Roads, 2006 Firedanger Applications of NCEP's Downscaled** CFS Forecasts In the mid 90’s, USFS...

23
Roads, 2006 Applications of NCEP's Downscaled** CFS Forecasts In the mid 90’s, USFS researchers (Fujioka) asked if we could develop routine weekly to seasonal fire danger predictions. Fortunately, we had already started collaborating with NCEP modelers (Juang and Kanamitsu), who were developing an experimental global to regional forecast system (GSM/RSM) that we thought was highly appropriate for our experimental effort. FWI Predictions (ca. 1997-2000) ECPC predictions Initial FDI Efforts (ca. 2001-2004) ECPC predictions Current FDI Efforts (ca. 2005-2008) NCEP GSM/RSM** ensemble predictions **See Wang and Juang’s poster on the RSM J. Roads*, P. Tripp* H. Juang**, J. Wang**, S. Chen***, F. Fujioka*** *ECPC, **NCEP, ***USFS

Transcript of Roads, 2006 Firedanger Applications of NCEP's Downscaled** CFS Forecasts In the mid 90’s, USFS...

Page 1: Roads, 2006 Firedanger Applications of NCEP's Downscaled** CFS Forecasts In the mid 90’s, USFS researchers (Fujioka) asked if we could develop routine.

Roads, 2006

Firedanger Applications of NCEP's Downscaled** CFS Forecasts

• In the mid 90’s, USFS researchers (Fujioka) asked if we could develop routine weekly to seasonal fire danger predictions.

– Fortunately, we had already started collaborating with NCEP modelers (Juang and Kanamitsu), who were developing an experimental global to regional forecast system (GSM/RSM) that we thought was highly appropriate for our experimental effort.

• FWI Predictions (ca. 1997-2000)• ECPC predictions

• Initial FDI Efforts (ca. 2001-2004)• ECPC predictions

• Current FDI Efforts (ca. 2005-2008)• NCEP GSM/RSM** ensemble predictions

**See Wang and Juang’s poster on the RSM

J. Roads*, P. Tripp*H. Juang**, J. Wang**, S. Chen***, F. Fujioka***

*ECPC, **NCEP, ***USFS

Page 2: Roads, 2006 Firedanger Applications of NCEP's Downscaled** CFS Forecasts In the mid 90’s, USFS researchers (Fujioka) asked if we could develop routine.

Roads, 2006

Firedanger predictions The USFS NFDRS describes the potential for fire danger.

– The NFDRS is based on historical USFS wind tunnel, fuel bed research (Rothermal).

– The NFDRS uses input from weather stations. Our long-term goal has been to show that NWP models can provide long-range

forecasts of the NFDRS.– USFS and other land agencies could use accurate long-range forecasts for

planning purposes.– In fact, there is 1 such meeting each year in early Mar. to make this assessment.

Model resolution is a continual problem.– Long range forecasts use coarse resolution models– Application communities want high resolution input and have started to use some of

the regional analysis products– We have been using the NCEP GSM/RSM forecast system to connect to the fire

danger application community• The RSM provides relatively high resolution input to FDI• As shown previously, the GSM & RSM have at least comparable forecast skill.• We are hopeful the RSM will eventually be shown to provide additional useful,

skillful features

Page 3: Roads, 2006 Firedanger Applications of NCEP's Downscaled** CFS Forecasts In the mid 90’s, USFS researchers (Fujioka) asked if we could develop routine.

Roads, 2006

A higher resolution Fire Danger Code

And updated fire statistics (Western States!)

Page 4: Roads, 2006 Firedanger Applications of NCEP's Downscaled** CFS Forecasts In the mid 90’s, USFS researchers (Fujioka) asked if we could develop routine.

Roads, 2006

Vegetation in 25km Fire Danger Code

Model A Annually varying Western grasslandsModel B Mature dense fields of brushModel C Open pine standsModel D Southeast coastal pine standsModel F California chaparralModel G Dense conifer with heavy litterModel H Short needled conifersModel L Perennial grasses

Model N Florida sawgrass Model O Dense brushlike fuels of SoutheastModel P Closted stands of long-needled southern pinesModel Q Upland Alsaskan black spruceModel R Deciduous hardwordModel S Alaskan tundraModel T Great Basin sagebrush grassModel U Closed stands of western long-needled pines

Page 5: Roads, 2006 Firedanger Applications of NCEP's Downscaled** CFS Forecasts In the mid 90’s, USFS researchers (Fujioka) asked if we could develop routine.

Roads, 2006

Slopes in 25 km FD

Page 6: Roads, 2006 Firedanger Applications of NCEP's Downscaled** CFS Forecasts In the mid 90’s, USFS researchers (Fujioka) asked if we could develop routine.

Site DescriptionFuel ModelSlope Class

Live Fuel TypesClimate Class

AverageAnnual

Precipitation

1300 LSTObservation

24-HourObservations

Carryover FuelMoistures (FM)

Relative HumidityTemperatureCloudiness

Wind Speed

Fuel StickMoisture

Max/Min RH

Min Temp

PrecipitationDuration

PrecipitationAmount

100-Hour

1000-Hour

Live Woody

FM

1-hr FM 10-hr FM 100-hr FM 1000-hr FMKBDI

Live FM

Drought Fuel

MaximumTemperature

PeriodicMeasurements

Season Code &Greenness Factor

SpreadComponent

SC

Burning Index BI

Energy Release Component

ERC

Ignition Component

IC

Cal

cula

ted

Inp

ut

Ou

tpu

t

September 19,2000

Contribution of dead FM to SC

Contribution of dead FM to ERC

(88)

(88)

(88)

National Fire Danger Rating System Structure

Optional pathway

Latitude

Page 7: Roads, 2006 Firedanger Applications of NCEP's Downscaled** CFS Forecasts In the mid 90’s, USFS researchers (Fujioka) asked if we could develop routine.

Roads, 2006

USFS Fire Danger Indices

1. SC is an index of the forward rate of spread at the head of a fire and is sensitive to wind speed.

2. ER is a number related to the available energy per unit area within the flaming front at the head of a fire. ER is not affected a by wind speed.

3. IC is a rating of the probability that a firebrand will cause a fire requiring suppression action. SC is a component of IC.

4. BI is a number related to the contribution of fire behavior to the effort of containing a fire. SC and IC contribute to the BI.

5. KB is a stand-alone index that can be used to measure the affects of seasonal drought on fire potential.

6. FWI was derived by Fosberg (1978) who assumed constant fuel (vegetation=grass) characteristics as well as climatic initial conditions. The FWI is most easily applied in practice and provides a first look at fire danger globally. It is a grassland approximation to BI.

Roads, J., F. Fujioka, S. Chen, R. Burgan, 2005: Seasonal Fire Danger predictions for the USA. International Journal of Wildland Fire, Special Issue: Fire and Forest Meteorology, 14, 1-18.

Page 8: Roads, 2006 Firedanger Applications of NCEP's Downscaled** CFS Forecasts In the mid 90’s, USFS researchers (Fujioka) asked if we could develop routine.

Roads, 2006

Firedanger predictions All fire danger indices (except FWI) depend upon the previous FDI

history. We must therefore use the best available data to drive the validating and initializing fire code

We use 1 day RSM predictions, which are closely related to NCEP analyses, except we can easily access all the needed variables in near real time with the same model we use for longer-range predictions.

Analysis/Forecast precipitation has been a problem! Fortunately,– Daily CPC precipitation at .25 degrees is now available in near-real

time and this precipitation is used in place of predicted precipitation to update the fire danger code every day. (poor man’s LDAS)

We validate the fire danger seasonal forecasts with the validating/initializing fire danger values and now also

Fire occurrence data (counts, ln area burned), which are available at coarse temporal (monthly) and spatial (1-deg.) resolution (cf. A. Westerling).

Page 9: Roads, 2006 Firedanger Applications of NCEP's Downscaled** CFS Forecasts In the mid 90’s, USFS researchers (Fujioka) asked if we could develop routine.

Roads, 2006

NCEP Global to Regional predictions NCEP CFS T62L64 SST forces NCEP GSM T62L28 NCEP GSM T62L28 forces NCEP L28 RSM (US 50 km) A continuous series of 1-day runs have now been made from 1982-

present, to provide (along with Higgins precipitation) the initialization/validation data for fire danger code

Ten 7-month predictions made monthly starting from 0000 and 1200 UTC of the first 5 days of current month and last 5 days of previous month. – Experimental predictions began Oct. 2004 and have been continued

every month without fail since then. 3 hindcasts (the first two days of month and the last day of the

previous month) initialized from the NCEP/DOE reanalysis for the same month but each year 1982-2004, or 23*3 mon. hindcasts. – more hindcast members may be added later if model not upgraded. In

fact, many sensitivity experiments are underway– a new land model, increased vertical resolution, etc.

Page 10: Roads, 2006 Firedanger Applications of NCEP's Downscaled** CFS Forecasts In the mid 90’s, USFS researchers (Fujioka) asked if we could develop routine.

Roads, 2006

US WestTime Series5 month running means. Validation (dark lines) and 2-month lead forecasts, (red lines)

Note summer has largest values

CN

AC

SC

ERC

IC

KB

BI

FWI

Page 11: Roads, 2006 Firedanger Applications of NCEP's Downscaled** CFS Forecasts In the mid 90’s, USFS researchers (Fujioka) asked if we could develop routine.

Roads, 2006

US WestAnomaly Time Series5 month running means. Validation (dark lines) and 2-month lead forecasts, 2+5 (red lines) Note lower values in 80s and higher values in 90’s

CN

AC

SC

ERC

IC

KB

BI

FWI

Page 12: Roads, 2006 Firedanger Applications of NCEP's Downscaled** CFS Forecasts In the mid 90’s, USFS researchers (Fujioka) asked if we could develop routine.

Roads, 2006

MJJAS Valid Mar. 1 Fcst

Page 13: Roads, 2006 Firedanger Applications of NCEP's Downscaled** CFS Forecasts In the mid 90’s, USFS researchers (Fujioka) asked if we could develop routine.

RSM FDI fcst biases are fairly similar for all of the indices, note the positive biases along the CA coast and over the GP and the negative values over the US West.Bias errors on the order of 10% of mean values

Page 14: Roads, 2006 Firedanger Applications of NCEP's Downscaled** CFS Forecasts In the mid 90’s, USFS researchers (Fujioka) asked if we could develop routine.

RSM RH, T, and pcp have fairly similar biases. It is too cold and wet over the US West, and too hot and dry over the Great plains. The wsp is too high along the coast and over the southern GP and too low over the US West. CLDC not

influential

Tmax, Tmin

RHmax, RHmin

Precip, Dur.

WSP, CLDC

Page 15: Roads, 2006 Firedanger Applications of NCEP's Downscaled** CFS Forecasts In the mid 90’s, USFS researchers (Fujioka) asked if we could develop routine.

GSM biases are somewhat similar.GSM fcsts are too warm and dry over most of US. The wsp is too high along the coast and over the GP and too low over the US West. The CLDC bias is the same. Precip, unfortunately, was missing from initial GSM archives. We are now trying to recover GSM precip. to compare GSM & RSM FDIs. Need to determine the sensitivity to input errors

Tmax, Tmin

RHmax, RHmin

Precip, Dur.

WSP, CLDC

Page 16: Roads, 2006 Firedanger Applications of NCEP's Downscaled** CFS Forecasts In the mid 90’s, USFS researchers (Fujioka) asked if we could develop routine.

Roads, 2006

Seas. Valid MJJAS 1983 Mar. 1 Fcst

Page 17: Roads, 2006 Firedanger Applications of NCEP's Downscaled** CFS Forecasts In the mid 90’s, USFS researchers (Fujioka) asked if we could develop routine.

Roads, 2006

Seas. Valid MJJAS 1994 Mar. 1 Fcst

Page 18: Roads, 2006 Firedanger Applications of NCEP's Downscaled** CFS Forecasts In the mid 90’s, USFS researchers (Fujioka) asked if we could develop routine.

Roads, 2006

Correlations of MMJJA validations with Mar. 1 forecast

Different indices have slightly different preferred regions

Page 19: Roads, 2006 Firedanger Applications of NCEP's Downscaled** CFS Forecasts In the mid 90’s, USFS researchers (Fujioka) asked if we could develop routine.

Roads, 2006

FDI Correlations

3 month ave. 5 month ave.

Page 20: Roads, 2006 Firedanger Applications of NCEP's Downscaled** CFS Forecasts In the mid 90’s, USFS researchers (Fujioka) asked if we could develop routine.

Roads, 2006

Correlation of FDI and ln counts

Correlation of FDI and ln acres burned

Page 21: Roads, 2006 Firedanger Applications of NCEP's Downscaled** CFS Forecasts In the mid 90’s, USFS researchers (Fujioka) asked if we could develop routine.

Roads, 2006

Correlation of fcst FDI and ln counts

Page 22: Roads, 2006 Firedanger Applications of NCEP's Downscaled** CFS Forecasts In the mid 90’s, USFS researchers (Fujioka) asked if we could develop routine.

Roads, 2006

FDI MJJAS Correlations

Mar. 1 Fcst

Page 23: Roads, 2006 Firedanger Applications of NCEP's Downscaled** CFS Forecasts In the mid 90’s, USFS researchers (Fujioka) asked if we could develop routine.

Roads, 2006

Summary• We are currently working with NCEP and USFS to develop routine US fire

danger forecasts from the CFS– Daily RSM products and observed precipitation from 1982-present now

provide a long term fire danger initialization/validation set for an upgraded fire danger model.

– This initialization/validation RSM set is used as the initialization/validations for 7-month and historical fire danger prediction ensembles (10+23x3).

• Preliminary results for Mar. 1 forecast of MJJAS are encouraging but analysis is continuing. – Need to include additional ensemble forecasts/hindcasts to increase skill and

estimate significance!– Need to determine added value of RSM to CFS/GSM– Need to determine added value of GSM/RSM to persistence– Need to determine whether it makes a significant difference if we use 1-day

GSM/RSM predictions and obs. Precip. instead of NARR for initialization and validation of FDI.

– Need to determine if other skill measures may be better measures of firedanger skill.