NHRP / Hazard themes / Weather, Flood and Coastal Hazards / Research Highlights 2014-15

Research Highlights 2014-15

A photograph of wildfire taken by Scion.

Photo: Scion

Mitigating Wildfire Risk – The Role of Weather Prediction

The NIWA-Scion fire weather system (FWSYS) provides practitioners and emergency managers with information that alerts them to potentially dangerous conditions and provides access to information to understand fire risk in the rural landscape and how that risk is expected to change over the coming days. 

This new FWSYS builds on weather prediction research and tool development carried out in the Weather Hazards theme, and a strong collaboration with Scion.

Estimating Fire Risk: The New Zealand Fire Danger Rating System

The Fire Weather Index system uses observations of temperature, relative humidity, wind speed  and rainfall to infer the effects of fuel moisture and weather on ignition potential and probable fire behaviour (Fig.1). 

A diagram depicting NIWA's NZ Fire Weather Index System.

 Fig 1. New Zealand Fire Weather Index (FWI) System - Four weather inputs, three fuel moisture codes and three fire behaviour indices. Higher code and index values indicate more severe fire potential.

The five fire danger classes (Low to Extreme) that we see on the side of highways are the outcome of the Head Fire Intensity (HFI) predictor for each of three major land classifications: Forest, Scrub and Grassland. HFI indicates the intensity of the head of a fire, which takes account of rate of spread and land attributes acting as  fuel. HFI can inform responders and/or decision-makers about the effectiveness of different types of suppression resources in containing a fire. 

The FWSYS computes rate of spread and head fire intensity for each land cover class separately, and as an integrated product that takes account of land classification type. As an example for Grasslands, the FWSYS can take account of the degree of grassland curing, a measure of the proportion of dead grass material present reflecting the stage of seasonal grass die-off.

The current fire danger class as well as the FWI codes and indices are calculated daily from data supplied from around 200 automatic weather stations spread across the country. Using geospatial processing techniques these site-specific data are interpolated onto a 5 km grid that covers the whole country, allowing estimates of current fire danger to be provided for any location.

Forecasting the Future Risk

Using numerical weather prediction (NWP) models, forecasts of FWI codes and indices, and of Fire Danger Class can be estimated, up to six days ahead. 

A diagram showing the interactions required to generate a 6-day ahead forecast for the Fire Weather Index.

Fig 2. The implementation of the NZFDRS with EcoConnect. The diamonds containing 'FWI' denote the system which generates the moisture codes, fire indices and danger classes from input weather variables.

Two different NWP models lie behind these calculations: NZLAM (the New Zealand Limited Area Model) is used to provide a high spatial resolution (12 km) forecast of weather conditions out to two days ahead, then a global NWP model with 20 km spatial resolution, to extend the forecast conditions out to six days ahead (Fig. 2). The NZLAM-based fire danger forecasts are expected to be the most accurate over the first 2 days, and then the Global model based forecasts for the following 4 days. 

A diagram depicting observed fire danger classes and that forecasts are encouragingly similar to actual observed data.

Fig3. Observed Fire Danger Classes.

Figure 3 shows example forecasts of national Fire Danger Class. Fig. 3a-c show the observed values for the days of 7,9,11 February 2015 ; Fig. 3d-e show the forecast Fire Danger Class using forecasts initiated on the 7 February (and available to end users on that date). The forecast for 9 February (d) is based on a 2-day weather forecast from NZLAM, while that for 11 February (e) is based on a 4-day forecast from the Global model. Accordingly Fig. 3a-c show what actually happened, and Fig. 3d-e show the forecast fire danger class up to 4-days ahead.  The forecast and actual fire danger classes are encouragingly similar, demonstrating the value of the forecast data to mitigate fire risk through management decisions in advance of days with forecast high (or low) fire danger.

Decision Tool

Using the EcoConnect FWSYS PC application, users can access a large range of information to monitor fire weather conditions, determine when to carry out fire prevention publicity efforts, setting of fire season restrictions, decide whether to issue burning permits, and setting readiness levels for fire suppression resources. 

The new FWSYS is flexible and automatically leverages advances in publicly funded research on increasingly more accurate weather prediction models. This will lead to even greater accuracy of both current and forecasted fire danger estimates in future. 

Contact: Michael Uddstrom, NIWA

Back to

Last updated 20 Jan 2016