EUROSTAT and GADM region classification.

Monitoring

Affected land cover: This layer shows the different land cover classes that have been affected by floods in the forecast period.

Affected population: This layer shows, in form of a thematic map, the number of people affected by floods within an area of 400 m2. In addition to the Sentinel-1 image outline with observed flooding less than 2 km2 and greater than 2 km2, made up of small scattered flooded areas, made up of midsized contiguous flooded areas and made up of large contiguous flooded areas.
Further layers include Exclusion mask, Observed flood extent, Reference water mask, Likelihood values, Sentinel1 Footprint, Sentinel1 Schedule, Advisory Flags and Observed Water Extent.

Additionally, GloFAS provides the possibility to import own data from an external WMS.

With GloFAS, you can get a quick overview of floods around the world. To view the flooded areas for a specific forecast date, under the Flood Risk tab, select the Rapid Flood mapping (Figure 7). In the map viewer, the potentially flooded areas will be highlighted in blue. Please keep in mind that this feature is experimental and results may not always be accurate enough for decision making.