A Little About Flood Forecasting

The Member States of the Caribbean Community are island states or continental countries with highly populated coastal plains lying below sea level. The Member States are subject to repeated flooding, some on an annual basis, with massive resultant loss in life and significant economic losses.

This project proposes to develop a robust, reproducible, and transparent approach to flood forecasting that utilizes both physically based and lumped hydrological with numerical weather prediction inputs. The approach overcomes many of the deficiencies encountered in more traditional approaches to flood forecasting in small watershed where there is a short time lag between precipitation events and the onset of flood.

Flood forecasting techniques currently implemented in some Caribbean territories often rely on (i) a series of precipitation measurements in the upland portions of watersheds to infer potential flooding downstream, and/or (ii) coarse precipitation forecasts from national meteorological services. In the former case, the time between the measured precipitation and the onset of flooding is often very short as most watersheds on Caribbean islands are small (10s of km2). In the latter case, coarse precipitation forecasts provide limited spatial and quantitative information to support a quantitative assessment of the potential for flooding. It is important to note that in the approaches mentioned little consideration is given to physical state of the watershed, which is a critical factor controlling flood development. This project was initiated through the Real-Time Flood Forecasting (RTFF) initiative which was sponsored by the Government of Japan through the CARICOM-Japan Friendship and Cooperation fund with the CARICOM Secretariat and the CIMH as the executing and implementing agencies respectively.

The current approach utilizes rainfall predictions from WRF-ARW for the Caribbean region. Predictions for selected watersheds are used as input to HEC-HMS to produce a flow prediction. Future hydrological prediction products will utilize the WRF-HYDRO code for a fully coupled approach which is capable of tracking changes in the hydrological characteristics of watershed that influence flood development, intensity, and duration. The WRF model is currently being used by CIMH staff to develop 48-hour weather predictions across the Caribbean at various spatial and temporal scales.

Rainfall Prediction Tools

WRF in a Nutshell

The Advance Research Weather Research and Forecasting model(hereinafter referred to as WRF) is an atmospheric simulation system that is portable on various computing platforms and allows for application across ta wide range of spatial scales (from tens of meters to thousands of kilometers) and temporal scales (seconds to hours). The WRF system is updated and maintained by the Mesoscale and Microscale Meteorology Division of NCAR. More information about this model can be found here.

Outputs are provided at a 4 km spatial resolution for 00Z, 06Z, 12Z and 18Z initialisations with two (2) runs for selected initialisations. The approach used is a one-way nesting approach. Two (2) products are available for viewing. These are:

(1) 48-hour aggregated accumulations following model initialization:

This product provides an hourly precipitation aggregate from model initialization. Hence, the final image provides an aggregated accumulation of the modelled precipitation for the 48-hour prediction period following initialization. This product can be used to support the identification of areas that may experience significant rainfall over the 48-hour prediction period.

(2) Disaggregated hourly accumulations for the model run:

This product displays hourly accumulations for the current 48-hour prediction period. This product supports prediction of intense rainfall events that can lead to flash flooding.

Flow Prediction Tools


The Hydrologic Modeling System (HEC-HMS) is designed to simulate the complete hydrologic processes of dendritic watershed systems. The software includes many traditional hydrologic analysis procedures such as event infiltration, unit hydrographs, and hydrologic routing. HEC-HMS also includes procedures necessary for continuous simulation including evapo-transpiration, snowmelt, and soil moisture accounting. Advanced capabilities are also provided for gridded runoff simulation using the linear quasi-distributed runoff transform (ModClark). Supplemental analysis tools are provided for parameter estimation, depth-area analysis, flow forecasting, erosion and sediment transport, and nutrient water quality. More information can be found here.

Disaggregated rainfall predictions for the watersheds of interest are extracted at runtime and reformatted to the native HEC-HMS gridded rainfall format. Flow predictions are produced at selected outlets for each watershed. Data are currently being collected and hydrometric networks expanded to support model optimisation. Currently, outputs are available for the Rio Cobre, Jamaica, Speightstown, Barbados and Layou, Dominica via links on the rainfall prediction pages for those countries.  More watersheds will be added as resources permit. Model outputs are uncalibrated. However, The CIMH is currently expanding its monitoring network through support from various initiatives. Data from the network will support calibration and validation procedures. Future work will also include the generation of multiple flow scenarios by sampling from a range of input model parameters.