Instructions for use

Title 北海道の気流系 : 風ベクトル合成法による解析

Author(s) 加藤, 央之

Citation 環境科学 : 北海道大学大学院環境科学研究科紀要, 6(2), 301-317

Issue Date 1984-03-30

Doc URL http://hdl.handle.net/2115/37158

Type bulletin (article)

File Information 6(2)_301-317.pdf

Hokkaido University Collection of Scholarly and Academic Papers : HUSCAP

301

i-"Environ. E..g.i;.l..Hokkaiclo l 6 (2) .11..301-317 l ll?.e.f.:..I.98.3......i

Air Current System in Hokkaido

-An Analysis Using Wind Vector Composition Method-

Hisashi Kato

Energy and Environment Laboratory, Environment Department, Atmospheric Environment Section, Central Research Institute of EIectric Power Industry, Tokyo 201

#k val iilg o St 3ii l%

jNKg i･ ivA., Jrtit2kec at 6getvi

lltl ue tLlq: ;2Z

'wsJt]Ft;VLLWI:eeErr=*ivlj'-NeY{i}ll7wsnjMS,Y<gliteeeereFYtiki

l. IRtroduction

In the previous paper (Kato, 1982) it was pointed out that the clear air current

iB any general wiRd direction should be described in order to clarify the regional

features of wind. Then in this paper the regional characteristics of air current

are described in detai] by means of an original method developed by the author.

SiRce the wind is a vector, its tyeatment is difficult. PCA (Princ!palcomponent analysis) has beeR applied to wind data in a few studies. Barnett (1977)

showed the wind distributioR patterns for u and v components, respectively. This

procedure is also used by Wal<amatsu and HataRo (1981) and Enfie]d (1981). On

the other hand, Hardy (1978) showed the wind distribution pattern as analysed

from £he vector component. However, the former studies did not show the actualair current systems and the }atter showed only a few air current systems.

TheR in order to clarify the various features of air current systems the follow-

lng methods are used. (1) The u and v compoRents are analysed by PCA, respec-

tively. (2) Distribution patterns of or. and v components are ca]culated using the

results of (1) when the correspondiRg Z-score has the maximum or the rninimum

value. (3) Each distribution pattern of tt(v) component is combiRed with that of

each v(u) componeRt referring to the corre]ation between Z-scores of zt, and v

components. (4) The air current systems are described using the results of (3).

(5) Pressure patterns corresponding to the systems are investigated. The air current

systems eonstructed by this method indicate the wind distribution pattern when

Received 8 Oct. 1983

1983 # 10 rs 8 -Sge

302 Environmental Scienee, Hokl<aido VoL 6, No. 2, 1983

the .creneral wind speed is strong in each direction.

2. Data and Methods

In this analysis the AMeDAS (Automated Meteorological Data Acquisition

System) wind data in 1978 are used. Basic data for u(T,) eomponent (134 stations×365 days), for example the u(v) axis con'esponding to E-W <N-S) direction, are

obtained from the daily mean of ?t(z,) component at each statlon. Missing data

which are interpo]ated by the three neighboring stations by the method of Suzuki

<1964) are O.66% of the total data and are regarded not to affect the results in

this analysis.

For the sake of brevity, the case with two variables is treated in the fol]owing

explanation. If we plot the wind data of each day on an orthogonal coordinate

system, O-xy, with Ui (daily ze component at one station) as x coordinate axis and

wjth Uli (daily u component at another station) as y coordinate axis, the points on

the plane are scattered in a el]iptical fashion (Fig. 1). In order to inspect the

relation between Ul and eq, principal component Zr. (first component) and Zh2t

(second component) are caiculated. Hence the character zt indicates u compoRent.

These principal components are given by

Zl,,=l,,･(L71+l,,･Ul, (1> Zle,,=:l2i･Ul+l22･q, (2)where lij is the coeflicient of the equations. These principal components indlcate

the rotation of the orthogonal coordinate system and are defined by

Zl v.e =Ui 'cos 0+ Uh 'sin0 <3) k. == -q･sin o+ q･cose (4)The direction of Zl,,, is in agreement with that of the apse line and also Zli,, with

the miRor axis. It is clear that the increase (decrease) of Zl,, shows the increase

(decrease) both of Ul and Q (see Fig. 1). Similarly, the increase (decrease) of

U2

11"x//-1/.f

..--- .V'- "x/'LN"-- e-@N･.A(ui,,l

.p(%!,l3)J22)yt'i/e.'!

1.

Fig. 1. An explanation of PCA in the case with two variables.

AirCurrentSysteminHokkaiclo ' 303

Zh,, shows both the decrease (increase) of Ul and increase (degrease) of ca. The

data of the day when Zl., has the maximum value corresponds to point A inFig. 1. As is evideRt from the figure, Ui and ca values of this day, i.e. UIA

and ca.b are written as '. '

Ul.= U,+k･cosO (s) q. =:= q+k･sino . ,.. (6)where Ui (q) is the ann-uaimeahl'of Ui (q). Substituting' (5) and (6) in (3), one

obtains

Zl.,.,,.::Ui･cosO+q･sinO+k (7)where Zl,,,,... shows the maximum value of Zi,, during the year. IR the equation

(7), Ul and q are ca]culated from the data. The values of cosO (xlH) andsinO(=li2) are also obtained as a resu}t of PCA. Since the total of the first

two terms in the right side of equation (7) equals the annual mean of Zl., Zi?,,

is used in the actual calculation. If the Zi,,,.,,. is obtained, the value of fe is

determined. Thus, substituting the k value into (5) and (6), one obtalns UIA and

Uliri. In this analysis Zl.,-score is calcu]ated for each day using equation (1), and

its frequency is approximately regarded as Normal distribution. Zi,,,.... is deter-

mined from the distribution as the value which occurs once a year.

In much the same manner as this, zt components of two statlons corresponding

to the minimum score of Zl,,, and to the maximum and mlnimum scores of' Zli.,

are ca]cula£ed. One a]so obtains the z) components to each Z-score. These pro-cedures are applicable in a case with a large number of variables. WheR the num-

bex of variables is 72, it or z) componeRts at each statlon are a]} calculated cor-

responcling to the maximum and minimum values of Zt,, (i-nv1,,2,･･･,n) and ldt`

(j'=1, 2, ･･･, n), respectively.

Next, the coeflicients of correlation between Z,i.. aRd Zj, are ca]cu]ated aRd

the couples with high correlations are picked up. For examp]e, we assume that

the corre]ation between Zi,,, and Zht) has a high positive va]ue in the case of two

variables. Since the time variations of Zl･,, ancl Zh, are similar to each other,

Zli. is apt to be c]ose to its maximum value when Jill. has its maximum value.

In other words the distribution of u. component, Uli ancl Ul2 (at Zl.,.ax), tends

to occuy when the v component has the distribution pattern of V2i and Vb2,

corresponding to Zb,,.,.. Hence the athxed numbers of U(V) indicate the order

of principal compoRent (former) and the station number (latter). Then based on

the assumption that both Zlu.m,ix and Zhv.m.. occur at the same time, thewind at each ith station at that time is obtained by composing the vector com-

ponents Uli, and V2i,. This procedure can be applied in general cases (p-variab]es)

(see Fig. 2). The wind distribution patterns obtained by these procedures are

discussed in connection with the pressure patterns.

By these methods, wind distribution patterns determined by synoptic scale

pressure field and the major topographical features alone can be delineated, and the

304 Environmental Science,

Zlu･max

Ul1 -ww' p U12 - Ulp U13 , 't"""--"'-- .

. <--. g Ulp

Fig. 2.

`kiiix

Hokkaido VoL 6,

Z2v･max

V21

V22

V23

.

.

V2p

Ne. 2, 1983

;

t

;

:

pV2p;

;

pl

Z:rssg

x.

noise su.c.h as background the windre]ated to the heat of thesemethods.

3. Analysis

The proportions of each principal component on the case of 1) ?t(E-W) and

v(N-S) and 2) u(SE-NW) and v(NE-SW) are shown in Table 1, The accumu]atedproportion of the first two principal components exceeds by about 70% in both

cases of l) and 2). ' Fig. 3 shows an example for the frequencies of Zl,,- and Zlv-scores in the

case when u and v axes correspond to the E-W and N-S lines, respective]y. In

Table1. Proportionsoffirstfiveprincipa!components in the case of 1) tt(E-W)-'v(N-S) axes and

2) u(SE-NW)-'v(NE-SW) axes (%)

A schema for composing vector components, See the textfor detail.

fluctuatioR of wind peculiar to each station or

effect is removed. This is the outstanding merit

I1 2 3 4 5

1)

u-component

v-component2)

u-component

v-component

(E-W)

(N-S)

(SE-NW)

(NE-SW)

::

I

l

ii

i 56,6

62.4

64,9

52,1'

13.8

9.l

9.8

14.1

5.9

6.1

4,2

7.0

3.6

2.7

3.4

4.2

2.6

2.3

2.2

2,9

so

o

it

,

50

Air Current System in Hokkaido

t-

/..l .//

.

.

.

..

Zlu r -- 6,4

SD=13,2

l

3e5

llI ･l l-50 --40 --30 -20 -10

s.

]II O 10 20 iiu

.

ll30 40

ZIV # o,7

SD=t3.1

s

Ol l]ll1]IIl -50 -40 -30 -20 -10 O 10 20 30 40 Zlv

Fig. 3. Frequencies of Ziu, and Ziv for the case with u-component

along E-W direction. Dot shows each frequency. The curves show the Normal distribution derived from the mean ancl the standard deviatio'n for each ZLscore. Arrows indi-

cate the maximum and the minimum Z-scores determined from each Normal clistribution.

the figure the abscissa corresponds to the Z-score and ordinate to the frequency.

Dots indlca£e each frequency and the curve of the Normal distribution is describedby the mean and the standard deviation in each Z-score. Since the distributions

of Z-scores are similar to the calculated eurves including these cases, each maximum

(minimum) Z-score is determined from these Normal distributions as they occur

once a year (i.e., about O.27% of the total data). These are shown by arrows in

the figures. Then the distyibutions of u and w components are calculated at each

maximum and minimum va]ue by the equatlons (5), (6) and (7).

The correlation between Zt,, (irm-1, 2, ･･･, 5) and Zli, (]'--1, 2, ･･･, 5) for the two

cases 1) and 2) mentioned above are shown in Table 2. A positive or negative

correlation clearly exists between Zl,, and Zb., and betweeR ZIv and Zli,,, in both

cases 1) and 2). Then the ze ancl w components corresponding to the extreme

Z-scores are combiRed at each station (see Fig. 2), and the wind distribution patterns

are obtained. In each direction, the score of Zl.(Zl.) is highly correlative (greater

than O.98) to the areal mean of daily u(v) component. Namely, the Zl. (Zl.)-

score shows the strength of the areal mean wind speed with regard to the direction

of z((v) component, and the wind patterR corresponding to Zlac.max (Zlv.max)

appears when the areal mean of daily u(v) component shows a maximum value.In the case of it(E-W) and v(N-S), with the east direction as the positive, the

wind patterR corresponding to Zl..... shows the air current system when the

306

1) u(E-W),

Environmental Science, Hokkaido Vol. 6, No. 2,

Table 2.

･v(N-S)

Correlation coeffcients betNNieen first five

'of u(E-W> and v(N-S) components

1983

Z-$cores

ZlV '',.g'l'

O.78

-O.27

-O,28

O.05

･v(NE-SW)

liv 2lev Kiv kv

2,li

Z'ii i

ttt... .t.............tt..t...............t......

2) u(SE-NW),

-O,83

O.30

-e.11

-O.05

O.12

- O.I4

e.19

O.11

-O.04

-O.42

O.02

O.04

O.14

-O.12

O.11tttttttttttttttttttttt ttttttttttttttttttttttttttttttttt

-O.l.8

O,06

O.20

-O.12

e.27 ttttttttttttttttttttttttttttttttttttttttt ttttttttttt

Zl･u

ku.

Zb,,.

kuZ6u,

ttttt

lttt

i 1

i'

111･

ZlV

-o.I6'

-e.78

-024

O.07

O.32tttt ttttttttttttttttttttttttttt

Z2v Zbvttt tttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttt

-O.90 -O.15

O,13 -O,31 -e.13 -O.03 -OO.9 O.03 O,Ol -Or22ttttttttttttttt ttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttt ttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttttt

N . i4

i

.;:------.J･ M(rr,v)1-: il,'ll'i.;'ii,... ,:IIIi>- ,lx.･/.x,x

.`. '.'.:1=:-ti./t-;:t;'4K:sl`: '. ')N,2

-

Z4Vtttttttttttttttttttttttttttttt

-O,21

-O.05

O,45

O.11

-O.14

kvttt tt tt ttt tttttttttttttttttttttttttttttt tttttttttttt

-o.e4

-O.07

e.13

-O.03

-O.22

W . . 4

.

/t

t

r

1.t 4

Fig. 4.

)Xx,i;･l 1-f[il･ii/':i-tt/,itE,L'i Ittli X"ti 'i

- xNxlN.s..-.fti-.L.-//' I

4imfsec L'

g

Areal mean vector wind for each day (l978).

indicates the maximum wind speeddirection. See Section III-4-c for the

r･-･-･i-v) V=9N

CN

A

i

B

Bro1<enca}cu]ated

cliscussion.

x

J

for

E

i

v=v[

!l

line

each

Air Ctirrent Systern in Hokkaido 307

easterly component of areal mean wind has the maximum value. On the otherhand, the pattern at Zl.,.i. describes that 'for the westerly winds.

Furthermore, £he areal mean vector wind when each Z-score has extremevalue is examinecl. The areal mean vector wind for each day (1978) is shown in

Fig. 4. The clashed line indicates the maximum wind speed for each direction

calcu]atecl in the same manner as in the case of Fig. 3. Hence, point A (in the

rectaRgle surrounded by dashed dotted lines) shows the maximurn value of the area}

mean u component (easterly wind), and corresponds to Zl,,..... However, whenZliA has a maximum value, it ls assuixted that av=l::Zb,,m.x and Zlv::i-=-l21U, j.e.,

w:'i:v. Namely, the wind distribution pattern at Zl.=Zl.,... obtaiRed here is

actually that at point B (zt=I=u.max and w=l=o). In the figure, as the maximumvalue is that of point q the pattern at point B shows that uncler a slightly stronger

wind speed than the measured one. AIthough the errors actua}ly exist as in this

case, their influences seem to be very small in this analys}s. Then, each wlnd

distribution pattern obtained here is considered to be that when the area] mean

vec£or wind in each direction has the maximum wind speed of the year, although

some errors are present m some cases

Table 3 shows £he relation between the diyection of areal mean vector wind

Table 3. Reiation between tl']e areal ]nean vector wind and the pressure

gra(/Hent witli clirectlon in the case when the areal, mean wind

$peed is greater than 1.0 mlsec. The value in the table indicates

the freqtiency of eacli wincl data. The direction of pressure

gradient (e.g. 16; the pressure is higher in the northern side)

is calculatecl from tl]e data of Wald<anai, Kakodate ancl Nemuro

I)IRECTION OF PRESSURE GRADIENT

: llI"' tI

it

i/I

I

2 3 45E6 7 8

s9 10 11 12 W

Zo---

yomptF-i

aaKkpt

O<cara

><

13 14 15 16 N

1

2

3

4E5

6

7

8S9

le

11

12 W

13

zti

15

16 N

i{

:II

il

1ii

I

2 2

2

4

5

1

8

2 17 31 11 5

7

6

4

1

5

12

15

1

5

2L, 12

1 12 11 3

1

4

2

l

4

53

2

308 Environmental Science, Hokkaido VoL 6, No. 2, 1983

and the clirection of the pressure gradient ca]cu]ated from the sea level pressure

of Wal<kanai, Hakodate and Nemuro in the case when the areal mean wind speed

is greater than 1.0 mlsec. The value in the table indicates the frequency of each

wind. From the table, it is noted that the direction of areal mean wiRd is rotated

couRterc]ockwise about one direction (i.e. 22.50) from that of the gradient wind

estimatecl from the pressure gradient, and is actually similar to the directioR of

general wind. Thus the areal mean wind is regarded as the general wind in this

analysis. Ultimate]y, air current systems obtained here correspoRd to the wind

pattern when the general wincl speed in each direction has the annual maximum

value.

4. Results and Discussions

a. Air eurrent system

The air current systems when the general wind speed has the annual maximum

value in each direc£ion are showR in Figs. 5(a)-(h). Each general wind directlonis shown on the upper part of the figure. Thlck solid arrows correspond to the

wind speecl ovey 5m/sec and dashed arrows to that Rnder 1 rn/sec.

General]y speakiRg, each air current system seems to be determiRed both by

the general wind direction and the major £opographic features as pointed out byKawamura (1963, 1966). Referring to the wind speed, strong wind areas exlst

a!ong the airways. On the other hand the weak wind areas lie to the leeward

of obstructions such as mountalns.

Flgs. 5(a)-(h) correspond to Fig. 3 of Kawamura (1963). Both figures show

air current systems uRder strong genera] wind. However, since the definition of

general wind direction in this study has some clifferences from that of Kawamura

(1963), such figures can not be compared simply. For example, Fig. 5(g> cor-

responds to the case under general wind direction between W and NW in his

study. Furthermore, the data using in £his analysis (daily mean) is also differentfrom those of his study (wind at 0900 (JST>). Although some dlfferences exist

between these two studies, both figures show analogous features regarding some

air curreBt sys£ems. Each map of Fig. 5 seems to compensate for the features insome general wincl directions between those of Kawamura (1963).

Then including further discussions of the properties of these results, the features

of air current in severa] ayeas are investigated in the following.

1) Ishika.ri Lo'wland (Ishikari Plain-Yzl･fl`tsu Lo'wlcznd)

The distinct air current exists in the area under the southerly or northerly

genera] wind. Especially under the southerly general wind, it is also characterized

that the stream liRe is divided by the Mashike Mountains as mentioned by Owada

and Yoshino (1971) which are the same as the features clarified in the previous

paper (Kato, 1982). ORe of these air currents toward Ishikari PlaiR clearly exist

under southeasterly or southerly geReral wlnds, on the other hand aRother exists

under the southerly or southwesterly general wind.

Air Current System in Hokkaido 309

(a)

fN

tf6

--

M'/

!

"'

VN

`,N le 1

,

]e/$gpt-

L' /ee....lg

in, i

N

/dix

gr

l--nt+ "{Ve c

xXN Xk i"

ii¥il.li{e

9.K'

i)ll#w

.mer-

x-N

N

x" "

ll./

r,-･r'i'

O 100Xm

lex"

N

x

si/

;

x

mM moo-pa eeo--t2oo

g 4oo- soo

O O-400

o

(b)

g

Fig. S.

ors

i-

-!

M Pig

;es,gger

'S.ifZi'rR'lil,;#,{,i- -

;/ iSfliV

iiillili$til

I

&l

NE

dwivg ,N,i?,

E, N 1 ;/xN X s'i

N

/;,

;

"

#/es - oi el i c. s ･ i{Isr,f'

Seste

t-

.O a

(a) Air current

,N"

,

N

,

t-N;--

leokm

u ifr'

/,. /fiVi$Iii [i

.;x ;

mM 1200-

ee eoo-nooee 4oo-eoo

O O--400

v

map for northerly genera! wind with maximumwind speed fer the direction during a year. Thick solid arrows

correspond to the wind speed over 5 m!sec and broken arrowsto that under 1 in!sec. <b) Sarne as (a), except for northeasterly

general wind.

Under the southwesterly general wind westerly surface wind exists to the

north of the Ishikari Bay. The area of westerly surface wind seems to extend

from the northern Japan Sea side to the south as the general wind direction turns

from S to NW (see Figs. 5(e)-(h)). This phenomenon appears both by the change

of general wind direction and by the influence of Shiribeshi Mountains which

310

(c)

y

Environmental

e{iiig(ll

"fo

'

Science,

:ajnyNi'Ito' ''

xgsr<---

e

Hokkaido

c -S} "-V'N'

fiy

VoL 6, No. 2, 1983

"'.-.' -"tg.. "))'..

¥//iii'k'.///.rf,-,,,X'-"-.9$izaNi,,.pt-tFfr "g'iig

･N .' --s-. -" s- ltor '' m N v "120Q- va eoo-l2oo wa 400.- 800 O 100km

v

(d)

y

Fig.

form an obstac]e

The wind dlrections

about 90 degrees

the northern area

explain a part of

the surface wind

ln wmter.

?D

--

xxg,.tt/:sililltillilSl;llliiiiierII¥i i; x?x',iii.iiill.get--"E-th.

s,`-W. xeq

R- g",S. (c)-(d)Sameas(a),except general wind,

in the case of general '

in the center or

as the general wind

show hardly any change

the phenomenon shown directions of southern

t

l 21lj

xii,ililii, :'k,>.iiigiii･.=.', k. I:

T--W=L'r' ' Di:.te>.

.--- -, -e

SE

---.ge-.x.ttsgayZ21

Egl.;#$.v vx:g

J("i

x-.X geNN

-m .E lli2oo- g 100km ag400- : o- for easter}y andrespeetively.

winds.

south of Is

turns from W to (see Figs. 5(g)

by Owacla and area of Ishikari

x ts9 caS.xH ;/･l,sgelts`-fd'

N. Xti"s"aiag"rg x x-7snS' .Q., x

v

eoo=noo

eoo 400 southea$terly

hikari Lowland change for

NW, although those in and (h)). This fact can

Yoshino (1971), in which

Lowjand are unstable

(e)

y

(f)

. 4e ".il.IIII$su- $K/)t?i!l¥lii'-ij･i-

y gtskyllj, ,

Fig. 5. (e)-<f) Same as (a>,

general

2) Eastem PaciZfic siJde a.nd

Under the southerly wincl system,

beRt to NE by the Akan-Shiretoko

(1973). Furthermore, southerly wind

ranges (Fig. 5(e)). Uncler the

along £he mountains to the north of Near Rausu, on the east of

mcreases

Air Current System in Hol<1<aido

re fo.. ,･- ,,. s ij e ..'`-ei55iiili'e! "

...

.,,;6illil$gi;;.eg,,w,,,,,fiig<l/,/",ll,/$'$.ite"tw", a/g M,s, ,¥gi,',,,f, ,' ,di:'i"i,,

, m '"iiptIT,"f{i;s,, ll 1200- va eoo-noo

.At e-/OOkm ge400-･eOO [] O-40Q

ro .fo - g.ij";g -SW

l.th.2tg/i"nS'./gi'S'/k･-ptS-', tk!"i<ff

illi{,

g,sL )ils

lfiii

v

v

311

l'" iiliil)/! =i2omo-

ge eoo-12oo 100krn suAOO-eOO ' VO-400 except for southerly ttnd southwesterly wind, re$pectively.

Abashiri Dist7'ict

the fe.q.ture in which the stream lines are

Mountains corresponcls to [he report by Owada

seems to blow through the two mountain westerly fiow (Fig. 5(g)) the wind tends to blow

the southeastern side of Hokkaido. Shiretoko Peninsula, it is clear that the wind speed

under the wester]y or northwesterly general wind. The strong wind blow-

312

(g)･

e

(h)

{ tS`k..sfii>?tttt.Q

Fig. S. (g)-(h) Same as (a), general wind,

ing here is referred to as "Rausu-Kaze"

wind occurs under particular weather

suitable for the generation of strong

(see Figs. 5(g) and (h)).

On the Okho£sk Sea side nearrelatively strong and the stream line

wind. These results are used for

studied by Yoshino et al. (1972).

the wind speed increases when the

Environmenta! Science, Hokkaido Vol. 6, No. 2,

) SS - ij'"e W

t.ept.s-/lg,,i,1,g･i/¥. //-.--.f,,i,,-g ,,,st

a'.'.- ":'--' Hp . ' ' ' Ts mli:k2"le ' -i2oo- XiS}eq ta eoo-12oosc'.a -/ookm g400-eoo a o-4oo

bti)1[>,s.-:. Nw"""

N $gij ･･si."ilii}lilltsZ.2...NtstiLa' Y

eet el,tSN

v r's

x N x

usts･ .,,ksSl

i¥t¥ X9" -

1983

℃

r-- m w" - 1200- : 600-1200 lookm g4oo-eoo G o-4oo except for westerly and respectively.

by Arakawa (l969,

condi£ion, the wind under these

Abashiri, the southerly

seems to be clear in the

the explanation of the

However,

general wind becomes

v

g

northwesterly

1971). Although the topography is assumed to be

general wlnd directions

or norther]y wind is

case of £hese strong southerly strong windat Mombetsu, west-northwest of Abashiri,

westerly or easterly. It

Air Current System in Hokkaido 313

is suggested that the wind system <air current system) changes in £his vicinity.

3) To,eachiDistrict

As dlscussed in the previous paper, (Kato, 1982), the features of westerly fiow,

in particular in the "l)ol<achl P]ain also exist clearly in the figttres (Fig. 5(g)).

Especially, it is interesting that the main flow towarcl the convergence zone near

Karikachi Pass changes from southerly to nortlaerly as the general wind direction

varies from W to NW. The features of Hidaka-Shimokaze reported by Arakawa <l969, 1971) aye net

marked clearly except for the convergence stream line near Ural<wa, on the south-

west of the Hidaka Mountains (Figs. 5(c) and (d)), because the area of such strong

winds is small both in space and ln time.

4) Klae74ontatsuncti Lozuland

The annual mean wind speed at Suttsu, to the north of Kuromatsunai Lowland,

is the strongest in Hol<kaido, ancl Tanaka (1958) showed that the wind speed

lncreases especially under £he southerly general wind here. From the figures suchfeatures are clearly marked, and furthermore, the strong winds are also recognized

under the northerly general winds. These results show the funneling effect of the

topography.

As mentioned above, the wind distribution patterns or the features of air cur-

reRts are clarified by this analysis. It is one of the outstanding features of this

method because the wincl pattems are obtained for a]] clirections of genera] winds.

b. Regional Characteristics of Air Current

Table 4 shows the wind speed at 22 meteorological stations when the general

wind speed has maximal values throughout the year for each direction. For each

statioR these va]ues are normalized xrsTith the maximum wind speed correspondiRg

to 10. In the table the normalized values ]ess than 6 are showR by "-", aRd

are regarded as weak wind speed. In other words, this table shows the general

wind direction when a strong surface wind occurs at each statioR. To examine

these results, the first five maxlmum wind speeds are selected from the daily data

of 1979 and 1980, i. e. the total is 10 cases, for each station. The corresponding

geneyal wind directioRs for each case are shown by circles in Table 4. Thick solid

circ]e indicates that the strong wind occurred more than two times in that general

wind direction.

At ail stations these strong winds tend to occur under the general wincl direc-

tion at which the normalized surface wind speed is more than 7, aRcl this table

seems to show the actual cases. AIthough the result at Wakkanai has some errors,

it depends on the fact shown by Narita and Masuyama (i955) that the surface

wind changes by the force of pressure gradient even if the dlrection of the pressure

gradient remains unchanged. HeRce it is iRteresting to note that the results of

Table 4 also c]ose]y resemble the figures for the probability of excess of wind speed

Eor each station shown by Murakami et al, (1979).

314

Table 4.

Environmental Science, Hokkaido VoL 6, No. 2, 1983

Calculated wind speed for each general wincl direction on 22 meteorologicai stations when the general wind speed has

the inaxhnuni value (pre$umed) for each clirection during 1978. For each station the values are normalizecl a$ tiie maximum wind spee6E eorresponds to 10, and the values }ess

than 6 are shown by "--". Clrcles indicate the generai wind

clirections corresponding te the first five maximum wincl speeds (daily data) for each statien both in 1979 ancl 1980.

Sqtiares shows that the strong wincl occurred more than two

times at tl)at general wincl clirectien

1 2

GENERAL3 f.....5 ..6

Wl,ND

78 8 10

79 - IZI

78 @ Lgl,

8

88 @ PIi･

---- 7

9@ 9 Q)

89

g'D ll"9'j

@e o-

DIRE(ITION 9 !O 11 12 13 14 I5 l6

WAKKANAIKI'IrAMI E.

ASAH,IKAWAHABOROSAPPOROIWAMIZAWAOTARUKUTCHANSIJTTSU

OHMUMOMBETSUABASHIRI

NEMUROKUSHIROOBIHIRO

HIROO'roMAKoMAIMURORANURAKAWAOSHAMANBEHAKOI)ATEESASHI

i.z

8

o

@(s)

7

@

o

7

Cl)

7

7

8

o@

-- 89-･- O- 7-- --@-- 8

-- -- 77 -･- 7 87- -- ---- -- --- 7

o ---- --- @

-- 777 - CD

999 (!l)

9 le9 10e ---9

8779979899

-- 79@

-e-@

7

@

@

ttt..tl

8

8

e

isi

eo

@@

F..gl]

9

8

8

@l.st

e@

7

8@ o 9 l9]

9 9 aj

op

7 9

9 op

8

89 10 (b (gi} igl

the

to

areare

NW or of ' whicother

also

(g)

e10

9

e

8

[en

8

9

8

8

10

8

8

M7

¢

8

o8

(ll)

F.g]

9

o7

10

o

Sincetheresu]tsrepresentedinTable4seemtoexpress regionalchar-acteristics of wind, these values are used for the c]uster analysis determine the

regions with similar features of wind. Three clustering regions shown in

Fig.6(a).Averagedfeaturesoiwindvariationforeachgroup showninFig.6(b>, i.e., the wind speed increases when the general wind direction is about 1)

SorNforthegroupI,2)SWorNEforthegroupII,and3) SEforthegroup III, respectively. It is worthy of note that the boundary wind variation

type between two groups exists in the north of Ishikari Plain h was also

revealedinthepreviouspaper(Kato,1982).Furthermore,the boundaryofwindvariatioRtypeisc]earlymarl<ednearMombetsuwhichis notedinthe

(a)

p

+:e:o:

l

II

m

Air

v

s

ts

a-

Current

{LXE3kKigif= tt fit %s ･

System in

e+ e?ee

E.,i iieil2g

ts#?ee

Hold<aido

e+e e+++

+

S- ."+ w

+

Sko- -- =-++S E--. ' ts-Iill,Si +

-trewi.."..s- .+

£E -.t.

if=L agCP$

KvdS¥l.iv. e"

p

v

-･-'n-ff -e']･rm s' n'- l'

'u U dios .v

- ee

ct' it

+ +..1:tgg

,,trdrtttsxtLrS7Tcbp.

:- +

9

O 100km

mM t200-

ee eoo-nooX 400-800

O O-400

315

?

(b) aLl.l 10tumuut

ailli

) 5awNxi2o

s

N

Nx NxX

xx !X"x-'x //XN

/xl

..!;,v'x

x/t/

E

-/A

ls1

× NN '

' Xl ' h- N. 1' ×'vt･-'

s

r'X Z1 / / !

N.tt

L"

1

.><-'NN

X NI 'x

'x. ll

'x 'NM

"i

Fig.

16123456789 10 11 12 13 14 15 16 1 GENERAL WIND DIRECTION6. (a) Regionalizat/ion clepending upon the clustering of the

calculated wincE speed for eacl] general wind direction,

and (b) the charaeterist'ics of "rincl v'ariation for each

group. See the text for detai],

prevlous sectlon.

Indeed Fig. 6 closely resembles Fig. 7 of Kato (1982) in spite of the difference

of the method. This fact indicates tlaat each area with homogeneous regiona]ity

of wind variation type exist, for example, Ishikari Plain, Kamikawa Basin, Kuro-

matsunai I.owland and so on, although the boundaries are not clear in inany cases.

These results will be ttsed not only for the explanation of the distribution pattern

of the other c]imatic e]ements l]ut also for the study of wind energy, for example

in reported by Komine et al. (1980).

316 Environmental ScSence, Hokkaido VoL 6, No. 2, 1983

5. Cenclusions

In order to clarlfy the various features of air current systems, AMeDAS data

are ana]ysed by means of the original method using PCA. The air current maps

obtained in this study correspond to the wind pattern when the general wind

speed in each direction has the annual maximum value, and reflect the actual wind

patterns. From these results, the structure and the extension of the air currents

for each general wind direction were revealed.

Using the calculated wind speed at the stations for each general wind direction,

the regionality of wind was clarified by means of c]uster aRalysis. The boundaries

(transitional zones) of the wind variation type agree with the ]imitation of the

extent of some air currents.

The regiona] characteristics of air current system in I'Iokkaido are mainly

governed by the speed of westerly general winds. Secondly characteristlcs depend

on the fact £hat the topography surrounding the region is suitable for generatingthe strong air curreRt under the southeasterly or the southwesterly genera] winds.

Such characteristics agree well with the previous study (Kato, 1982).

By these methods, wind distribution patterns determined by synoptic scale

pressure field and the major topographical features alone can be delineated, and

the noise such as background fluctuation of wind peculiar to each station or the

wind related to the heat effect are removed. This is the outstanding merit of

these methods.

Hereafter, it is necessary to investigate the wind pattern under the weak

pressure gradient, i.e., the wlnd pattern depend on the heat effect. Furthermore,

these methods may be applied to the other regions, and the propriety of the results

should be a]so examined.

Acknow!edgeraents

The author wishes to thank Prof. Hiroshi Kadomura, Graduate School ofEnvironmental Science, Hold<aido University, for his helpful suggestions, encourage-

ment and guidance in the course of this work. The author also wishes to thank

Assoc. Prof. Hidenori Takahashi, Graduate School of Environmental Science,Hokkaido University, for his helpful discussion and advice. Special thanks is also

given to Prof. Katsuhiro Kikuchi, Department of Geophysics, Hol<kaido Uni-versity, for his friendly aclvice. The author must not forget the assistance from

the sections of Japan Meteorological Agency for supplying the AMeDAS data tapes.

This paper is a part of the doctoral dissertatioR submited to Hokkaido Uni-

versity. Numerical computation was done at the Hokkaido University Computing

CeRter. The original data were compiled at Tokyo University Computing Center.

References

Arakawa, S. (1969); Climatological and dynamical studies on the local strong winds, mainly

in Hokkaido, Japan. Geopt)hys. Mag., 34, 359-425,

Air Current System ln Hokl<aido 317

Arakawa, S. (1971): On the local strong winds. 7"enki, 18, le3-115".

Barnett, T. P. (1977): rl'he principal tiine and space scales of the Paciflc trade wind fields. .ll

Attnos. Sci., 34, 221-236.

Ilnfield, D. B. (1981): Annual ancl nonseasonal variability of monthly low-level wind fields over

the Southeast'ern Tropical I'acific. Adon. IVea. Rev., le9, 2177-2190.

Hardy, D. M. (1978): Principal components ttnalysis of vector wind measurements. .Jl /!ppl.

A4keteo,;, 1.7, 1153-1162.

Kato, l'{. (1982): Regionality of surface wind in T{ol<kaido: An analysis by means of PCA.

En2,i7wn. Sci., Ifokfeaido, 5(2), 293-304S"*,

Kawamura, T. (1963): Surface wincl distribution in Hokl<aido in winter. iL A4et. Res., 15, 533-

537**.

Kawamura, T. (1966): Surface wind system over central Japan in the winter season. Geogr.

Re7,. ,ftipan, 39, 538-554*".

Komine, K. Mural<ami, S., Shibata, H. and Matsuno, N. (1980): Prediction of surface wind

velocity at any point, using regression analysis of topographic effects on wincl velocity.

Cl'ettki, 27, 849-861'.

Murakami, S. (1979): :l'Vie slucU>, on the geogi'aphical disti'ibt(tioit and the seasonal 'i,aiAiation of'

zvind eneig3, in tJaPan. Committee of Study on the Geographical Distribution and the

Seasonal Variation of Wind Energy in Japan, 163p*.

Narita, G. and Masuyama, R. (1955): On the influence of topography upon the wind direction

at Wakkanai. ,L A4k't. Res., 7, 797-799'i:.

Owada, M. and Yoshino, M. M. (1971): Prevailing winds in the Ishikari Plain, Hokkaido.

Geogr. ReT,. .Jdpan, 44, 638-652**.

Owada, M. (1973): Prevailing winds in the Konsen-Genya, Southeastern Hokkaido, Geogr. Re'u.

.lbpan, 46, 505-515*".

Tanaka, '!'. (1958): Some characters of gale in Suttsti District, Hokkaido. J. Adfet. ,Res., 10, 596-

602*.

Wakamatsu, S. ancl l'{Iatano, S. (1981): Local wind analysis using P.C.A.. .J. ,IZipan Soc. .f'lir

llollut., 16, 379-386*".

Yoshino, M. M., Hoshino, M. and Owada, M. (1972): On tl]e local wind distribution and the

wind break density in the Shari-Abashiri regien, Hokkaido. ,L Agr. Met; ,Jdpan, 27, 145-

152.

* in IIapanese.

** in Japanese with English abstract.

Abstract

In orcler to clarify the various features of air current systems, AMeDAS clata are analysed

by means of an original. method using PCA. The air currents shown on the maps correspond

to the wind patterns when the general wincl speed in each direction has the annual maximum

value, and refiect the actual wind patterns, From these results, the structure ancl the extension

of the air currents for each general wincl direction were revealed.

The regional characteristics of air current system (wind variation type) in ffokkaido are

mainly governed by the speed of westerly general winds. Secondary characteristics depend on

the fact that the topography surrounding the region is suitable for generating the strong air

current under the southeasterly or the southwesterly general winds.