Reconciling Fugitive Dust Emissions with Ambient Measurements: Along the Unpaved Road V. Etyemezian*, J. Gillies, H. Kuhns, D. Nikolic Division of Atmospheric Sciences Desert Research Institute Las Vegas, NV and G. Seshadri, J. Veranth University of Utah SLC, UT Sponsors: WESTAR, DoD/SERDP Outline • Background – What is the phenomenon being examined? – Why is it important? • Modeling approaches – Gaussian plume model (ISC style) – Box Model (Gillette style) – Deposition velocity • Model results – What do initial results tell us? – How does this compare with measurements? – Where do we need improvements? • Conclusions and Future Work Relevance to Fug Dust Emissions • Not accounting for near-field deposition ⇒ – Overestimate fug dust regional contribution (e.g. Chow and Watson, 2000; Countess, 2001) – Erroneous estimate of size distribution of regionally transportable fraction (important for visibility) – Cannot resolve PM10 contributions from varying fug dust sources • Is it the nearby unpaved road, the farm 5 km upwind, or China? • Over-accounting for deposition ⇒ – Similar problems Downwind of Dust Source Downwind Regimes Removal of PM10 dust by vegetation/land cover • Region “A”: – Plume height ~ vegetation height – Work in progress • Region “B”: – Mixed height > Plume height > vegetation height – Addressed in this study • Region “C”: –PM10 ~ constant with height – Addressed in regional scale models Gaussian Plume Approach • Use Gaussian plume model to simulate dispersion • Assumptions: o Gaussian profile is reasonable o Deposition is “slow” compared to dispersion. I.e. can distribute removal over entire plume o Can use bulk deposition models even though concentration gradients high very near the source Gaussian Approach Cont. • Basic equation similar to ISC3: 2 QVD y C = ()1×10−6 exp− 0.5 2πu σ σ σ s y z y Q = pollutant emission rate (g/s) V = a vertical term that includes reflection, deposition, and mixing D = chemical decay term (equal to 1 since dust non-reactive) σ σ y, z = standard deviation of vertical and lateral concentration us = mean wind speed at release height. Gaussian Approach Cont. Assume line source ⇒ Integrate crosswind: Q&V C = ()1×10−6 2π usσ z Vertical term: 2 2 z − h z + h V = exp− 0.5 r + exp− 0.5 r + []..... σ σ z z If h = 0, z = 0, V=2 Gaussian Approach Cont. σ z Gaussian Approach Cont. Vertical Standard Deviation: b σ z = ax Constants a, b from Turner (1970) dependent on atmospheric stability and distance downwind Guassian Approach Cont. Additional Considerations: • Initial height ⇒ + x’ to downwind distance ' b σ z = a()x + x = IH ≈ Hvehicle • Assume WS uniform with height. I.e. WS×Time=Distance • Limited accuracy near ground and near source Box Model Approach • Proposed by Gillette (2002) • Mass balance on long box downwind of unpaved road. Particle flux only at entrance, top, and bottom • Dispersion through top ∝ Concentration dm = K ∗C K = Au* dt top • Deposition at ground ∝ Concentration dm = Vd ∗C dt depos Wind dm/dtup dm/dtceil dm/dtambin dm/dtambout dm/dtroad X X=Xo dm/dtdepos dm/dtceil dm/dtambin dm/dtout ∆Z dm/dtroad Z=0 ∆L=1 Box Model Cont • Resultant Equation for Transportable Fraction dmup V K Φ = dt = 1− d = dm road ()Vd + K ()Vd + K dt * where Vd and K (= 0.06 u ) are constant Dry Deposition • Flux to Ground ∝ Concentration ≡ Fdep = Vd ∗C, Vd Deposition velocity • Framed for removal from uniform concentration in bulk medium (e.g. Slinn, 1982) – Ground release is opposite situation – Roughness elements < concentration profile – Assumptions may apply ~10’s m downwind • Assume removal due to impaction mostly Results: ISC Fraction in Suspension vs. distance 1 0.9 0.8 0.7 0.6 u* = 0.1 m/s 0.5 u* = 0.3 m/s 0.4 u* = 0.5 m/s 0.3 u* = 0.8 m/s 0.2 u* = 1.2 m/s Stability: Neutral Remaining In Suspension Fraction of 8 micron Particles 0.1 0 100 500 1,000 5,000 10,000 Distance Downwind (m) Results: Box Model Prediction 1 u*=0.1 u*=0.3 0.9 u*=0.5 u*=0.8 0.8 u*=1.2 g 0.7 0.6 0.5 0.4 0.3 Fraction Remainin 0.2 0.1 0 × A=0.04 A=0.08 A=0.12 A=0.2 A=0.3 K=A u* Dispersion Parameter in Box Model Results: Box Model Difficulties • Deposition O.K. dm • For first order dispersion, I.e. = K ∗ C dt top Limited applicability • To specify K correctly, require specification of distance downwind Results: Stability and Transportable Fraction u* = 0.3 m/s u*=0.5 m/s Stability 500 m 5,000 m 500 m 5,000 m Very Stable 0.95 0.87 0.72 0.43 Stable 0.96 0.91 0.79 0.57 Neutral 0.97 0.94 0.83 0.66 Moderately Unstable 0.98 0.98 0.91 0.86 Very Unstable 0.99 0.99 0.93 0.92 Note: Particle size differences similar to u* differences Results: Concentration not indicator of removal o 1 0.8 0.6 ISC-Normalized concentration ISC-Fraction in suspension 0.4 Normalized concentration 0.2 Fraction of Particles in Suspension 0 0 200 400 600 800 1000 Distance Downwind (m) Measurements: Daytime desert -DT -SA -DT Top View DT_3 DT_2 DT_1 Legend 12 12 DT: DustTrak Trailer with 20 20 visibility GR: GRIMM equipment SA: Sonic Anemometer LIDAR 3,000 Generator meters : Wind Vane 200 meters : Cup Anemometer : Laptop computer -DT -GR -DT -GR -DT 57 57 0 0 51 7 -DT -GR -DT -DT-GR 26 26 0 0 26 6 12 12 -DT -GR -DT -GR 12 -DT 5 5 8 -DT 76 40 -DT 700 5,000 10,000 Measurements: Nighttime urban? 180 m 9876543210 A Array of cargo B containers. C Dimensions (m): D 2.5 × 2.4 × 12.2 E 176 m F 4 Effective z0: 0.1 m G H Effective u*: 0.4 m/s I 95 m J K 3 m L 2 3 Unpaved Test Road 30 m True 1 Y X North Results: Measurement vs. model • Daytime Desert fraction PM10 removed: – Measured: 0% – Modeled: 3% • Nighttime urban fraction PM10 removed: – Measured: 85% – Modeled: 30% Conclusions • Fraction of PM10 removed near source – Decreases with WS initially, increases at higher WS – Minimum for unstable conditions – Minimum for small roughness height – Not proportional to concentration • Box model has limitation in dispersion Future Work • Gaussian model simple and holds promise • Essential to have field data – Multiple atmospheric conditions – Multiple roughness (vegetative/land cover type and density) • Examine “Region A” removal .