Anda di halaman 1dari 30

6 atmosfer Dispersi

dispersi atmosfer adalah proses yang sangat penting dalam polusi udara. Penggunaan tumpukan tinggi untuk
meminimalkan dampak dari emisi polutan udara re fl ected yang mengatakan bahwa “ solusi untuk polusi adalah
pengenceran. ” Meskipun ini pepatah ternyata salah karena beberapa alasan, termasuk efek kumulatif dari
sejumlah besar sumber individu dan pembentukan polutan sekunder pada skala regional yang besar, dispersi
atmosfer adalah tetap salah satu proses kunci yang mengatur tingkat polusi udara. Bab ini menyajikan proses
dasar dari dispersi atmosfer, teori dasar mereka, serta kelebihan dan kekurangan dari berbagai jenis model
dispersi atmosfer.

6.1 Pertimbangan Umum Atmosfer Dispersi

Polutan tersebar, dan karena itu diencerkan, di atmosfer sebagai akibat dari turbulensi. turbulensi ini mencakup
berbagai skala spasial, termasuk pusaran skala besar, pusaran kecil, dan seterusnya sampai dispersi energi
kinetik pada skala molekul. Lewis Fry Richardson (1881 - 1953), seorang fisikawan Inggris, diringkas kaskade ini
proses dengan cara yang lebih puitis:

“ Pusaran besar memiliki berpusar kecil yang pakan pada kecepatan mereka,

dan berpusar sedikit memiliki berpusar lebih rendah dan sebagainya untuk viskositas. ”

Karena tidak mungkin untuk mewakili turbulensi dengan cara deterministik matematika, perlu untuk
memperkenalkan representasi perkiraan dari proses utama yang terlibat. Karena kehadiran Bumi ' s permukaan,
yang menciptakan anisotropi di lapisan bawah atmosfer, adalah tepat untuk membedakan antara turbulensi di
arah vertikal dan horisontal.

Hasil turbulensi vertikal dari proses yang disebutkan dalam Bab 4, yang meliputi turbulensi mekanis dan
turbulensi termal. Yang terakhir sangat bergantung pada struktur termal dari lapisan batas planet (PBL); sangat
besar ketika suasana tidak stabil dan lemah ketika suasana stabil. fenomena konveksi, yang terjadi pada PBL
serta di troposfer bebas, campurkan juga paket udara (dan, karena itu, polusi udara) dalam arah vertikal, tetapi
mereka biasanya sesuai dengan gerakan yang koheren dan, karena itu, tidak diperlakukan secara rinci di sini .

turbulensi horisontal dapat dibuat dengan proses mekanis dalam PBL, serta oleh proses termal. Hal ini lebih
menantang untuk parameterisasi proses fisik yang relevan,
95

download dari https://www.cambridge.org/core . University of Texas Perpustakaan, pada 8 Desember 2019 di 8:58:32, tunduk pada Cambridge Inti persyaratan penggunaan, tersedia di
https://www.cambridge.org/core/terms . https://doi.org/10.1017/9781108674614.006
96 atmosfer Dispersi

dibandingkan dengan mereka yang mengatur gerakan vertikal. Oleh karena itu, pemodelan matematika yang bisa dif fi kultus, terutama pada

skala spasial yang sangat besar.

proses turbulen dapat diatasi baik melalui Lagrangian atau pendekatan Eulerian. Dalam pendekatan Lagrangian, salah
satu kegunaan sistem referensi yang bergerak dengan angin dan turbulen berarti gerakan yang dipelajari sehubungan
dengan sistem referensi bergerak. Oleh karena itu, salah satu mengacu relatif difusi turbulen, karena relatif terhadap sistem
referensi ini. Dengan demikian, berkelok-kelok yang membanggakan sebuah, yang mungkin hasil dari gerakan
besar-besaran yang mempengaruhi sistem referensi, tidak termasuk. difusi berkoresponden relatif terhadap ukuran
membanggakan seketika, karena rata-rata ukuran bulu-bulu dari waktu ke waktu akan sesuai dengan termasuk beberapa
berkelok-kelok membanggakan dalam ukuran membanggakan. Dalam pendekatan Eulerian, ansambel dari gerakan
turbulen dianggap sehubungan dengan fi sistem referensi yang tetap. Oleh karena itu, salah satu mengacu mutlak difusi
turbulen. Kedua pendekatan menjadi setara ketika pendekatan Lagrangian mencakup rentang penuh dari pusaran bergolak.

6.2 Representasi Lagrangian dari Turbulence

Dalam representasi Lagrangian turbulensi, satu mengikuti evolusi statistik jarak antara dua partikel (atau
molekul) yang dipancarkan dari sumber yang sama (Csanady,
1973). Jika partikel hadir di lokasi x ' pada waktu t = 0, gerak acak dapat diwakili dengan kepadatan probabilitas p
(x - x ', t) seperti yang p (x - x ', t) d x probabilitas bahwa gerakan partikel selama waktu t lead kedatangan di ruang d x
berpusat di lokasi x. Kemudian, konsentrasi partikel hadir pada saat t di lokasi x ( yaitu, massa partikel dalam
volume d x) adalah sebagai berikut:

C ð x; t j x 0 Þ ¼ S 0 p ð xx 0; t Þ ð 6: 1 Þ

dimana S 0 adalah massa partikel yang dipancarkan pada saat 0 di lokasi x '. Asumsi dari proses difusi acak
bergolak mengarah ke distribusi gaussian konsentrasi partikel dengan standar deviasi σ dan maksimum pada
lokasi sumber x '.
Dalam tiga dimensi, standar deviasi dari distribusi partikel menjalani bergolak diffusionmay menjadi de fi ned
dalam satu arah (di sini x) untuk seluruh populasi partikel sebagai berikut (dengan asumsi bahwa sumber adalah
di x '= y '= z '= 0):
ððð

σx2¼ 1 x 2 C ð x; y; z; t Þ d x d y d z ¼ ð ð ð x 2 p ð x; y; z; t Þ d x d y d z ¼ x 2 ð 6: 2 Þ
S0

Oleh karena itu, standar deviasi dari distribusi partikel dalam satu arah setara dengan akar rata-rata kuadrat dari
gerakan partikel ke arah itu. Jika turbulensi seragam dan isotropic, maka: σ x = σ y = σ z = σ. Jarak yang ditempuh oleh
partikel pada kecepatan
u selama waktu t adalah:

x ð t Þ ¼ ðt u ð t0 Þ d t0 ð 6: 3 Þ

download dari https://www.cambridge.org/core . University of Texas Perpustakaan, pada 8 Desember 2019 di 8:58:32, tunduk pada Cambridge Inti persyaratan penggunaan, tersedia di
https://www.cambridge.org/core/terms . https://doi.org/10.1017/9781108674614.006
97 6.2 Representasi Lagrangian dari Turbulence

Oleh karena itu, perubahan kuadrat dari jarak yang ditempuh sebagai fungsi waktu adalah sebagai berikut:

ðt
d x2 ð t Þ
u ð t Þ u ð t0 Þ d t0 ð 6: 4 Þ
dt¼2xdx dt¼2
0

Mengambil ensemble rata-rata untuk kedua sisi persamaan:


ðt
d x2 ð t Þ
u ð t Þ u ð t0 Þ d t0 ð 6: 5 Þ
dt¼2
0

Autokorelasi antara kecepatan dari dua partikel di kali t dan t '= t + τ L adalah sebagai berikut:

R ð τL Þ ¼ u ð t Þ u ð t þ τL Þ ð 6: 6 Þ
u2

Oleh karena itu, perubahan dari deviasi standar sebagai fungsi waktu ditulis sebagai berikut (d t ' menjadi d τ L):

ðt
d σ2 ð t Þ
R ð τL Þ d τL ð 6: 7 Þ
d t ¼ 2 u2
0

Ungkapan ini dikenal sebagai Taylor ' s Teorema (1922) (Geoffrey Ingram Taylor adalah seorang ahli fisika Inggris
dan matematika). Setelah integrasi:

ð tð t 0
σ2 ð t Þ ¼ 2 u2 R ð τL Þ d τL d t0 ð 6: 8 Þ

0 0

Integrasi parsial menyediakan solusi untuk kuadrat dari standar deviasi dari distribusi partikel untuk proses
dispersi mapan:
24 35
ðt ðt
σ2 ð t Þ ¼ 2 u2t R ð τL Þ d τL τL R ð τL Þ d τL ð 6: 9 Þ

0 0

Jadi:

ðt
σ2 ð t Þ ¼ 2 u2 ðt τL Þ R ð τL Þ d τL ð 06:10 Þ

dimana σ adalah standar deviasi dari distribusi partikel dalam satu arah ( σ x di arah angin rata-rata, σ y arah
crosswind horisontal, σ z dalam arah vertikal), t adalah waktu sejak emisi, u adalah kecepatan angin, dan R ( τ L) adalah
fungsi korelasi Lagrangian untuk seragam dan turbulensi stasioner. Untuk situasi seperti ideal (konstan dalam
ruang dan waktu), salah satu dapat menggunakan proses Markov untuk mewakili nilai-nilai yang berurutan dari
kecepatan partikel tersebar. (Dalam proses Markov, prediksi masa depan adalah

download dari https://www.cambridge.org/core . University of Texas Perpustakaan, pada 8 Desember 2019 di 8:58:32, tunduk pada Cambridge Inti persyaratan penggunaan, tersedia di
https://www.cambridge.org/core/terms . https://doi.org/10.1017/9781108674614.006
98 atmosfer Dispersi

dilakukan hanya menggunakan informasi tentang keadaan sekarang dengan tidak ada informasi tentang negara-negara sebelumnya;

meskipun tidak sepenuhnya benar untuk bergolak sebuah fl ow, itu adalah pendekatan yang baik kecuali untuk waktu yang sangat

singkat) Kemudian, fungsi korelasi dapat diwakili oleh fungsi eksponensial.:

R ð τ L Þ ¼ exp ð? τ L Þ ð 06:11 Þ
tL

dimana t L adalah skala waktu Lagrangian. Hal ini de fi didefinisikan sebagai integral dari fungsi korelasi:

tL¼ ð∞ R ð τL Þ d τL ð 06:12 Þ

Skala waktu Lagrangian berada di urutan 60 sampai 100 s di lapisan batas planet. Dekat sumber, yakni sebagai t ->
0, R ( τ L) -> 1 dan solusi untuk deviasi standar:

ðt
t2
σ2 ð t Þ ¼ 2 u2 ðt τL Þ d τL¼ 2 u2t2 ð 06:13 Þ
2
0

Karena itu:
σ2 ð t Þ ¼ u2t2 ð 06:14 Þ

Standar deviasi σ sebanding dengan standar deviasi dari kecepatan angin seketika, di sini σ u, dan untuk waktu
berlalu sejak emisi (atau jarak yang ditempuh oleh partikel dari sumber jika mean kecepatan angin adalah
konstan, x = ut).
Jauh dari sumber, yakni sebagai t -> ∞, R ( τ L) -> 0 dan solusi untuk deviasi standar:

ð∞ ð∞ ð∞
σ2 ð t Þ ¼ 2 u2 ðt τL Þ R ð τL Þ d τL¼ 2 u2t R ð τL Þ d τL2 u2 τL R ð τL Þ d τL ð 06:15 Þ

0 0 0

Mari kita de fi ne:

t0¼ ð∞ R ð τL Þ d τL

0
ð 06:16 Þ
ð∞
t1¼ 1 τL R ð τL Þ d τL
t0
0

Kemudian, σ 2 ( t) dapat ditulis sebagai fungsi dari t, t 0, dan t 1:

σ2 ð t Þ ¼ 2 u2t0 ð t t1 Þ ð 06:17 Þ

Untuk fungsi korelasi eksponensial, diperoleh:

t0¼ tL ð 06:18 Þ

download dari https://www.cambridge.org/core . University of Texas Perpustakaan, pada 8 Desember 2019 di 8:58:32, tunduk pada Cambridge Inti persyaratan penggunaan, tersedia di
https://www.cambridge.org/core/terms . https://doi.org/10.1017/9781108674614.006
99 6.3 Euler Representasi Turbulensi

Nilai dari t 1 diperoleh dengan integrasi parsial:

t1¼ tL2 ¼ tL ð 06:19 Þ


t0

Karena itu:
σ2 ð t Þ ¼ 2 u2ð tLt tL2 Þ ð 06:20 Þ

dan karena t » t L:

σ2 ð t Þ ¼ 2 u2tLt ð 06:21 Þ

Standar deviasi σ sebanding dengan akar kuadrat dari waktu berlalu sejak emisi atau akar kuadrat dari jarak x bepergian
dengan partikel dari sumber jika mean kecepatan angin adalah konstan ( x = ut). Oleh karena itu, Lagrangian
representasi mengarah ke proses dispersi yang meningkat lebih cepat di dekat sumber (sebanding dengan t atau
x) dari jauh hilir (sebanding dengan t 1/2 atau x 1/2). Pada tahun 1922 artikelnya, Taylor menyebutkan bahwa hasil ini
konsisten dengan pengamatan yang diperoleh di laboratorium dan fi pengamatan lapangan dari bulu stack.
Memang, bulu-bulu di dekat sumber ditampilkan bentuk kerucut (yaitu, σ sebanding dengan x), sedangkan jauh
hilir bentuk bulu-bulu menyerupai bahwa dari paraboloid eliptik (yaitu, σ sebanding dengan x 1/2). Perubahan dalam
perumusan standar deviasi sebagai fungsi jarak dari sumber dibahas lebih lanjut dalam Bagian 6.3.2, ketika
representasi Lagrangian dan Eulerian turbulensi dibandingkan.

representasi Lagrangian ini digunakan asumsi implisit bahwa distribusi kecepatan sesaat tentang kecepatan
rata-rata adalah gaussian. Asumsi ini umumnya sesuai untuk arah horisontal; Namun, jarang veri fi ed dalam arah
vertikal karena Bumi ' s lead permukaan ke ketidaksimitrisan penting antara gerakan ke atas dan ke bawah.
gerakan ke atas umumnya terjadi di sepanjang kolom sempit, sedangkan gerakan ke bawah cenderung terjadi
dengan kecepatan yang lebih rendah dan lebih kolom yang lebih luas dari udara.

6.3 Euler Representasi Turbulensi

6.3.1 The Atmospheric Di ff usion Persamaan

Persamaan kontinuitas untuk polutan udara (partikel atau molekul) ditulis sama dengan yang ditulis untuk suhu di
Bab 4 (di sini, salah satu tidak memperhitungkan emisi, deposisi, dan transformasi):

∂C
ð 06:22 Þ
∂ xÞ þ ∂ ð vC ∂Þy þ ∂ ð toilet ∂Þz ¼ 0
∂ t þ ∂ ð UC

Persamaan ini kadang-kadang disebut sebagai persamaan difusi atmosfer. difusi molekuler tidak diperhitungkan
di sini, karena diabaikan di atmosfer dibandingkan

download dari https://www.cambridge.org/core . University of Texas Perpustakaan, pada 8 Desember 2019 di 8:58:32, tunduk pada Cambridge Inti persyaratan penggunaan, tersedia di
https://www.cambridge.org/core/terms . https://doi.org/10.1017/9781108674614.006
100 atmosfer Dispersi

untuk difusi turbulen, kecuali permukaan dekat (lihat Bab 11). Akuntansi untuk sifat bergolak dari fl ow, yang
mempengaruhi tidak hanya momentum dan suhu (lihat Bab 4), tetapi juga konsentrasi polutan:

C ¼ C þ C 0; u ¼ u þ u 0; v ¼ v þ v 0; w ¼ w þ w 0 ð 06:23 Þ

Oleh karena itu, sejak UC 0 ¼ 0; u 0 C ¼ 0; dll ::

∂C
þ ∂ ð v C þ v0 C0Þ þ ∂ ð toilet þ w 0 C 0 Þ ¼0 ð 06:24 Þ
∂ t þ ∂ ð u C þ u ∂0 C
x 0Þ ∂y ∂z

Menggunakan persamaan kontinuitas untuk momentum (lihat Persamaan 4,19) mengarah ke penyederhanaan yang fi persamaan ed:

∂C
ð 06:25 Þ
∂ t þ u ∂ C∂ x þ ∂ ð u 0 C
∂ x0 Þþ v ∂ C ∂ y0 Þþ w ∂ C
∂ y þ ∂ ð v0 C ∂ z þ ∂ ð w0 C Þ0
∂ z0 ¼

Selanjutnya, seperti dalam Bab 4, notasi berikut digunakan: C ¼ C; u ¼ u; dll: A fi Penutupan rstorder dilakukan oleh
anal ogy dengan Fick ' s hukum untuk difusi molekul (Bird et al.,
2006). Misalnya, untuk istilah u 0 C 0:

∂C ∂C ∂C
u 0 C 0 ¼? K xx ð 06:26 Þ
∂ x K xy ∂ y K xz ∂z

dimana K xx, K xy, dan K xz adalah koefisien bergolak difusi fi koefisien. Secara umum diasumsikan bahwa istilah
non-diagonal dari matriks dari koefisien bergolak difusi fi koefisien adalah nol ( K xy = K xz = 0). Kemudian:

∂C
u 0 C 0 ¼? K xx ð 06:27 Þ
∂x

istilah serupa diperoleh untuk v 0 C 0 dan w 0 C 0. Perhatikan bahwa analogi dengan difusi molekul tidak memiliki arti
fisik yang benar. difusi molekul diatur oleh gradien konsentrasi (nilai-nilai koefisien difusi molekul fi sien bervariasi
sedikit di antara spesies kimia). Dalam kasus difusi turbulen, istilah ini juga sebanding dengan gradien
konsentrasi, tetapi itu adalah sifat bergolak dari fl ow, yang mengatur pentingnya istilah ini dalam persamaan difusi
atmosfer. Syarat “ penyebaran ” biasanya digunakan sebagai sinonim dari difusi turbulen. Persamaan difusi
atmosfer kemudian ditulis sebagai berikut:

∂C ∂ ∂C ∂ ∂C ∂ ∂C
ð 06:28 Þ
∂ t þ u ∂ C∂ x ∂ xK xx ∂ x þ v ∂ C∂ y ∂ YK Y y ∂ y þ w ∂ C∂ z ∂ z K zz ∂ z ¼ 0

Menulis ulang dengan mengelompokkan adveksi dan difusi turbulen hal:

∂C ∂C ∂C ∂C
ð ∂ ∂ xK xx ð 06:29 Þ
∂ t þ ð u ∂ C∂ x þ v ∂ C∂ y þ w ∂ C∂ z Þ? ∂ x þ ∂ ∂ YK Y y ∂ y þ ∂ ∂ z K zz ∂ z Þ ¼ 0

Demi kesederhanaan, kita menganggap akhirat bahwa kecepatan dan arah angin yang konstan sepanjang x arah
(yaitu, v = w = 0), sehingga:

download dari https://www.cambridge.org/core . University of Texas Perpustakaan, pada 8 Desember 2019 di 8:58:32, tunduk pada Cambridge Inti persyaratan penggunaan, tersedia di
https://www.cambridge.org/core/terms . https://doi.org/10.1017/9781108674614.006
101 6.3 Euler Representasi Turbulensi

∂C ∂C ∂C ∂C
ð∂ ð 06:30 Þ
∂ t þ u ∂ C∂ x ∂ xK xx ∂ x þ ∂ ∂ YK Y y ∂ y þ ∂ ∂ z K zz ∂zÞ¼0

6.3.2 sesaat Titik Sumber

Studi kasus dari sumber titik sesaat, yang digunakan untuk representasi Lagrangian, juga digunakan di sini fi pertama
untuk representasi Eulerian. Oleh karena itu, kondisi awal dan batas adalah sebagai berikut:

C ð x; y; z; 0 Þ ¼ S 0 δ ð xx s Þ δ ð Y y s Þ δ ð z zs Þ
ð 06:31 Þ
C ð x; y; z; t Þ ¼ 0 untuk x; y; dan z → ∞

dimana S 0 adalah massa (g) yang dipancarkan dari sumber, yang terletak di ( x s, y s, z s), dan δ adalah fungsi Dirac.
Misalnya, untuk δ ( x - x s):

δ ð xx s Þ ¼ 0 untuk x ≠ x s

δ ð xx s Þ ¼ þ ∞ untuk x ¼ x s
þð∞ ð 06:32 Þ

δ ð xx s Þ d x ¼ 1

Solusi dari persamaan kontinuitas adalah konsentrasi dalam kepulan dipancarkan oleh sumber:
! ð 06:33 Þ
S0 ð xx s Þ 2 ð Y ys Þ2 ðz zs Þ2
C ð x; y; z; t Þ ¼ p exp
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
8 π32t32 K xx K Y y K zz 4 K xx t 4 KYy t 4 K zz t

Persamaan ini menunjukkan distribusi gaussian dari konsentrasi. Memperkenalkan standar deviasi dari distribusi
gaussian, σ x, σ y, dan σ z:
!
S0 ð xx s Þ 2 ð Y ys Þ2 ðz zs Þ2
C ð x; y; z; t Þ ¼ exp ð 06:34 Þ
ð 2 π Þ32σ x σ y σ z 2 σx2 2 σy2 2 σz2

Sebuah perbandingan sederhana antara Persamaan 6.33 dan 6.34 menunjukkan bahwa Eulerian bergolak difusi
koe fi koefisien terkait dengan standar deviasi dari distribusi konsentrasi dengan hubungan berikut:

p
2 K xx t
σ x ¼ ffiffiffiffiffiffiffiffiffiffiffi
p
2 KYy t
σ y ¼ ffiffiffiffiffiffiffiffiffiffiffi ð 06:35 Þ
p
2 K zz t
σ z ¼ ffiffiffiffiffiffiffiffiffiffi

Dalam formulasi Euler ini, standar deviasi yang sebanding dengan akar kuadrat dari waktu yang telah berlalu
sejak emisi (atau akar kuadrat dari jarak yang ditempuh dari sumber jika mean kecepatan angin adalah konstan).
Oleh karena itu, hubungan ini adalah sama dengan yang diperoleh dengan representasi Lagrangian untuk kasus
ketika salah satu jauh melawan arah angin dari sumber (Persamaan 6.21). Di sisi lain, di dekat sumber,
representasi Lagrangian menyebabkan deviasi standar yang sebanding dengan waktu yang telah berlalu sejak
emisi atau jarak yang ditempuh dari sumber (Persamaan 6.14).

download dari https://www.cambridge.org/core . University of Texas Perpustakaan, pada 8 Desember 2019 di 8:58:32, tunduk pada Cambridge Inti persyaratan penggunaan, tersedia di
https://www.cambridge.org/core/terms . https://doi.org/10.1017/9781108674614.006
102 atmosfer Dispersi

dispersi Eulerian

σ=2Kt

Log (Standard deviation of the concentration distribution)


rezim menengah

Dispersi puff dengan


dispersi
pusaran turbulen
Lagrangian
awal
σαt

Log (Jarak dari sumber)

Gambar 6.1. Skema representasi dari evolusi standar deviasi konsentrasi dalam pu ff sebagai fungsi jarak dari sumber.

Perubahan tingkat dispersi dengan jarak dari sumber dapat dijelaskan secara konseptual sebagai berikut.

Dekat sumber, puff polutan sangat kecil. Relatif dispersi dari hasil engah hanya dari pusaran turbulen yang
lebih kecil dari puff. Sebagai puff tumbuh dalam ukuran karena dispersi, ada lebih dan lebih pusaran yang lebih
kecil dari puff. Lebih jauh melawan arah angin, ketika ukuran kepulan cukup besar untuk menutupi seluruh
spektrum ukuran eddy, menstabilkan dispersi engah dan rumus Eulerian berlaku. Ada rezim sementara antara
dispersi dekat-sumber, yang diwakili oleh dispersi Lagrangian dekat-sumber, dan dispersi jauh-melawan arah
angin, diwakili oleh dispersi Eulerian (serta rumus Lagrangian untuk t » t L), dimana standar deviasi sebanding
dengan t α di mana 0,5 < α < 1. Perubahan ini dari fungsi yang mengatur standar deviasi dari distribusi konsentrasi
engah secara skematis diperlihatkan pada Gambar 6.1.

6.3.3 berkelanjutan Titik Sumber

Untuk sumber titik terus menerus dalam kondisi stasioner dan uniformmeteorological (yaitu, dengan kecepatan
angin konstan dan arah), konsentrasi steady-state dari polutan di bulu-bulu dihitung dengan persamaan
konservasi massa berikut:

∂ ∂C ∂ ∂C ∂ ∂C
u∂C zs Þ ð 06:36 Þ
∂x ∂ xK xx ∂ x ∂ YK Y y ∂ y ∂ z K zz ∂ z ¼ S δ ð xx s Þ δ ð Y y s Þ δ ð z

download dari https://www.cambridge.org/core . University of Texas Perpustakaan, pada 8 Desember 2019 di 8:58:32, tunduk pada Cambridge Inti persyaratan penggunaan, tersedia di
https://www.cambridge.org/core/terms . https://doi.org/10.1017/9781108674614.006
103 6.4 Model Lagrangian dari Atmospheric Dispersi

dimana S adalah tingkat emisi sumber (gs - 1). Turbulensi di x arah dapat diabaikan dengan asumsi bahwa
transportasi atmosfer arah itu didominasi oleh angin rata-rata; ini disebut ramping membanggakan pendekatan ( K
xx = 0):

∂ ∂C ∂ ∂C
u∂C zs Þ ð 06:37 Þ
∂x ∂ YK Y y ∂ y ∂ z K zz ∂ z ¼ S δ ð xx s Þ δ ð Y y s Þ δ ð z

Solusinya adalah sebagai berikut:


!
S ð Y ys Þ2 ðz zs Þ2
C ð x; y; z Þ ¼ p exp
ffiffiffiffiffiffiffiffiffiffiffiffiffi ð 06:38 Þ
4 π ut K Y y K zz 4 KYy t 4 K zz t

dimana t = x / u. Persamaan ini dapat ditulis sesuai dengan bentuk gaussian nya:
!
S ð Y ys Þ2 ðz zs Þ2
C ð x; y; z Þ ¼ exp ð 06:39 Þ
2 π u σyσz 2 σy2 2 σz2

Dengan demikian, konsentrasi segumpal gaussian dipancarkan dari sumber yang berkelanjutan berbanding terbalik dengan

kecepatan angin.

6.4 Model Lagrangian dari Atmospheric Dispersi

6.4.1 Pertimbangan Umum

dispersi atmosfer dapat dimodelkan menggunakan pendekatan Lagrangian, yaitu, sehubungan dengan sumber,
atau menggunakan pendekatan Euler, yaitu, tanpa referensi khusus pada lokasi sumber. Dalam kasus yang
pertama, model harus memperhitungkan jarak yang ditempuh oleh polutan dari sumber (atau waktu berlalu sejak
emisi) dan kondisi meteorologi dan fisiografi con fi gurasi, yang mempengaruhi turbulensi. Dalam kasus terakhir,
hanya kondisi meteorologi dan fisiografi con fi gurasi diperhitungkan.

Model dispersi Lagrangian sering disebut sebagai model gaussian, karena distribusi gaussian konsentrasi
sekitar puff atau bulu-bulu pusat diasumsikan. Jika kondisi meteorologi diasumsikan stasioner dan seragam
(yaitu, konstan dalam ruang dan waktu), solusi analitis dapat diperoleh, yang identik dengan yang diperoleh
untuk persamaan Euler:

!
y2 ðz zs Þ2
C ð x; y; z Þ ¼ S exp ð 06:40 Þ
2 π u σy σz 2 σy2 2 σz2

dimana z s adalah ketinggian sumber, sumber yang terletak di (0, 0, z s), dan dispersi sepanjang arah angin
diasumsikan diabaikan (ramping membanggakan pendekatan). Kemudian, konsentrasi di tengah bulu-bulu diperoleh
untuk y = 0 dan z = z s:

download dari https://www.cambridge.org/core . University of Texas Perpustakaan, pada 8 Desember 2019 di 8:58:32, tunduk pada Cambridge Inti persyaratan penggunaan, tersedia di
https://www.cambridge.org/core/terms . https://doi.org/10.1017/9781108674614.006
104 atmosfer Dispersi

C ð x; y; z Þ ¼ S ð 06:41 Þ
2 π u σy σz

Dekat sumber, pendekatan Lagrangian mengarah ke standar deviasi yang sebanding dengan jarak dari sumber
(Persamaan 6.14):

σy ð x Þ α t ¼ x ð 06:42 Þ
u

dan juga untuk σ z. Oleh karena itu, konsentrasi di tengah bulu-bulu terkait dengan jarak dari sumber sebagai
berikut:

C ð x; y; z Þ α S ð 06:43 Þ
x2

Konsentrasi maksimum dalam bulu-bulu berkurang awalnya sebagai kebalikan dari kuadrat jarak dari sumber.

6.4.2 Plume Naik

Nilai tinggi membanggakan awal tidak hanya tergantung pada ketinggian sumber, tetapi juga pada kenaikan membanggakan
sehubungan dengan sumber itu. Jika ef yang fl uent pada sumber memiliki beberapa kecepatan awal, momentum ini
mengarah membanggakan untuk naik di atas sumber. Juga, jika suhu membanggakan lebih besar dari suhu lingkungan,
bulu-bulu naik karena daya apung, hingga beberapa ketinggian di atas sumber mana mencapai kesetimbangan dengan
lingkungan sekitar. Perhitungan kenaikan membanggakan, Δ z s, dapat dilakukan dengan menggunakan rumus semi-empiris.
Rumus CONCAWE sederhana:

h
Δ z s ¼ 2:71 Q 0: 5 ð 06:44 Þ
u 0:75

dimana Q h adalah tingkat emisi panas (J s - 1). Tingkat emisi panas ini dapat dihitung berdasarkan jumlah gas yang
dipancarkan dan suhu mereka:

Qh¼ nsCPΔ T ð 06:45 Þ

dimana n s adalah tingkat emisi di mol yang dipancarkan per satuan waktu (mol s - 1), C P adalah rata-rata kapasitas panas
molar dari gas yang dikeluarkan (J mol - 1 K - 1) dan Δ T = T s - T ( K) adalah perbedaan antara ef fl suhu uent dan suhu
lingkungan. Rumus CONCAWE ini memperhitungkan kenaikan membanggakan karena daya apung membanggakan
(yaitu, gerak vertikal yang membanggakan karena perbedaan suhu antara bulu-bulu dan lingkungannya), tetapi tidak
memperhitungkan momentum membanggakan awal.

The Briggs rumus (1984) lebih lengkap karena memperhitungkan baik efek: apung dan kinetik awal energi ef fl uent:

Δ z s ¼ ð 3 F Sebuah x
00:36 u 2 þ 04:17 F b x 2 u3Þ1=3

ds2
F Sebuah ¼ T v 2 saya ð 06:46 Þ
Ts 4
ds2 TT
F b ¼ gv saya Þ
4 ð Ts s

download dari https://www.cambridge.org/core . University of Texas Perpustakaan, pada 8 Desember 2019 di 8:58:32, tunduk pada Cambridge Inti persyaratan penggunaan, tersedia di
https://www.cambridge.org/core/terms . https://doi.org/10.1017/9781108674614.006
105 6.4 Model Lagrangian dari Atmospheric Dispersi

dimana d s adalah diameter stack (m), T s dan T adalah suhu ef fl uent pada sumber dan suhu lingkungan (K),
masing-masing, dan v saya adalah kecepatan vertikal awal ef fl uent (yaitu, plume atau engah) pada sumbernya (ms -
1), biasanya di pintu keluar stack. Rumus ini berlaku sampai dengan jarak x t, yang de fi didefinisikan sebagai berikut:

x t ¼ 49 F 5 = 8 b
untuk F b ≤ 55m 4 s - 3
ð 06:47 Þ

x t ¼ 119 F 2 = 5 b
untuk F b ≥ 55m 4 s - 3

Itu fi tinggi nal dari bulu-bulu di atas permukaan tanah disebut ketinggian sumber yang efektif dan sesuai dengan fi nilai nal
dari ketinggian membanggakan, z s, f = z s + Δ z s.
formula yang lebih rinci yang menjelaskan stabilitas atmosfer telah dikembangkan. Mereka tidak diberikan di sini,
tetapi tersedia dalam literatur. Misalnya, Carson dan Musa (1969) telah membandingkan beberapa rumus kenaikan
membanggakan dengan data eksperimen yang tersedia di lokasi industri. Ini rumus kenaikan membanggakan
dikembangkan awalnya untuk model dispersi gaussian. Mereka sekarang digunakan juga dalam model lain, seperti
model kualitas udara Eulerian.

6.4.3 Model Gaussian Operasional Sumber Titik

Sebuah bulu yang dipancarkan ke atmosfer masuk ke kontak dengan tanah di beberapa arah angin jarak sumber
(segera jika sumber di permukaan tanah, z s = 0). Kemudian, solusi untuk konsentrasi membanggakan harus
memperhitungkan fakta bahwa konsentrasi tidak dapat dihitung untuk z < 0. Secara umum diasumsikan bahwa
bulu-bulu sepenuhnya re fl ected di tanah. Kemudian, solusinya diperoleh dengan menggunakan sumber virtual
yang simetris dari sumber yang sebenarnya sehubungan dengan tanah untuk mewakili re ini fl cerminan dari
bulu-bulu di tanah. Solusi untuk konsentrasi membanggakan adalah sebagai berikut:

! þ exp !!
y2 ðz z s; f Þ 2 ð z þ z s; f Þ 2
C ð x; y; z; t Þ ¼ S exp exp
2 π u σyσz 2 σy2 2 σz2 2 σz2

ð 06:48 Þ

Demikian pula, jika ketinggian emisi di bawah inversi suhu, mungkin ada re fl cerminan dari bulu-bulu terhadap dasar
lapisan inversi ini. Konsep sumber virtual juga dapat digunakan untuk akun untuk re membanggakan ini fl ection
tinggi-tinggi. Sumber Avirtual yang simetris dari sumber yang sebenarnya sehubungan dengan dasar inversi
digunakan untuk mewakili re bulu-bulu fl ection terhadap dasar lapisan inversi tinggi z saya permukaan tanah di atas:

! þ exp !!
y2 ðz z s; f Þ 2 ð z þ z s; f 2 z saya Þ 2
C ð x; y; z; t Þ ¼ S exp exp
2 π u σyσz 2 σy2 2 σz2 2 σz2

ð 06:49 Þ

Kedua persamaan di atas dapat dikombinasikan ke akun untuk kedua re fl cerminan dari bulu-bulu di tanah dan melawan

dasar dari tinggi-tinggi lapisan inversi. Kemudian, salah satu keharusan di samping

download dari https://www.cambridge.org/core . University of Texas Perpustakaan, pada 8 Desember 2019 di 8:58:32, tunduk pada Cambridge Inti persyaratan penggunaan, tersedia di
https://www.cambridge.org/core/terms . https://doi.org/10.1017/9781108674614.006
106 atmosfer Dispersi

Fanning Stabil

coning
Netral

lofting Netral Stabil


tinggi-tinggi di bawah ini

Stabil tinggi-tinggi
mengasapi tidak stabil di bawah

looping labil

dispersi Plume profil vertikal temperatur potensial

Gambar 6.2. pro khas fi les dispersi vertikal bulu dalam berbagai kondisi atmosfer.

account untuk re fl ections dari bulu sesuai dengan sumber virtual (misalnya, re fl ection di permukaan tanah dari
bulu-bulu dari sumber virtual yang digunakan untuk mewakili re fl ection terhadap dasar lapisan inversi). Oleh karena
itu, solusinya mencakup serangkaian istilah yang mewakili orang-orang berbagai re fl ections:

!
y2 X þ Nr ðz z s; f þ 2 N k z saya Þ 2
C ð x; y; z; t Þ ¼ S exp exp
2 π u σyσz 2 σy2 2 σz2
Nr
!!
ð z þ z s; f 2 N k z saya Þ 2
þ exp ð 06:50 Þ
2 σz2

dimana N r adalah jumlah re fl ections ( N k = 1,. . ., N r). Sebuah solusi yang memuaskan diperoleh dengan re tunggal fl ection
( N r = 1) dan fi ve istilah ( N r = 5) adalah suf fi sien untuk mendapatkan solusi yang akurat. Formulasi ini berguna untuk
mewakili berbagai jenis dispersi atmosfer bulu di PBL tergantung pada berbagai kondisi stabilitas dan kehadiran
lapisan inversi. Gambar 6.2 menggambarkan jenis utama dari dispersi vertikal bulu yang dipancarkan dari
tumpukan tinggi.

penyimpangan Bila menggunakan formulasi gaussian Lagrangian, standar σ y dan σ z


harus de fi ned. Mereka biasanya de fi ned secara empiris menggunakan data dari fi eksperimen lapangan. data tersebut
sesuai biasanya untuk rata-rata kali dari 5 sampai 15 menit; Oleh karena itu, mereka termasuk beberapa berkelok-kelok
membanggakan selain difusi membanggakan relatif. Ini standar deviasi umumnya disebut dispersi koe fi koefisien. Di
antara koefisien dispersi Gaussian fi koefisien yang paling banyak digunakan, salah satu mungkin mengutip orang-orang
dari Pasquill - Gifford dan orang-orang dari McElroy - Pooler. The Pasquill-Gifford koe fi koefisien disajikan pada Gambar
6.3. Mereka juga dilaporkan sebagai rumus matematika analitis untuk penggunaan langsung dalam pemodelan numerik.
Tabel 6.1 daftar rumus analitis untuk Pasquill-Gifford dan

download dari https://www.cambridge.org/core . University of Texas Perpustakaan, pada 8 Desember 2019 di 8:58:32, tunduk pada Cambridge Inti persyaratan penggunaan, tersedia di
https://www.cambridge.org/core/terms . https://doi.org/10.1017/9781108674614.006
107 6.4 Model Lagrangian dari Atmospheric Dispersi

SEBUAH

10 5

10 4

SM

σ z ( m)
10 3

2 DEF

10 10

0,1 1 10

x ( km)

10 3

10 2
σ y ( m)

DEF

10

0,1 1 10 AB

x ( km)

Gambar 6.3. Empiris coe dispersi FFI koefisien dari Pasquill-Gi ff ord sebagai fungsi jarak dari sumber untuk di ff erent kelas stabilitas atmosfer (mulai dari A, sangat

tidak stabil, untuk F, sangat stabil). coe ini FFI koefisien dikembangkan oleh Pasquill menggunakan data yang dilaporkan oleh Gi ff ord (1961). Vertikal

coe dispersi FFI koefisien, σ z ( teratas fi angka), dan coe dispersi horisontal FFI koefisien, σ y ( bawah fi angka).

McElroy-Pooler koe fi koefisien. rumus empiris lainnya telah dikembangkan untuk dispersi koe fi koefisien, seperti yang
dari Briggs dan orang-orang dari Doury (yang terakhir berlaku untuk jarak yang lebih jauh daripada yang berlaku
untuk formula lain), serta orang-orang berdasarkan teori kesamaan (Irwin dan Venkatram) dan mereka dikembangkan
untuk tumpukan tinggi (Hanna dan Paine , Gillani dan Godowitch).

Asumsi kondisi atmosfer stasioner dan seragam, yang diperlukan bila menggunakan rumus membanggakan
gaussian, dapat dihapus dengan menggunakan sebuah ensemble dari puff bukan segumpal tunggal. Tingkat
emisi terus menerus kemudian discretized

download dari https://www.cambridge.org/core . University of Texas Perpustakaan, pada 8 Desember 2019 di 8:58:32, tunduk pada Cambridge Inti persyaratan penggunaan, tersedia di
https://www.cambridge.org/core/terms . https://doi.org/10.1017/9781108674614.006
108 Atmospheric Dispersion

Table 6.1. Analytical formulas for the dispersion coe ffi cients of Pasquill-Gi ff ord and Briggs-McElroy-Pooler,
σ y and σ z ( m), as a function of distance from the source, x ( m), and atmospheric stability (A to F).

Pasquill-Gifford

Stability σ y = exp ( a y + b y ln( x) + c y ( ln( x)) 2) σ z = exp ( a z + b z ln( x) + c z ( ln( x)) 2)

ay by cy az bz cz
A − 1.104 0.9878 − 0.0076 4.679 − 1.172 0.2770
B − 1.634 1.0350 − 0.0096 − 1.999 0.8752 0.0136
C − 2.054 1.0231 − 0.0076 − 2.341 0.9477 − 0.0020
D − 2.555 1.0423 − 0.0087 − 3.186 1.1737 − 0.0316
E − 2.754 1.0106 − 0.0064 − 3.783 1.3010 − 0.0450
F − 3.143 1.0148 − 0.0070 − 4.490 1.4024 − 0.0540

Briggs-McElroy-Pooler

Stability σy σz

Rural conditions A
0.22 x ( 1 + 0.0001 x) − 1/2 0.20 x

B 0.16 x ( 1 + 0.0001 x) − 1/2 0.12 x

C 0.11 x ( 1 + 0.0001 x) − 1/2 0.08 x ( 1 + 0.0002 x) − 1/2


D 0.08 x ( 1 + 0.0001 x) − 1/2 0.06 x ( 1 + 0.0015 x) − 1/2
E 0.06 x ( 1 + 0.0001 x) − 1/2 0.03 x ( 1 + 0.0003 x) − 1
F 0.04 x ( 1 + 0.0001 x) − 1/2 0.016 x ( 1 + 0.0003 x) − 1
Urban conditions A &
B 0.32 x ( 1 + 0.0004 x) − 1/2 0.24 x ( 1 + 0.001 x) 1/2
C 0.22 x ( 1 + 0.0004 x) − 1/2 0.20 x

D 0.16 x ( 1 + 0.0004 x) − 1/2 0.14 x ( 1 + 0.0003 x) − 1/2


E&F 0.11 x ( 1 + 0.0004 x) − 1/2 0.08 x ( 1 + 0.0015 x) − 1/2

terhadap waktu dan setiap tingkat emisi discretized ditugaskan untuk puff sebagai tingkat emisi seketika. Puffs
yang dipancarkan pada berturut-turut (umumnya, tetapi tidak harus, konstan) langkah waktu. Jadi, jika arah
angin bervariasi dalam waktu dan / atau ruang, bulu-bulu diwakili oleh sebuah ensemble dari puff dan tidak akan
mengikuti lintasan lurus. Oleh karena itu, model kepulan memberikan representasi yang lebih realistis dari
transportasi dan dispersi atau udara polutan yang dipancarkan dari sumber dari model membanggakan standar,
ketika kondisi meteorologi tidak stasioner dan seragam. Formulasi gaussian dapat digunakan dalam model
engah. Namun, karena masing-masing engah dipengaruhi oleh kondisi sendiri atmosfer (kecepatan dan arah
angin, stabilitas atmosfer), bulu-bulu, yang diwakili oleh sebuah ensemble dari puff, tidak gaussian. Ini
representasi yang lebih realistis dispersi membanggakan melibatkan beban komputasi yang lebih besar, seperti
yang diperlukan untuk menghitung transportasi dan dispersi dari sebuah ensemble puff. Ada ada beberapa
model kepulan Lagrangian operasional. Beberapa yang paling umum digunakan termasuk SCIPUFF dan versi
dengan SCICHEM kimia atmosfer, HYSPLIT, dan FLEXPART.

Downloaded from https://www.cambridge.org/core . University of Texas Libraries, on 08 Dec 2019 at 08:58:32, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms . https://doi.org/10.1017/9781108674614.006
109 6.4 Lagrangian Models of Atmospheric Dispersion

35

NO (no turbulence) NO
(turbulence) NO
r = 0.97

NO Concentration (ppb)
(observed)

(a)

0 2 4 6 8 10 12 14 16 18 20 22 24 26

80

NOy (model)
60 NOy (observed)
NO y Concentration (ppb)

r = 0.97

40

20 (b)

0
0 2 4 6 8 10 12 14 16 18 20 22 24 26

10
10 5 0
8 SO2 (model) SO2
(observed)
SO 2 Concentration (ppb)

r = 0.85
6 25 20 15
30

(c)
2

0
0 2 4 6 8 10 12 14 16 Crosswind distance18 20 22 24 26

(km)

Figure 6.4. Air pollutant concentration pro fi les of a coal- fi red power plant plume. Measured values are represented by circles and simulated values are

represented by solid lines. The simulation represented by a dotted line corresponds to the case where the in fl uence of atmospheric turbulence

on the chemical kinetics of pollutant reactions was taken into account. Source: Reproduced with permission from Karamchandani, P. et al.

(2000).

The gaussian formulation for plume dispersion is theoretical. Nevertheless, it is a good representation of the
atmosphere when conditions are relatively stationary. Figure 6.4 shows a comparison of measured and simulated
concentrations of nitrogen oxides (nitric oxide, NO, and the ensemble of nitrogen oxides, NO y, see Chapter 8 for
the de fi nition of NO y) and sulfur dioxide (SO 2) in the plume of a coal- fi red power plant. The measured
concentrations were obtained with an instrumented aircraft fl ying across the plume. The simulated concentrations
were obtained with the SCICHEM puff model. Although the measured concentration pro fi les are not exactly
gaussian, the model simulation reproduces satisfactorily the observed concentrations.

Figure 6.5 shows the results of ground-level concentrations calculated with a gaussian plume model for a
point source (stack). In these simulations, the SO 2 emission rate is 5 g s − 1. In the top fi gure, the stack height ( z s) is
5 m; it is 30 m in the bottom fi gure. In each

Downloaded from https://www.cambridge.org/core . University of Texas Libraries, on 08 Dec 2019 at 08:58:32, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms . https://doi.org/10.1017/9781108674614.006
110 Atmospheric Dispersion

zs= 5 m

4000

Neutral, Ts = 25 ˚C

Neutral, Ts = 50 ˚C
Concentration ( µ g m − 3) 3000

2000
Unstable, Ts = 25 ˚C

Unstable, Ts = 50 ˚C

1000

0 0.2 0.4 0.6 0.8 1

Distance (km)

z s = 30 m

Unstable, Ts = 25 ˚C
400
Unstable, Ts = 50 ˚C
Concentration ( µ g m − 3)

300

Neutral, Ts = 25 ˚C

Neutral, Ts = 50 ˚C
200

100

0 0.2 0.4 0.6 0.8 1

Distance (km)

Figure 6.5. Simulation of SO 2 ground-level concentrations ( μ g m − 3) with a gaussian plume model. Top fi gure: stack height = 5 m; bottom fi gure:

stack height = 30 m.

case, four scenarios were simulated: two separate atmospheric conditions (unstable and neutral) and two initial
plume temperatures (25 and 50 °C, with an ambient temperature of 20 °C).

When the stack is only 5 m high, the largest concentrations are obtained under neutral conditions. Although
the wind speed is typically greater under neutral and stable conditions compared to unstable conditions (here 2 m
s − 1 for unstable conditions and 4 m s − 1 for

Downloaded from https://www.cambridge.org/core . University of Texas Libraries, on 08 Dec 2019 at 08:58:32, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms . https://doi.org/10.1017/9781108674614.006
111 6.4 Lagrangian Models of Atmospheric Dispersion

neutral conditions), the dispersion coef fi cients are much lower for neutral conditions than for unstable conditions,
thereby leading to a more concentrated plume. Therefore, the ground-level plume concentrations are greater. If
the ef fl uent temperature increases, the plume rise (4 m for neutral conditions and 7 m for unstable conditions)
leads to lower ground-level concentrations near the source. The SO 2 hourly concentration standard is about 200 μ
g m − 3 ( 75 ppb) in the U.S. and 350 μ g m − 3 in France (see Chapter 15). Thus, these air quality standards are
exceeded for all conditions.

When the stack is 30 m high, the largest concentrations are obtained under unstable conditions. Under neutral
conditions, the plume remains aloft and shows ground-level impacts only beyond 100 m with a maximum
concentration between 400 and 500 m downwind. On the other hand, unstable conditions are conducive to
vertical dispersion,
i.e., they dilute the plume concentrations, but mix the plume to the ground at a short distance from the source. An
increase in the ef fl uent temperature leads to the same plume rise as above and minimizes the ground-level
impacts. Then, the U.S. air quality standard is exceeded only under unstable conditions and the French air
quality standard is not exceeded if a high ef fl uent temperature is used.

In summary, stable and neutral conditions lead to the largest ground-level impacts for short stacks, whereas
unstable conditions lead to the largest ground-level impacts for tall stacks (see Figure 6.2). Furthermore, plume rise
may reduce the ground-level impacts signi fi cantly. Plume rise may be obtained either by increasing the ef fl uent
initial velocity (i.e., increased momentum) or by increasing the ef fl uent temperature (i.e., increased buoyancy). This
type of analysis led to an air quality management approach based on the reasoning that “ the solution to pollution is
dilution. ” Thus, many large industrial sources were given tall stacks. Some of those stacks can be as high as 300 m
above ground level. Also, under some con fi gurations (stacks located near a building for example), minimum stack
height requirements are used to minimize ground-level impacts (for example, “ good engineering practice ” in the
U.S.). However, this approach pertains only to primary pollutants, i.e., those pollutants emitted from the stack. It is
not relevant to secondary pollutants, such as acid rain, ozone, and fi ne particles, because these secondary
pollutants are associated with a regional pollution, which does not depend strongly on stack height. Therefore, in
the 1990s, emission controls started to be required for some industrial sources that had tall stacks, but contributed
to regional pollution. These emission controls of tall stacks were driven fi rst by acid rain regulations (see Chapters
10 and 13) and later by ozone regulations (see Chapter 8).

6.4.4 Operational Gaussian Models for Line Sources

The gaussian equation presented in Section 6.4.3may be applied to non-point sources, i.e., line, area, and volume
sources. However, the integration over a line, area or volume rarely leads to an analytical solution. An exception is
the integration over a line source when the wind is perpendicular to the line source:

! ! ! ð 6:51 Þ
z2 ð y y1 Þ ð y y2 Þ
C ð x; y; z Þ ¼ S l p u σz exp erf p σffiffiffi
y erf p σffiffiffi
y

2 ffiffiffiffiffi 2 σz2 2 2

Downloaded from https://www.cambridge.org/core . University of Texas Libraries, on 08 Dec 2019 at 08:58:32, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms . https://doi.org/10.1017/9781108674614.006
112 Atmospheric Dispersion

where S l is the line source emission rate (g km − 1 s − 1), y 1 and y 2 are the coordinates of the extremities of the line
source and erf is the error function. For other wind directions, it is necessary to either use an approximate
analytical solution or to obtain a numerical solution. A numerical solution is usually computationally expensive
and, therefore, analytical solutions, even if they are approximate, are preferred. The HorstVenkatram (HV) model
is a good example of an approximate solution of the gaussian equation for a line source (Venkatram and Horst,
2006). It was later improved by Briant et al. (2011) (BKS), who minimized the errors due to the use of analytical
functions and combined it with a numerical solution for cases where the wind is almost parallel to the line source
(the error of the analytical solution becomes too large for such wind/line source con fi gurations, see the
discussion of Equation 6.52).

The analytical form of the HV-BKS model is as follows:

Sl z2
C ð x; y; z Þ ¼ p u cos ð α Þ σ z ð d eff Þ exp

2 ffiffiffiffiffi 2 σ z 2 ð d eff Þ
! !!
ð y y 1 Þ cos ð α Þ x sin ð α Þ ð y y 2 Þ cos ð α Þ x sin ð α Þ
erf p σffiffiffi
y ð d1 Þ erf p σffiffiffi
y ð d2 Þ
2 2

1
ð 6:52 Þ
L BKS ð x wind Þ þ 1 þ E BKS ð x wind; y wind; z Þ

where α is the angle between the wind direction and the direction perpendicular to the line source, x wind and y wind are
the receptor coordinates in the wind direction coordinate system, d eff = x/ cos( α), and d i = ( x – x i) cos( α) + ( y – y i) sin( α), i
= 1 or 2. The fi rst part of the formula corresponds to the HV model and L BKS and E BKS are the analytical functions
added in the BKS version of the model to further minimize the error. When the wind direction becomes parallel to
that of the line source, α tends toward 90 ° and cos( α) tends toward 0, so that C tends to in fi nity. Then, a
numerical solution is used instead. It consists of discretizing the line source with an ensemble of point sources
when the wind direction is within 10 ° of the direction of the line source. This type of formula may be used, for
example, to assess the air quality impacts of on-road traf fi c emissions.

The application of such models to on-road traf fi c showed that air pollution due to on-road traf fi c decreases
rapidly with the distance from the roadway. At about 200 to 300 m from the roadway, the air pollution due to the
local traf fi c has decreased enough that the air pollution is commensurate with the background air pollution.
Results of simulations conducted with such models are in reasonable agreement with measurements of air
pollution conducted in the vicinity of roadways. The modeling of the local impact of on-road traf fi c on air pollution
is discussed in greater detail by Venkatram and Schulte (2018).

6.4.5 Operational Gaussian Models for Area Sources

Two main approaches are used to simulate atmospheric dispersion from area sources. The lagrangian
atmospheric dispersion equation may be integrated numerically (for

Downloaded from https://www.cambridge.org/core . University of Texas Libraries, on 08 Dec 2019 at 08:58:32, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms . https://doi.org/10.1017/9781108674614.006
113 6.4 Lagrangian Models of Atmospheric Dispersion

example, using Romberg ’ s method). This approach has been used to take into account the width of a roadway,
for example. Then, the formula for a line source presented in Section 6.4.4 is integrated along the width of the
roadway. On the other hand, one may use a virtual point source located at a distance upwind of the area source
such that the plume has the width of the area source over the area source. This approach is computationally
cheaper than the former one because no numerical integration is needed. It presents, however, two
shortcomings: (1) the location of the virtual source and the width of the plume over the area source depend on
wind speed and atmospheric stability (initial source dimensions may be used instead) and (2) the concentration
pro fi le over the source area is gaussian. Nevertheless, this latter approach is widely used, for example for
industrial sources, because of its low computational cost (EPA, 1995).

6.4.6 Operational Gaussian Models for Volume Sources

The most commonly used approach for volume sources is similar to the latter one mentioned for area sources.
Thus, a virtual source is located at a distance upwind of the volume source such that the plume has the
dimensions of the volume source when it reaches that source. This approach has been used, for example, to
model fugitive emissions of volatile organic compounds from re fi neries (Kim et al., 2014).

6.4.7 Gaussian Models with Removal of Air Pollutants in the Plume

The concentrations of an air pollutant may decrease in a plume (or a puff), because of chemical reactions,
scavenging by rain or dry deposition on surfaces. In the case where simple parameterizations are used to model
such processes, it may be possible to take them into account explicitly in the operational formulas already
presented. The addition of these removal terms is presented here in the case of a point source, but the formulas
for other source types can be modi fi ed similarly.

Chemical kinetics is treated in the following chapters. Here, one considers the case where the air pollutant
reacts with a chemical species of constant concentration (for example, an oxidant). Then, its concentration
decreases with time according to an exponential function (see Chapters 7 and 8):

C ð t Þ ¼ C 0 exp ð k ½ X t Þ ¼ C 0 exp ð k 0 t Þ ð 6:53 Þ

where C 0 is the initial concentration, k is the rate constant of the chemical reaction, [X] is the oxidant
concentration, and k ’ is the pseudo- fi rst-order rate constant expressed in s − 1

( k ’ = k [ X]). The solution of the gaussian equation taking into account this removal by chemical reaction is as
follows:

! þ exp !!
y2 ðz z s;f Þ 2 ð z þ z s;f Þ 2
C ð x; y; z Þ ¼ S exp exp exp ð k 0 t Þ
2 π u σyσz 2 σy2 2 σz2 2 σz2

ð 6:54 Þ

Downloaded from https://www.cambridge.org/core . University of Texas Libraries, on 08 Dec 2019 at 08:58:32, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms . https://doi.org/10.1017/9781108674614.006
114 Atmospheric Dispersion

where t = x/u.
Scavenging by rain (wet deposition) may be parameterized empirically using an exponential function (see
Chapter 11):

C ð t Þ ¼ C 0 exp ð Λ t Þ ð 6:55 Þ

where Λ is the scavenging coef fi cient expressed in s − 1. The solution is similar to that obtained for the removal by
chemical reaction:
! þ exp !!
y2 ðz z s;f Þ 2 ð z þ z s;f Þ 2
C ð x; y; z Þ ¼ S exp exp exp ð Λ t Þ
2 π u σyσz 2 σy2 2 σz2 2 σz2

ð 6:56 Þ

where t = x/u.
Removal by dry deposition is usually parameterized with a deposition velocity, v d,
expressed in m s − 1 ( see Chapter 11). The solution is more complicated because dry deposition only affects the air
pollutants that are in contact with the surface (chemical reaction and rain scavenging affect air pollutants in the
entire plume). In other words, a solution that removes the air pollutants uniformly throughout the plume is
inappropriate here. A suitable solution is (Overcamp, 1976):

! þ α d exp !!
y2 ðz z s;f Þ 2 ð z þ z s;f Þ 2
C ð x; y; z Þ ¼ S exp exp
2 π u σyσz 2 σy2 2 σz2 2 σz2

ð 6:57 Þ

where α d is a function of the pollutant dry deposition velocity:

2 vd
αd¼ 1 ð 6:58 Þ
v d þ ð u z s;f σ z 1 Þ d σ z
dx

If the dry deposition velocity is zero, then, α d = 1, and the solution becomes identical to the standard gaussian
formula. On the other hand, if the dry deposition velocity tends toward in fi nity, then, α d tends toward − 1 and the
surface acts as an irreversible sink for the pollutant in the plume.

6.4.8 Atmospheric Stability Categories

The parameterizations of the dispersion coef fi cients σ y and σ z that are used in operational gaussian plume and
puff models depend on atmospheric stability. Therefore, it is useful to be able to determine the atmospheric
stability category from simple meteorological data. Table 6.2 provides information to determine the atmospheric
stability category given data on wind speed, solar radiation (during daytime) and cloudiness (at night). Table 6.3
provides information to determine the atmospheric stability categories given the vertical temperature gradient, the
bulk Richardson number or theMonin-Obukhov length over open terrain.

Downloaded from https://www.cambridge.org/core . University of Texas Libraries, on 08 Dec 2019 at 08:58:32, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms . https://doi.org/10.1017/9781108674614.006
115 6.5 Eulerian Models of Atmospheric Dispersion

Table 6.2. Atmospheric stability categories a as a function of simple meteorological data, after Turner (1970).

Day Night

Solar radiation b Cloudiness c

Wind speed (m s − 1) at 10 m Strong Moderate Weak ≥ 4/8 ≤ 3/8

<2 A A or B B F F
2 to 3 A or B B C E F
3 to 5 B B or C C D E
5 to 6 C C or D D D D
>6 C D D D D

(a) Stability categories: A (very unstable), B (unstable), C (moderately unstable), D (neutral),


E (stable), F (very stable). (b) Solar radiation: strong (>700Wm − 2), moderate (between 350 and 700Wm − 2), weak (<350Wm −
2)

(c) Cloudiness: fraction of the sky covered by clouds.

Table 6.3. Atmospheric stability categories as a function of the bulk Richardson number, the vertical
temperature gradient, and the Monin-Obukhov length.

Atmospheric stability Bulk Richardson Vertical temperature Monin-Obukhov length c


number a
category gradient b ( °C m − 1) (m)

A Ri b < – 0.86 Δ T/ Δ z < – 1.9 − 10 < L MO < 0


B − 0.86 < Ri b < – 0.37 − 1.9 < Δ T/ Δ z < – 1.7 − 15 < L MO < – 10
C − 0.37 < Ri b < – 0.10 − 1.7 < Δ T/ Δ z < − 1.5 − 100 < L MO < − 15
D − 0.10 < Ri b < 0.053 − 1.5 < Δ T/ Δ z < – 0.5 L MO < – 100; 100 < L MO
E 0.053 < Ri b < 0.134 − 0.5 < Δ T/ Δ z < 1.5 30 < L MO < 100
F 0.134 < Ri b 1.5 < Δ T/ Δ z 0 < L MO < 30

(a) The observed temperature must be corrected using a vertical gradient of − 1 °C / 100 m to obtain
the potential temperature. See Equation 4.62 for the de fi nition of Ri b.
(b) The temperature gradient must be measured over a difference in altitude that is signi fi cant (at least
100 m)
(c) For a roughness length of 0.1 m (open terrain). For a greater value of the roughness length, the
absolute values of the ranges increase, since mechanical turbulence becomes more important. See Equation 4.59 for
the de fi nition of L MO.

6.5 Eulerian Models of Atmospheric Dispersion

6.5.1 General Considerations

Lagrangian dispersion models are speci fi c to a single source and, therefore, cannot be applied to a large number
of sources, as required when simulating air pollution in an urban

Downloaded from https://www.cambridge.org/core . University of Texas Libraries, on 08 Dec 2019 at 08:58:32, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms . https://doi.org/10.1017/9781108674614.006
116 Atmospheric Dispersion

area or at a regional, continental or global scale. In such cases, an eulerian chemical-transport model is generally
used. Air pollutant concentrations are calculated in a three-dimensional (3D) grid and they are not de fi ned with
respect to a source in particular. Therefore, atmospheric dispersion is calculated based on the atmospheric
conditions and is independent of the source-based dispersion process. In other words, within a model grid cell,
one assumes that the dispersion process includes the full spectrum of the turbulent eddies. The turbulent diffusion
coef fi cients in the horizontal direction, K h, and vertical direction, K z, are de fi ned solely as functions of meteorological
conditions.

6.5.2 Vertical Dispersion

The vertical dispersion coef fi cient is important, because it governs the mixing of air pollutants within the PBL. It is
a function of meteorological conditions, and several formulations are available. The two main categories of
formulations include the local and non-local parameterizations. Local parameterizations consist of a standard fi rst-order
closure of turbulent diffusion (so-called K-theory formulations). Thus, vertical dispersion is represented via mass
transfer between adjacent model layers. Non-local parameterizations apply to strongly convective conditions (i.e.,
with important vertical mass transfer), when the atmosphere becomes mixed signi fi cantly over large depths.
Then, the transfer of air pollutants may occur among non-adjacent layers of the model. Figure 6.6 depicts
schematically these different formulations.

The convective dispersionmodel presented here is the asymmetric convectivemixingmodel (ACM). Mass transfer
from the bottom layer occurs upward into all layers above. On the other

K-type turbulent diffusion Asymmetric convective mixing Asymmetric convective mixing with
(ACM) K-type diffusion (ACM2)

Figure 6.6. Schematic representations of vertical dispersion in eulerian models. Left fi gure: standard K-theory turbulent di ff usion model; middle fi gure:

convective dispersion model; right fi gure: model combining convective dispersion and K-theory turbulent di ff usion.

Downloaded from https://www.cambridge.org/core . University of Texas Libraries, on 08 Dec 2019 at 08:58:32, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms . https://doi.org/10.1017/9781108674614.006
117 6.5 Eulerian Models of Atmospheric Dispersion

hand, downward mass transfer occurs between adjacent layers (some earlier models used formulations allowing
downward mass transfer among non-adjacent layers). The assumption used in ACM is that the upward mass
transfer is due to strong convection, whereas downward transfer results from a slow subsidence and, therefore, is
more gradual. Both approaches (K-theory and ACM) have been combined to provide a more complete
representation of vertical mixing processes; the corresponding model is called ACM2 (Pleim, 2007).

Several parameterizations have been developed to represent the vertical turbulent diffusion coef fi cient, K z, for
models using K-theory for turbulence closure. For example, similarity theory may be used for the PBL to develop
parameterizations of K z ( see Chapter 4). By analogy with the parameterization of the turbulent kinematic viscosity
(i.e., the coef fi cient of turbulent transfer of momentum) presented in Equation 4.31, the following equation may be
used for K z:

2
z
Kz ð z Þ ¼ κ u z1 ð 6:59 Þ
Φ z0
zi
L MO

where κ is the von Kármán constant (0.4), u* is the friction velocity (see Chapter 4.3), L MO
is the Monin-Obukhov length, which characterizes atmospheric stability (see Chapter 4.4),
z is altitude, z i is the PBL height, and z ’ is equal to z, except for unstable conditions when z ’
is the surface layer depth (0.1 z i) for z > z ’. The PBL height, z i, is usually obtained from a meteorological numerical
simulation; however, parameterizations are also available to estimate it (see for example, Troen and Mahrt,
1986). The function Φ varies depending on the atmospheric stability. Those presented in Equations 4.68 and
4.69 for heat transfer may be used for mass transfer. Figure 6.7 shows the vertical pro fi le of this vertical
dispersion coef fi cient for various atmospheric stability conditions.

500
Unstable

Neutral
Altitude (m)

300
400

Stable

200

100

20 40 60 80 100 120
K z ( m 2 s − 1)

Figure 6.7. Vertical pro fi les of the eulerian vertical dispersion coe ffi cient (similarity theory formulation). Meteorological conditions: stable (dashed

line), neutral (dash-dotted line), and unstable (solid line). The PBL height is 500 m. The friction velocity ( u*) is 0.3, 0.4, and 0.6 m s − 1 for

stable, neutral, and unstable conditions, respectively.

Downloaded from https://www.cambridge.org/core . University of Texas Libraries, on 08 Dec 2019 at 08:58:32, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms . https://doi.org/10.1017/9781108674614.006
118 Atmospheric Dispersion

Other parameterizations of vertical dispersion have been developed for the PBL and the free troposphere. For
example, the parameterization of Louis (1979) is used inmanymodels for the PBL. Some other parameterizations
use the convection velocity, rather than the friction velocity (e.g., Troen and Mart, 1986).

6.5.3 Horizontal Dispersion

Various formulations are available to parameterize the horizontal dispersion coef fi cient in eulerian
chemical-transport models. These formulations depend usually on the horizontal grid size of the eulerian model.
However, the dispersion coef fi cients depend on model grid size in ways that may lead to confusion: one
formulation uses a proportional relationship between the dispersion coef fi cient and the grid surface area,
whereas another formulation uses an inversely proportional relationship. The most widely used formulation is that
of Smagorinsky (1963):

K h;Smagorinsky ¼ 0:16 ð S Γ 2 þ S Λ 2 Þ 1 2 ð Δ x Þ 2

∂v
SΓ¼ ∂ u
∂x ∂y ð 6:60 Þ

SΛ¼ ∂ v
∂ x þ ∂ ∂u y

where S Γ represents the stretching deformation and S Λ the shearing deformation of the wind fi eld; Δ x is the
horizontal grid size (m). This formulation leads to a horizontal dispersion coef fi cient, which is proportional to the
horizontal grid cell surface area. Therefore, horizontal dispersion will become very large for a model simulation
where a large horizontal grid size is used (say >10 km). Since numerical diffusion increases with grid size, it is
likely that the horizontal dispersion term will become super fl uous when the horizontal model resolution is coarse.
Thus, it appears that this formulation should be limited to cases where the model horizontal grid size, Δ x, is on the
order of a few km.

Another parameterization, called Unif, has been proposed to counter numerical diffusion in cases where a
coarse horizontal resolution is used (Byun and Schere, 2006):

K h;Unif ¼ 2000 4000 ð 6:61 Þ


Δx

This formulation leads to a horizontal dispersion coef fi cient, which is inversely proportional to the horizontal
grid cell surface area, ( Δ x) − 2. Therefore, its value decreases when
Δ x increases, i.e., when numerical diffusion increases. However, this formulation is not applicable to simulations
where a very small grid size is used, because K h tends toward in fi nity when the grid size decreases (for example,
it is 32,000 m 2 s − 1 for Δ x = 1 km).
An evaluation of these two formulations was performed using a fi eld experiment conducted in 1999 in Texas.
Tracer gases were released from different locations and measured downwind with a monitoring network. The
model simulation used two imbedded domains with horizontal grid sizes of 12 and 4 km, respectively. The results
suggested that, for this

Downloaded from https://www.cambridge.org/core . University of Texas Libraries, on 08 Dec 2019 at 08:58:32, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms . https://doi.org/10.1017/9781108674614.006
119 6.6 Street-canyon Models

10 5

3 × 10 4

10 4 K h,Unif K h,Smagorinsky

3 × 10 3

Horizontal dispersion coefficient (m 2 s − 1)


10 3

3 × 10 2
Kh

10 2

30

10

10 20 30 40 50

Grid cell horizontal size (km)

Figure 6.8. Eulerian horizontal dispersion coe ffi cients: K h ( solid line), K h,Smagorinsky ( dash-dotted line), K h,Unif ( dashed line). A value of 10 − 4 s − 1 was used

for the wind fi eld deformation.

model con fi guration, the Smagorinsky formulation led to better model performance than the Unif formulation (Pun
et al., 2006). Byun and Schere (2006) proposed the following formulation, which combines the two formulations
previously described:

1
Kh¼ ð 6:62 Þ
ð Kh;Smagorinsky
1
þ K1 h;Unif Þ

When Δ x tends toward 0, ( K h,Unif) − 1 tends toward 0 and K h tends toward K h,Smagorinsky;
when Δ x tends toward in fi nity, ( K h,Smagorinsky) − 1 tends toward 0 and K h tends toward K h,Unif.
Figure 6.8 illustrates how these eulerian horizontal dispersion coef fi cients depend on the horizontal grid size.

6.6 Street-canyon Models

The application of gaussian models to the urban canopy is strongly limited by the presence of buildings, which
affect the air fl ow and, therefore, the transport and dispersion of air pollutants. In addition, the air pollution in a
street canyon results mostly from local sources and it cannot be modeled by eulerian models, which have a grid
resolution of at least 1 km,
i.e., much greater than the dimension of a street canyon. Computational fl uid dynamics (CFD) models can solve
the equations that govern the air fl ow and the dispersion of the pollutants while taking into account the effect of
buildings; however, the associated computational costs are too large to make CFD models applicable to large
domains for long periods. Therefore, other approaches had to be developed to simulate the transport and

Downloaded from https://www.cambridge.org/core . University of Texas Libraries, on 08 Dec 2019 at 08:58:32, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms . https://doi.org/10.1017/9781108674614.006
120 Atmospheric Dispersion

dispersion of pollutants in an urban canopy with parameterizations that take into account the presence of
buildings.
The most commonly used formulations are currently the Operational Street Pollution Model
(OSPM; Berkowicz, 2000) and SIRANE (Soulhac et al., 2011).
The latter is the most recent and it is summarized here. This formulation uses the assumption that the air
pollutant concentration is uniform within a street canyon (or a segment of a street canyon). The turbulent transfer
of air pollutants between the street canyon and the atmosphere above the urban canopy is modeled using a
mass transfer coef fi cient at roof level. The horizontal transport of air pollutants within the street canyon is
modeled using a wind that is parallel to the street axis, with a wind speed calculated using an exponential vertical
pro fi le. Then, the concentration of an air pollutant within the street canyon, C ( g m − 3), is calculated via a
steady-state mass balance, which takes into account the pollutant emissions in the street as well as deposition
processes:

u s h b W s C u þ S l L s ¼ u s h b W s C þ F d L s ð W s þ 2 h b Þ þ σ w W s Lp sπffiffiffi
ð C Cb Þ ð 6:63 Þ
2

where u s is the mean wind speed within the street canyon (m s − 1), σ w is the standard deviation of the vertical wind
speed at roof level (m s − 1), h b, W s, and L s are the average height of the buildings, the average street width, and the
length of the street-canyon segment, respectively (m), S l is the emission rate of the pollutant in the street-canyon
segment (g m − 1 s − 1), C u is the concentration of the pollutant transported form the upwind street-canyon intersection
(g m − 3), C b is the background concentration of the pollutant above the urban canopy (g m − 3, concentration obtained
either from air monitoring stations or from a model simulation), and F d ( g m − 2 s − 1) is the average dry deposition

fl ux on urban surfaces (street, buildings . . . ). The average wind speed within the street is calculated using the
assumption of an exponential vertical wind pro fi le (see Equation 4.43) and accounting for the angle between the
wind direction above the urban canopy and the direction of the street-canyon axis. The fi rst term of the equation
represents the pollutant in fl ow rate from the upwind intersection and the second term represents the emission
rate within the street. The third term represents the out fl ow rate toward the downwind intersection. The fourth
term represents dry deposition on surfaces. Scavenging by rain is not included here, but may easily be added if
needed. The fi fth term represents pollutant transfer between the street canyon and the background atmosphere
above the urban canopy. It depends only on atmospheric turbulence (represented by σ w)

and does not account, for example, for traf fi c-generated turbulence or the effect of vegetation on atmospheric
turbulence. Deposition fl uxes may be calculated with parameterizations described in Chapter 11; they may be a
function of surface type (street, sidewalk, building walls, windows, etc.).

SIRANE simulates the horizontal transport of air pollutants at street intersections by taking into account the
variability of the wind direction and by conducting a mass balance of the fl ows through the intersection. Thus,
some turbulent mass transfer at roof level is used to compensate for the horizontal in fl ows from and out fl ows to
the street canyons connected with the intersection. The set of street-canyon equations is solved

Downloaded from https://www.cambridge.org/core . University of Texas Libraries, on 08 Dec 2019 at 08:58:32, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms . https://doi.org/10.1017/9781108674614.006
121 6.7 Numerical Modeling of Pollutant Transport in the Atmosphere

numerically, and the pollutant concentrations are calculated for each street-canyon segment.

If the wind is perpendicular to the street-canyon axis, an air recirculation zone may develop on the upwind
side of the street while a ventilation zone may appear on the downwind side. Therefore, if an air pollutant is
emitted within the recirculation zone, its concentration will be greater than that of an air pollutant emitted within
the ventilation zone. The OSPM formulation takes into account the presence of recirculation and ventilation
zones within a street-canyon con fi guration. However, the horizontal con fi guration of the street-canyon network
may also generate recirculation zones and an academic representation of recirculation and ventilation zones may
not always apply. Then, a CFD model is needed to simulate the air fl ow and calculate air pollutant concentrations
in cases of complex con fi gurations.

6.7 Numerical Modeling of Pollutant Transport


in the Atmosphere

Gaussian plume models and street-canyon models have analytical solutions, which do not present any particular
numerical dif fi culties. However, the calculation must be conducted for each source and receptor (i.e., the location
where the air pollutant concentration is calculated) or for each street-canyon segment. Therefore, a large number
of sources, receptors, and street-canyon segments may lead to signi fi cant computational costs.

Puff models require more computations, because puff trajectories must be tracked according to a wind fi eld
obtained from a numerical meteorological model. However, the computation of the pollutant concentrations within
the puffs does not lead to any more numerical dif fi culties than those of gaussian plumes (puffs that treat chemical
reactions may lead to numerical dif fi culties due to fast kinetics, see Chapter 7).

Eulerian chemical-transport models require a numerical solution of the set of the mass conservation equations
representing the processes governing the concentrations of the air pollutants. Two distinct terms appear in the
partial differential equation (excluding source, sink, and transformation terms): a term representing transport of
the air pollutants by the mean wind and a turbulent dispersion term.

The fi rst term (transport by the mean wind) is hyperbolic and presents a numerical dif fi culty: it leads to
numerical diffusion, which arti fi cially spreads concentrations throughout the model gridded system. Numerical
diffusion may be reduced by using a fi ner spatial resolution, but, then, the computational cost increases
considerably. Therefore, numerical algorithms have been developed to minimize numerical diffusion while
keeping the computation costs manageable. However, some algorithms may have negative secondary effects
such as the propagation of concentrations upwind (e.g., Bott ’ s algorithm) or non-conservation of mass (e.g.,
some semi-lagrangian algorithms based on spatial gradients). Decisions on the choice of a numerical
algorithmmust be made based on the application at hand. The Smolarkiewicz algorithm is widely used; it
presents few

Downloaded from https://www.cambridge.org/core . University of Texas Libraries, on 08 Dec 2019 at 08:58:32, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms . https://doi.org/10.1017/9781108674614.006
122 Atmospheric Dispersion

shortcomings, however, it is more diffusive than some other algorithms. The piecewise parabolic method is used
in several models, because it presents the advantage of being positive de fi nite and monotonous (Colella and
Woodward, 1984). The algorithm of Walcek and Aleksic (1998) is also monotonous and mass conserving; in
addition, it may be used via operator splitting for both transport terms (mean wind transport and dispersion).
Comparisons of numerical algorithms have been conducted, which provide valuable information on the pros and
cons of various available algorithms (e.g., Dabdub and Seinfeld, 1994).

The second term (dispersion) is parabolic and, therefore, does not present any particular numerical dif fi culty.
Standard numerical algorithms, such as semi-implicit methods, may be used (e.g., von Rosenberg, 1969).

Problems

Problem 6.1 Dispersion of a gaussian plume


A plume is emitted from an incinerator stack. Atmospheric conditions are assumed to be stationary, and a
gaussian plume model is used to simulate the atmospheric dispersion of this plume. The dioxin emission rate is 6
× 10 − 4 g s − 1, the stack height is 20 m, atmospheric stability is neutral, the wind speed is 5 m s − 1, and plume rise is
assumed to be negligible. Calculate the dioxin concentration, C, at a location 1 km downwind, underneath the
plume at a height of 1.5 m corresponding to a person ’ s exposure.

Problem 6.2 Atmospheric dispersion of air pollutants


Air pollutants emitted from a tall stack lead to exceedance of an ambient air quality standard at a nearby
monitoring station. Reducing the emissions with some control equipment would be too costly, and increasing the
stack height is not allowed by the local regulatory agency. Suggest another approach to increase the dilution of the
air pollutants in order to attain the ambient air quality standard.

Problem 6.3 Plume rise


Calculate the plume rise for a power plant stack with the following characteristics and atmospheric conditions: the
stack diameter is 7.3 m, the plume initial vertical velocity at the stack exit is 18 m s − 1, the plume initial temperature
is 77 °C, the ambient temperature is 20 °C, and the horizontal wind speed is 5 m s − 1.

Problem 6.4 Lagrangian and eulerian representations of atmospheric dispersion


The gaussian dispersion coef fi cients of McElroy-Pooler under rural conditions and neutral atmospheric stability
are used here (Table 6.1) with a wind speed of 10 m s − 1.

a. Calculate the corresponding eulerian coef fi cients.


b. At which downwind distance from the source can these eulerian dispersion coef fi cients
be used?
c. Check whether the values of these eulerian dispersion coef fi cients are consistent with
the values given in this chapter.

Downloaded from https://www.cambridge.org/core . University of Texas Libraries, on 08 Dec 2019 at 08:58:32, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms . https://doi.org/10.1017/9781108674614.006
123 References

References

Berkowicz, R., 2000. OSPM – A parameterized street pollution model, Environ. Monitoring
Assessment, 65, 323 – 332.
Bird, R.B., W.E. Stewart, and E.N. Lightfoot, 2006. Transport Phenomena, John Wiley & Sons,
Inc., USA.
Briant, R., I. Korsakissok, and C. Seigneur, 2011. An improved line source model for air pollutant
dispersion from roadway traf fi c, Atmos. Environ., 45, 4099 – 4107.
Briggs, G.A., 1984. Plume rise and buoyancy effects, Atmospheric Science and Power Production,
U.S. Department of Energy, USA.
Byun, D. and K.L. Schere, 2006. Review of the governing equations, computational algorithms, and
other components of theModels-3 CommunityMultiscale Air Quality (CMAQ) Modeling System,
Appl. Mechanics Rev., 59, 51 – 77.
Carson, J.E. and H. Moses, 1969. The validity of several plume rise formulas, J. Air Pollut. Control Assoc., 19, 862 – 866.

Colella, P. and P.R. Woodward, 1984. The piecewise parabolic method (PPM) for gas-dynamical
simulations, J. Comp. Phys., 54, 174 – 201.
Csanady, G.T., 1973. Turbulent Diffusion in the Environment, D. Reidel Publishing Company, Dordrecht, The Netherlands.

Dabdub, D. and J.H. Seinfeld, 1994. Numerical advective schemes used in air quality models –
Sequential and parallel implementation, Atmos. Environ., 28, 3369 – 3385.
EPA, 1995. User ’ s guide for the industrial source complex (ISC3) dispersion models, U.S.
Environmental Protection Agency, Research Triangle Park, NC, USA; available at: www
.epa.gov/ttn/scram/userg/regmod/isc3v2.pdf.
Gifford, F.A., 1961. Use of routine meteorological observations for estimating atmospheric
dispersion, Nucl. Safety, 2, 47 – 51.
Karamchandani, P., L. Santos, I. Sykes, Y. Zhang, C. Tonne, and C. Seigneur, 2000. Development and
evaluation of a state-of-the-science reactive plume model, Environ. Sci. Technol., 34, 870 – 880.
Kim, Y., C. Seigneur, and O. Duclaux, 2014. Development of a plume-in-grid model for industrial
point and volume sources: Application to power plant and re fi nery sources in the Paris region,
Geosci. Model Dev., 7, 569 – 585.
Louis, J.-F., 1979. Aparametric model of vertical eddy fl uxes in the atmosphere, Bound.-Lay. Meteor.,
17, 187 – 202.
Overcamp, T.J., 1976. A general Gaussian diffusion-deposition model for elevated point sources,
J. Appl. Meteor., 15, 1167 – 1171.
Pleim, J.E., 2007. A combined local and non-local closure model for the atmospheric boundary layer.
Part 1: Model description and testing, J. Appl. Meteor. Climatol., 46, 1383 – 1395.
Pun, B., C. Seigneur, K. Vijayaraghavan, S.-Y. Wu, S.-Y. Chen, E. Knipping, and N. Kumar, 2006.
Modeling regional haze in the BRAVO study using CMAQ-MADRID. 1. Model evaluation,
J. Geophys. Res., 111, D06302, doi:10.1029/2004JD005608. von Rosenberg D., 1969. Methods for the Numerical Solution of
Partial Differential Equations,
Elsevier, New York.
Smagorinsky J., 1963. General circulation experiments with the primitive equations: 1. The basic
experiment, Mon. Wea. Rev., 91, 99 – 164.
Soulhac, L., P. Salizzoni, F.-X. Cierco, and R.J. Perkins, 2011. The model SIRANE for atmo-
spheric urban air pollution dispersion: Part I, presentation of the model, Atmos. Environ., 45,
7379 – 7395.
Taylor, G.I., 1922. Diffusion by continuous movements, Proc. London Math. Soc., A20, 196 – 212. Troen, I.B. and L. Mahrt, 1986.
A simple model of the atmospheric boundary layer: Sensitivity to
surface evaporation, Bond.-Lay. Meteor., 37, 129 – 148.

Downloaded from https://www.cambridge.org/core . University of Texas Libraries, on 08 Dec 2019 at 08:58:32, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms . https://doi.org/10.1017/9781108674614.006
124 Atmospheric Dispersion

Turner, J.S., 1970. Workbook of Atmospheric Dispersion Estimates, U.S. Environmental Protection Agency, Research Triangle
Park, NC, USA.
Venkatram, A. and T. Horst, 2006. Approximating dispersion from a fi nite line source. Atmos.
Environ., 40, 2401 – 2408.
Venkatram, A. and N. Schulte, 2018. Urban Transportation and Air Pollution, Elsevier,
New York.
Walcek, C.J. and N.M. Aleksic, 1998. A simple but accurate mass conservative, peak-preserving,
mixing ratio bounded advection algorithm with Fortran code, Atmos. Environ., 32, 3863 – 3880.

Downloaded from https://www.cambridge.org/core . University of Texas Libraries, on 08 Dec 2019 at 08:58:32, subject to the Cambridge Core terms of use, available at
https://www.cambridge.org/core/terms . https://doi.org/10.1017/9781108674614.006

Anda mungkin juga menyukai