Observational and theoretical investigations of turbulent structures generated by low-Intensity prescribed fires in forested environments

X. Bian, W. Heilman, J. Charney, J. Hom, K. Clark, N. Skowronski, M. Gallagher, and M. Patterson S. Zhong, M. Kiefer, and N. Duan

USDA Forest Service, East Lansing, MI Michigan State University, East Lansing, MI

Results and Conclusion These tests demonstrate quicker upward transport of ground-level heat and momentum fluxes by the FIPFs as compared to the eddy diffusion scheme currently used …

This is the model used in the comparisons described in

Based on field experiment data

The analyses of turbulence data collected before, at, and post fire front indicate that

•The LIPF can generate turbulence

that is much greater than turbulence in the ambient atmosphere.

•The production of turbulence at the surface near the fire front

was associated with increased variance in the 3D winds, specially in vertical direction, while buoyancy was strongest at higher levels within the fire

plume. •Turbulence at 10-m level

nearly doubled that at 3-m level, indicating that the assumption of ‘constant flux layer’ near the surface is no longer valid in the LIPF environment.

•The residual term is generally small in the ambient

environment, but quite large at the fire front and somewhat large immediate before and after the fire front, indicating that other factors in addition to shear and buoyancy are important to local turbulence behavior. Further analyses are needed to identify these specific factors.

IntroductionLow-intensity prescribed fires (LIPF) can be a viable tool for managing forest ecosystems. However, LIPF may significantly modify the atmospheric environment by inducing strong fire-atmosphere interactions. These interactions can lead to intense turbulence production in and around the fire front, which may affect the dispersion of smoke. As part of a broad Joint Fire Science Program (JFSP) project to develop modeling tools for predicting smoke dispersion from LIPFs, the USDA Forest Service - Northern Research Station conducted two LIPFs in 2011 and 2012 on forested plots in the New Jersey Pine Barrens to collect meteorological and air-quality data that can be used for model validation. The data have been used to investigate, among other things, the modification of LIPF on turbulence mixing processes and provide further insight into the complex physical processes that govern turbulent energy production and turbulent diffusion of momentum and heat in the vicinity of LIPFs within forest vegetation layers. In this paper, we describe a turbulence parameterization (flux-profile relationship) developed using the 10 Hz measurements of temperatures and wind speeds using fine-wire thermocouples and sonic anemometers at multiple levels on a 20 me tower during the LIPFs. This parameterization takes into account of modification of LIPF on the mean and turbulent flows in the surface boundary layer .

Theoretical ModelA Non-local Turbulent Closure Model (NTCM) for vertical turbulent mixing in Fire Induced Convective Boundary Layers (FICBL) has been developed for applications of fine-scale atmospheric models to LIPF environments. The NTCM is based on the concept that vertical mixing/transport within the FICBL is inherently asymmetrical and that turbulent flux consists of a local and a non-local component. The non-local upward/downward mixing and transport by buoyant fire-smoke plumes originating in the surface layer is simulated by mixing from the lowest model layer directly to all other layers (Fig 1) in the FICBL. Following Holtslag and Moeng (1991), the vertical turbulent flux of a scalar quantity C can be written as:

where is eddy transport coefficient, is the local vertical gradient for the mean quantity C, and represents the non-local transport. The eddy transport coefficient can be expressed as

where z is height above surface and h is boundary layer height. In the surface boundary layer, z is about 10% of h, and thus , resulting in a simplified formula for eddy transport coefficient

u* is friction velocity that is related to turbulent momentum fluxes by

is a non-dimension function of atmospheric stability, which is calculated following Louis (1979)

The Monin-Obukov length L is defined as

In this study, we focus on turbulent heat flux,

Figure 1. Temperature at 10 and 3 meter levels and heat fluxes (top); the ratio of non-local vs local component in turbulent heat flux before, during and after the fire front passing the tower location (bottom).

Data and MethodUsing high frequency (10Hz) data from the fine-wire thermocouples and sonic anemometers on the 20 m tower in the burn plot during the LIPF in the New Jersey Pine Barrens in March 2011 and 2012, we estimated the ratio of local vs non-local contribution to the total measured sensible heat flux before, during, and after the fire fronts pass the tower location. Specifically, the measured turbulent heat flux, which is the total heat flux, can be decomposed into

where the local contribution can be calculated using

Acknowledgments.

Funding for this research was provided by the U.S. National Fire Plan , the U.S. Joint Fire Science Program (Project # 09-1-04-1), and the USDA Forest Service (Research Joint Venture Agreement # 09-JV-11242306-089).

0.01

0.1

1

10

-40 -20 0 20 40φ(z

/L)

z/L

prefireinfirepostfire

-5

0

5

10

15

20

25

0

10

20

30

40

50

60

Sens

ible

Hea

t Flu

x (K

m s-1

)

Tem

pera

ture

(o C)

Tm10 Tm03 Flux

-10

0

10

15:29 15:49 16:09 16:29 16:49 17:09 17:29 17:49 18:09

Ratio

= N

on-lo

cal/

Loca

l

Time of 2011-03-20 (EST)

Pre-fireIn-firePost-fire

Blackadar (1978, 4th Symp. on Atmospheric Turbulence, Diffusion and Air Quality, pp. 443–447, Reno, Am. Meteorol. Soc.) Following Holtslag and Moeng (1991 J.A.S: Eddy Diffusivity and Countergradient Transport in Convective Atmospheric Boundary)

)('' CC ZCKCW

CK zC C

2* )1(hzzuK

CC

1)1( 2 hz

C

CC

zuK

*

2/1)/91(74.0 LzC

2/1222* ])''()''[( vwuwu

'')/(

2*

wguL

C

localnonlocal WWW ''''''

Fig. 1. Illustration of local vs non-local contribution to measured turbulent fluxes

)(''Z

KCW local

The mean vertical gradient can be calculated using the measured potential temperature at two layers, and eddy transport coefficient for heat can be calculated using measured momentum and heat fluxes. The ratio of local to non-local components can thus be estimated as

zK

zKW

WWWR

local

local

''

''''''

Results