Extraordinary Persistence of Foehn Observed in the

Journal of the Meteorological Society of Japan, Vol. 80, No. 4, pp. 579--594, 2002 579

Extraordinary Persistence of Foehn Observed in the Hokuriku District of

Japan in the 1999 Summer

Hiroyuki INABA, Ryuichi KAWAMURA

Department of Earth Sciences, Toyama University, Toyama, Japan

Takahiro KAYAHARA

National Research Institute for Earth Science and Disaster Prevention, Tsukuba, Ibaraki, Japan

and

Hiroaki UEDA1

Meteorological Research Institute, Tsukuba, Ibaraki, Japan

(Manuscript received 12 November 2001, in revised form 21 April 2002)

Abstract

An unusually prolonged foehn was observed at Toyama in the Hokuriku district, located on the coastof the Sea of Japan of central Japan, from 30 July to 3 August 1999. A 5-day foehn is extraordinarilypersistent and was not observed during the 24-year period 1975–1998, which makes the 1999 summerquite exceptional. While an anticyclone east of Japan is intensified over the 5-day period, a combinationof the anticyclone with a rapidly developing cyclone over southeastern Siberia and a typhoon migratingnorthward into the East China Sea induces a definite intensification of southerly geostrophic winds overcentral Japan, a favorable condition for the occurrence of unusually prolonged foehn.

Observational and model results show that on a monthly mean basis, a noticeable east-west pressuregradient around Japan is present due to the prominence of the Pacific and Japan (PJ) teleconnectionpattern in response to tropical convective heating, resulting in reinforcing southerly geostrophic windsacross central Japan. Once large-scale circulation anomalies are initiated and sustained as a result ofsuch an extratropical response, the development and movement of adjacent synoptic-scale disturbancesare largely regulated by those anomalies. A combined effect of the excitation of the PJ pattern and asso-ciated synoptic-scale disturbances is crucial for the extraordinary persistence of foehn along the coast ofthe Sea of Japan. The PJ pattern that appeared in July 1999 has an unusual geographical location andconfiguration because enhanced cumulus convection over the warm pool region of the western NorthPacific is significantly displaced about 20� westward and 2–3� northward, compared to the typical hotsummer that Japan experiences. This displacement is presumably attributed to a similar shift of dis-tinctive warm SST anomalies over the warm pool region.

1. Introduction

The foehn phenomenon, one of the intriguinglocal winds appearing frequently in the moun-tain region as well as the bora, has been in-vestigated by a number of meteorologists andclimatologists around the world. According to

Corresponding author: Ryuichi Kawamura, De-partment of Earth Sciences, Toyama University,3190 Gofuku, Toyama 930-8555, Japan.E-mail: [email protected]

1 Present affiliation: Institute of Geoscience, Uni-versity of Tsukuba, Tsukuba, Ibaraki 305-8571,Japan.E-mail: [email protected]( 2002, Meteorological Society of Japan

Yoshino (1976), who reviewed local winds, es-pecially bora, the definitions of the foehn andbora are made simply or practically as follows:the foehn is a downward wind on the leewardside of a mountain range. When it blows, theair temperature on the leeward slope rises. Thebora also is a downward wind on the leewardside of a mountain range, but when it commen-ces, the air temperature on the leeward slopefalls. Such local winds depend strongly on theconfiguration of the land, the pressure field atthe level of the mountain peaks, and the atmo-spheric stability (e.g., Ikawa and Nagasawa1989; Saito and Ikawa 1991; Saito 1993). Theoccurrence of extremely strong foehns in mid-summer brings about the outstanding increasein daily maximum temperature and the de-crease in daily minimum humidity, eventuallydamaging a variety of crops including grains,vegetables, and fruits. A combined effect of thefoehn occurrence and the heat island phenome-non in urban areas also creates discomfort andincreases daily electricity consumption due tothe continuous use of air conditioners. Daily lifeand economic activities thus are vulnerable to

the occurrence of foehn that causes extremeweather.

Since the Toyama Plain in the Hokuriku dis-trict of Japan is surrounded by the Hida Moun-tains and other high mountains to the south,as shown in Fig. 1, it tends to frequentlyundergo salient foehns on the leeward slope,due to strong pressure gradient across centralJapan when well-developed extratropical ortropical cyclones pass through the Sea of Japan(e.g., Arakawa 1988). An extremely strongfoehn has been observed at Toyama from theend of July to the beginning of August 1999.Foehns prevail only a couple of days in mostcases, but this exceptional foehn predominatedlonger than at least 3 days and the daily maxi-mum temperature at Toyama reached 38.8�Con July 31. Was such persistence of foehn ab-normal? If so, what caused the persistentfoehn? Was the dominance of foehn confinedonly to the Toyama Plain? Was the frequentoccurrence of foehn significantly associatedwith a severe hot summer in Japan?

Our main purposes of this study are (1) toverify whether or not the prolonged foehn ob-

Fig. 1. Geographical map of central Japan analyzed in this study.

Journal of the Meteorological Society of Japan580 Vol. 80, No. 4

served at Toyama in the midsummer of 1999is an exceptional case in recent years and (2)to identify what dynamic processes make thefoehn persist longer than at least 3 days, usingvarious observational data and an atmosphericgeneral circulation model (AGCM). Section 2describes data used and the analysis procedure.In Section 3 we briefly document interannualevolution of the summertime occurrence fre-quency of foehn days observed at Toyama inrecent years. Section 4 shows regional featuresof surface air temperature, wind, and precipi-tation fields when the foehn prevails in the Ho-kuriku district. In Section 5 we examine whatsynoptic-scale disturbances in the vicinity ofJapan are crucially responsible for the persis-tent foehn. Section 6 examines how the anoma-lous circulations, which provide a favorablecondition for the persistence, are induced in themidsummer of 1999 on the basis of model re-sults. Discussion and concluding remarks arefound in Section 7.

2. Data used and analysis procedure

2.1 Extraction of extreme foehnsBased on the Japan Meteorological Agency

(JMA) observational station data during theperiod 1975–1999, we defined a pronouncedfoehn, seen at Toyama city in midsummer(July–August) on a daily basis, using the fol-lowing four criteria:

i) daily maximum temperature is greaterthan 35�C,

ii) daily minimum relative humidity is lessthan 45%,

iii) daily mean wind speed is greater than3 m s�1, and

iv) daily most frequent wind direction is SSW,S, SSE or SE.

The above second and third criterions are quitereasonable because the foehn is characterizedby unusually dry and strong winds on the lee-ward side of the mountain range. Since it is an-ticipated that the foehn contributes frequentlyto the occurrence of unusual hot weather ex-ceeding 35�C, the first criterion is also in-troduced. We further applied the fourth crite-rion to the definition of foehn because theToyama Plain is surrounded by high mountainsto the south, and strong southerly winds onthe leeward side are an important indication of

foehn in this region. According to the aboveclassification, we identified the number of foehnday occurrences for each midsummer and ob-tained interannual variation of the summer-time occurrence frequency of foehn days atToyama from 1975 through 1999. Note that wedid not consider any foehns that occurred be-fore 1975, owing to the lack of daily most fre-quent wind direction data.

2.2 Observational dataTo clarify regional features of foehns in the

1999 summer, we use surface air temperature,wind velocity and direction, and precipitationdata derived from 778 AMeDAS (AutomatedMeteorological Data Acquisition System) sta-tions of JMA on mainland Japan, and sondedata from two selected stations: Wajima sta-tion, the nearest to Toyama, and Hamamatsustation on the coast of the Pacific Ocean. As thenext step, to examine anomalous circulations inthe vicinity of Japan relevant to the extraordi-nary persistence of foehn, we use the GANAL(JMA global objective analysis) data with aspatial resolution of 1.25� in longitude and lat-itude and 3-hourly infrared equivalent black-body temperature (TBB) with a spatial resolu-tion of 1� � 1� derived from the Japanesegeostationary meteorological satellite, provid-ing an appropriate measure of anomalous con-vection in lower latitudes. Also used for thecorresponding period are 10-day mean sea sur-face temperature (SST) data of JMA, with aresolution of 1� in longitude and latitude overthe western North Pacific from 100�E to 180�

between the equator and 60�N. We furtheruse the National Centers for EnvironmentalPrediction/National Center for AtmosphericResearch (NCEP/NCAR) global atmosphericreanalysis dataset; monthly outgoing longwaveradiation (OLR) data with a spatial resolutionof 2:5� � 2:5� observed from NOAA satellites;and the Global Ice and SST (GISST) dataset(Parker et al. 1995) for a comparison betweenthe 1999 summer and a typical hot summer.

2.3 Model simulationFor this study we used a T42 AGCM version

of JMA (JMA-GSM89) global model with 21levels (JMA 1993; Sugi et al. 1990). The hori-zontal resolution is equivalent to approxi-mately 2.8� longitude by 2.8� latitude. The pa-rameterization of deep convection in the model

H. INABA, R. KAWAMURA, T. KAYAHARA and H. UEDA 581August 2002

employs a modified Kuo scheme. In radiationprocesses, longwave cooling and solar heatingwith interactive cloud are taken into account.The land surface process utilizes a simple bio-sphere model. The surface exchanges employthe Louis scheme for surface moisture, heat,and momentum fluxes, and a Moller and Ya-mada level-2 closure model for vertical diffu-sion. Kawamura et al. (1998) have alreadyconducted an ensemble of three independent40-year integrations from 1955 to 1994, em-ploying prescribed global SSTs. These three in-dependent integrations are extended to August1999 using the SST data stated above to payspecial attention to the summer of 1999. Thethree parallel simulations are subjected to thesame SST boundary conditions. The ensemblemean process is expected to be very usefulfor inspecting the influence of anomalous SSTforcing from the tropics in extratropical circu-lation anomalies in the vicinity of Japan be-cause the internal variability of the extratropi-cal model atmosphere is considerably reduced.

3. Occurrence frequency of foehns

Based on the definition of foehn stated earlier,we extracted significant foehns observed atToyama during the midsummer season (July–August). Figure 2 shows the number of foehn

day occurrences at Toyama for each midsum-mer from 1975 to 1999 (bar), together with thetime sequence of the July–August mean sur-face air temperature (solid line). Foehn occurs 7days in the summer 1999, which is the largestnumber of the foehn days in the analyzed pe-riod. A sequence of 5 days out of the 7 days oc-curs from July 30 to August 3, 1999, as shownlater. The second largest number of the occur-rences is observed in the summers of 1978 and1985. The number of summers when no markedfoehns prevailed is 10. We see no definitefoehns during the years from 1979 to 1983.The relationship between the foehn’s occur-rence frequency and the mean temperature isnot necessarily significant. Actually, althoughthe 1994 summer is warmest summer in recentyears, and the mean temperature is 28.0�C, theoccurrence of foehn is only 3 days, while the1999 summer when foehns appeared most fre-quently is not an extremely hot summer. Whatwe would like to stress here is that the 1999summer is an exceptional case in the last twodecades in terms of the occurrence frequency offoehns.

Figure 3 shows the time evolutions of dailymaximum temperature, daily minimum rela-

Fig. 2. Time series of the number offoehn day occurrences at Toyama inmidsummer (July–August) during the25-year period 1975–1999 (bar). Alsoshown are the interannual variation ofthe July–August mean surface air tem-perature at the same station (solid line)and its 25-year average (dashed line).

Fig. 3. Time series of daily maximumtemperature (thick line), daily mini-mum relative humidity (thin line), dailymean wind speed (dashed line), anddaily most frequent wind direction (ar-row) at Toyama from July to August1999. Upward arrow denotes southerlywind. Opened arrow represents thewind direction that satisfies the fourthcriterion (see text).


tive humidity, daily mean wind speed, anddaily most frequent wind direction at Toyamafrom July to August of 1999. The occurrence ofsevere hot weather exceeding 35�C is 10 daysover the whole period. Seven days out of the 10days are classified as foehn days (July 30–August 3, August 19 and 31). The daily mini-mum relative humidity is remarkably low, lessthan 30%, from late July to early August, espe-cially on July 31 and August 2. The daily meanwind speed shows high values for almost thesame period, reaching about 6 m s�1 from July30 to August 3. Thus, extreme values for thesethree variables are concentrated mostly in theperiod from the end of July to the beginningof August. Moreover, the most frequent dailywind direction is SSW or S during this period.It is understood, as a consequence, that a se-quence of 5 foehn days (July 30–August 3)made the 1999 summer quite exceptional. Suchextraordinary persistence of foehn with 5 daysis not observed during the 24-year period1975–1998.

4. Regional features

This section focuses on the regional featuresof surface air temperature, wind speed and di-

rection and precipitation fields derived fromAMeDAS stations over central Japan duringthe period when the 5 sequential foehn days isobserved at Toyama. As is well known, localwind patterns over central Japan are gener-ally very complicated, due to the prominenceof various local winds such as land and seabreeze. To specifically highlight the regionalfeatures only relevant to the foehn, we com-puted daily mean surface air temperature anddaily mean surface wind speed and direction,resulting in the reduction of their diurnal cy-cles. Also considered are daily total precipita-tion distributions to infer what type of foehndominates. These regional patterns show a sig-nificant difference between the early and laterperiods of the foehn sequence of 5 days. Thus,we demonstrate the regional features of July 31and August 2 as a typical case for each period.

Figure 4 shows the distributions of dailymean surface air temperature and wind speedand direction in central Japan on July 31. Themost notable feature in this figure is that veryhot areas exceeding 28�C extend entirely alongthe mainland coast of the Sea of Japan, whichis quite different from the Pacific Ocean side.The temperature difference between the two

Fig. 4. Spatial distribution of daily mean surface air temperature and wind speed and direction incentral Japan on 31 July 1999. Contour interval is 2�C, and shading denotes regions greater than28�C. Reference arrow is 3 m s�1.


sides is about 3 to 4�C. Several cities (e.g.,Toyama and Komatsu) in the Hokuriku districtexperience severe hot weather, with greaterthan 30�C daily mean temperature. It turns outthat these hot areas along the coast of the Seaof Japan are accompanied by significantlystrong southerly winds over 3 m s�1. The Chu-goku district is also one of the regions wherethe foehn tends to appear frequently on thecoast of the Sea of Japan when strong south-erlies prevail across comparatively high moun-tains. These features suggest that the foehnpredominates not only at the Toyama Plain,but also in the Hokuriku and Chugoku dis-tricts. Very interestingly, the high temper-atures also cover the Kanto Plain, especiallythe Tokyo area, over 28�C, with strong south tosouthwesterly winds. Although we understandthat the heat island phenomenon due to large-scale urbanization contributes to such hightemperatures, this effect is beyond the scope ofthis study.

Figure 5 shows the distributions of dailymean surface air temperature and wind speedand direction in central Japan on August 2. Ina similar fashion, very hot areas extend over awide region on the coast of the Sea of Japan. Ifwe look at the Hokuriku district, Toyama and

Komatsu are observed to further heat up to32�C. A noticeable difference between July 31and August 2 is the local wind pattern in thewestern half of central Japan in terms of winddirection. An easterly wind component is moreevident on August 2 than on July 31, althoughsoutherly winds predominate on both days. Thedaily mean wind speed exceeding 5 m s�1 isalso observed in a number of stations on Au-gust 2. These changes suggest that the behav-ior of synoptic-scale disturbances in the vicinityof Japan significantly differs for each period.

Another remarkable difference between July31 and August 2 is the daily total precipitationdistribution on the Pacific Ocean side of centraland western Japan. Figures 6 and 7 display thedaily total precipitation pattern on July 31 andAugust 2, respectively. On July 31, weak pre-cipitation, with values of about 10–20 mm, isobserved in the Shikoku and Kinki districts,but there is no significant precipitation aroundthe Hida Mountains, which may imply thatthe foehn prevailing in the Hokuriku districtis identified with a dynamically-induced foehnaccompanied by no precipitation on the wind-ward side of a mountain range (e.g., Scorer andKlieforth 1959; Ikawa and Nagasawa 1989).In contrast, on August 2 heavy precipitation

Fig. 5. As in Fig. 4 but for 2 August 1999.


Fig. 6. Spatial distribution of daily total precipitation in central Japan on 31 July 1999. Large,medium, and small circles denote precipitation with values of 100 mm, 50 mm and 10 mm,respectively.

Fig. 7. As in Fig. 6 but for 2 August 1999.


exceeding 100 mm occurs, especially in theShikoku and part of Kyusyu districts, and wealso see somewhat weak precipitation less than50 mm on the Pacific Ocean side of Tokai dis-trict to the south of the Hida Mountains. Asshown in Figs. 4 and 5, hot weather conditionsin the Hokuriku district on a daily mean basisare more severe on August 2 than on July 31.Furthermore, hot areas over 28�C expand intothe Kinki district. It appears that the so-calledclassical foehn, which is a thermodynamicalfoehn accompanied by precipitation on thewindward side of a mountain range, contrib-utes to further reinforcement and extensionof severe hot conditions in the Hokuriku andKinki districts.

Figure 8 shows the vertical profiles of airtemperature and potential temperature at 09JST 31 July observed at Wajima and Hama-matsu sonde stations, which were selected asreference stations on the leeward and wind-ward sides of high mountain ranges in centralJapan, respectively. Both stations exhibit avery similar vertical structure in the layerabove about 2000 m in height, but a quite dif-ferent structure can be found in the boundarylayer, especially below 1500 m in height. Aconvective mixed layer is seen below about800 m in height at Hamamatsu station, but it

is capped by a very stable layer between 800 mand 1500 m. At Wajima station, in contrast, anextremely stable layer is established belowaround 800 m in height. Although there existsa remarkable difference in the stratification be-low 1500 m between the two stations, it is un-certain whether the foehn observed at Toyamaextends up to Wajima station in the Noto Pen-insula, about 100 km from the Hida Moun-tains. Unfortunately, we cannot make surewhich type of foehn, a dynamical or a thermo-dynamical foehn, was more dominant duringthe 5 days from July 30 to August 3 due to ashortage of sonde station data.

5. Synoptic-scale features in the vicinityof Japan

In this section we focus on synoptic-scale dis-turbances in the vicinity of Japan associatedwith the extraordinary persistence of foehn.Figures 9 and 10 illustrate the spatial patternsof twice daily 850-hPa height and wind vectorand TBB during the periods of July 30–31 andAugust 1–2, respectively. Taking a look at theearly period (Fig. 9), an intensified anticyclone,which is part of the North Pacific High, is ob-served to the east of Japan, whereas a rapidlydeveloping cyclone with a stationary compo-nent is located over southeastern Siberia of theAsian continent. We also have a cloud bandacross the Korean Peninsula at 21 JST 31 July,corresponding to a frontal system of the extra-tropical cyclone. It is obvious that the pressuregradient between the two systems across Japanbecomes pronounced, which implies the promi-nence of geostrophic wind just over Japan. Overcentral Japan the geostrophic wind has an evi-dent southerly component, eventually facilitat-ing the occurrence of foehn on the northernslope of a high mountain range, especially inthe Hokuriku district. Likewise, a typhoondevelops and moves northwestward along thesouthwestern periphery of the enhanced anti-cyclone system.

At the later period (Fig. 10) the anticycloneeast of Japan is still dominant, but the extra-tropical cyclone over southeastern Siberia be-gins to move eastward into the Sea of Okhotskand disappears rapidly, nearly concurrent withthe northward extension of the North PacificHigh. The typhoon, on the other hand, con-tinues to intensify and move northwestward

Fig. 8. Vertical profiles of air tempera-ture and potential temperature fromnear-surface to 6000 m in height atWajima station (solid line) and Hama-matsu station (dashed line) at 09 JST31 July 1999.


into the East China Sea. The cloud band struc-ture across the Korean Peninsula becomes lessclear with the weakening of the low pressuresystem over the continent by 21 JST 1 August,but on August 2 it is largely modified by theincreased moisture supply from the approach-ing typhoon. It is inferred that the enhancednorthward moisture transport at the lowlevelscontributes substantially to the occurrence ofheavy precipitation events over western Japan,especially Shikoku and Kyusyu districts, asshown in Fig. 7. A combination of the two majorsystems, i.e., the typhoon and anticyclone sys-tems, induces a new strong east-west pressure

gradient around Japan and enhances southerlygeostrophic winds across the high mountainrange.

We would like to emphasize here that thesoutherly geostrophic winds over central Japanare definitely intensified during both periods,due to the presence of strong pressure gradient,although synoptic-scale disturbances that in-duce such a pressure gradient differ appreci-ably for each period. This results in the ex-traordinary persistence of the foehn from July30 to August 3. A common feature in both peri-ods is the enhancement of the anticyclone eastof Japan. The strengthened anticyclonic circu-

Fig. 9. Spatial pattern of twice daily 850-hPa height and wind vector and TBB from 09 JST 30 Julyto 21 JST 31 July 1999. Contour interval is 30 m, and reference arrow is 20 m s�1.


lation blocks the penetration of the typhooninto the vicinity of Japan, and hence the ty-phoon is forced to migrate northward along thewestern periphery of the anticyclone. As a re-sult, a distinctive pressure gradient is inducedat the later period. The intensified anticyclone,moreover, might activate the stationary natureof the low-pressure system over southeasternSiberia at the early period, by preventing east-ward movement of the cyclonic disturbance.Even not so, the enhanced anticyclonic circula-tion provides a favorable condition for the for-mation of the marked pressure gradient acrossJapan between its circulation and a developingcyclonic disturbance over the continent. It

should be noted, therefore, that the intensifiedanticyclone east of Japan is responsible for theoccurrence of unusually prolonged foehn, in thesense that the system significantly influencesthe behavior of adjacent synoptic-scale distur-bances.

6. Large-scale circulation anomalies

In the previous section we showed how syn-optic-sale conditions in the vicinity of Japan fa-cilitate the extraordinary persistence of foehnobserved in the Hokuriku district. An impor-tant question that needs to be addressed is whysuch specific synoptic-scale conditions are in-duced and maintained. In particular, we should

Fig. 10. As in Fig. 9 but for from 09 JST 1 August to 21 JST 2 August 1999.


make clear a possible cause of the intensifica-tion of the anticylonic circulation east of Japanover the period. Of much interest is that its in-tensification commences several weeks beforethe occurrence of the exceptionally prolongedfoehn (figure not shown).

6.1 ObservationsFigure 11 shows the observed monthly mean

850-hPa height and wind vector and TBB

anomaly patterns for July 1999. Anomalies arerelative to the 1980 to 1999 average. A notablefeature is that an outstanding anticycloniccirculation anomaly is organized to the eastof Japan, while an anomalous cyclonic circula-tion dominates over an ocean area betweenthe Philippines and Japan. The pronouncedanticyclonic and cyclonic circulation anomaliesare accompanied by positive and negative TBB

anomalies, respectively. The center of theanomalous cyclonic circulation, which is essen-tially baroclinic in the vertical, is locatednorthwest of the center of intense convection.We can interpret the cyclonic circulation anom-

aly in the subtropics as Matsuno-Gill type re-sponse (Matsuno 1966; Gill 1980) of the lowertroposphere to enhanced convective heatingaway from the equator.

A pair of the above two circulation anomaliesis clearly identified with the Pacific and Japan(PJ) teleconnection pattern found by Nitta(1987). The development of the anticyloniccirculation, which is part of the PJ pattern,reinforces descending motion in the lower tro-posphere and brings about less cloud cover,resulting in the occurrence of a hot summer,especially in the northern and eastern parts ofJapan (e.g., Kurihara and Tsuyuki 1987; Ka-wamura et al. 1998).

It is crucially important that the PJ patternexcited by convective forcing generates a strongeast-west pressure gradient across Japan, be-cause southerly wind anomalies prevail justover central Japan. The persistent PJ pat-tern affects the development and movementof synoptic-scale disturbances. Actually, a ty-phoon develops and moves northwestward tothe south of Japan under the influence of theenhanced anticyclonic circulation east of Japan,as indicated in Fig. 10. It is quite possible thatPJ-related synoptic-scale disturbances give afavorable condition for the frequent appearanceof foehn on the coast of the Sea of Japan whenthe PJ pattern is predominant. We thus antici-pate that a combined effect of the excitation ofthe PJ pattern and associated synoptic-scaledisturbances is crucial for the occurrence of theexceptionally prolonged foehn from July 30 toAugust 3. As for August, the anticyclonic circu-lation east of Japan becomes weak in associa-tion with attenuated convective heating northof the Philippines (not shown).

Figure 12 also shows the observed SSTanomaly pattern for July 1999. An importantindication is the presence of significantly warmanomalies exceeding þ0:5�C over the tropicalwestern North Pacific east of the Philippines. Itis very likely that these warm anomalies leadto the enhanced convective activity north ofthe Philippines, which will be confirmed laterusing the model. Cool anomalies over the cen-tral equatorial Pacific indicate a signature of aLa Nina state. Another feature is that verywarm anomalies exceeding þ1:5�C expand overa wide area east of Japan, which coincides wellwith the positive TBB anomaly area (Fig. 11).

Fig. 11. Observed monthly mean 850-hPa height and wind vector and TBB

anomaly patterns for July 1999.Anomalies are relative to the 1980 to1999 average. Contour interval is 10 m,and negative height values are denotedby dashed lines. Heavy and light shad-ings denote significant negative andpositive TBB anomalies, respectively.Reference arrow is 9 m s�1.


Kawamura et al. (2001) pointed out that no-ticeable warm anomalies from the vicinity ofJapan to the central North Pacific, appearingwhen Japan undergoes a severe hot summer,are basically caused by the change in surfaceheat flux, especially short-wave radiation andlatent heat fluxes, relevant to the prominenceof an anomalous anticyclonic circulation east ofJapan. Their idea can apply to the above fea-ture quite reasonably.

6.2 AGCM simulationNext, we examine the question of whether an

AGCM can simulate the basic features of ob-served low-level circulation anomalies over thewestern North Pacific and East Asia, especiallythe vicinity of Japan. If the model is successfulin simulating their essential features, the ori-gin of the observed anomalous circulations isattributed to anomalous SST forcing in low lat-itudes. Figure 13 shows the model-simulated850-hPa height and wind vector and precipita-tion anomaly patterns for July 1999. Note thatthe simulated anomaly fields presented hereare derived from the ensemble mean of thethree parallel experiments. On the ensemblemean basis, the model simulates increasedprecipitation in the vicinity of the Philippines,in intimate association with the underlyingwarm SST anomalies, which indicates a signa-

ture of strong western North Pacific monsoon(Murakami and Matsumoto 1994), but the cen-ter of the simulated enhanced convection areashifts to the south compared to that observed(see Fig. 11 again). Another notable feature isestablishment of an anticyclonic circulationanomaly east of Japan. Japan is located on thewestern periphery of this anomalous circula-tion and covered by southerly wind anomalies,which is similar to the observations. The model,on the contrary, fails to reproduce a significantcyclonic circulation anomaly to the south ofJapan, although a weak cyclonic circulation issimulated just over the Philippines. This mayresult from a systematic error that the modeltends to underestimate tropical precipitationover the warm pool region of the western NorthPacific (Sugi et al. 1995). Another possible rea-son is that the AGCM simulation does notincorporate large-scale air-sea feedback pro-cesses. For instance, negative SST anomaliessouth of Japan (Fig. 12) might be interpretedas a result of anomalous surface heat flux and/

Fig. 12. Observed monthly mean SSTanomaly pattern for July 1999. Contourinterval is 0.5�C, and shading denotesregions of positive values.

Fig. 13. Model-simulated monthly mean850-hPa height and wind vector andprecipitation anomaly patterns for July1999, based on the ensemble mean ofthree parallel simulations. Contour in-terval is 10 m, and reference arrow is3 m s�1. Heavy and light shadings de-note regions greater than þ1 mm day�1

and less than �1 mm day�1, respec-tively.


or wind forcing relevant to convection-inducedanomalous circulation. The SST decrease mayin turn suppress convection. Yet, we disregardsuch feedback processes in the model.

Although systematic errors such as local pre-cipitation distribution are indeed present in themodel, enhanced southerly wind anomaliesalong the western periphery of the anomalouscirculation centered east of Japan, relevant tothe occurrence of the extraordinary prolongedfoehn, are basically simulated in terms ofanomalous circulation field. This indicates thatthe anomalous SST forcing is responsible forthe establishment and maintenance of the an-ticyclonic circulation anomaly east of Japan.Since only the three parallel simulations wereperformed, we cannot fully remove the internalvariability of extratropical model atmospherethrough the ensemble mean process. Thus, itsinternal variability may also influence such acirculation anomaly. It is worthwhile to notehere that the simulated anticyclonic circulationis accompanied by the underlying warm SSTanomalies, as shown in Fig. 12. We think itvery unlikely that the in situ warm SST anoma-lies generate an anomalous anticyclonic circu-lation in the lower troposphere through localair-sea heat exchanges. As previously stated, itcan actually be interpreted that the warm SSTanomalies are not the cause, but the result, of awell organized anticyclonic circulation anomaly(Kawamura et al. 2001). We are aware, ofcourse, that local air-sea interactions possiblymodify the anomalous anticyclonic circulation.

From the common features between themodel-simulated and observed anomalies, it isconcluded that the anomalous SST forcing fromtropics induces the PJ-like response of extra-tropical atmosphere over the western NorthPacific, resulting in reinforcing southerly geo-strophic winds across central Japan throughintensification of an east-west pressure gradi-ent. Once large-scale circulation anomalies areinitiated and sustained as a result of such anextratropical response, the behavior of asso-ciated synoptic-scale disturbances is largelyregulated by those anomalies. We emphasizethat a combination of the excitation of the PJpattern and the PJ-related synoptic-scale dis-turbances can easily provide a favorable condi-tion for the extraordinary persistence of foehnalong the coast of the Sea of Japan.

7. Discussion and concluding remarks

An important problem remains unsolved.Why is only the 1999 summer exceptional interms of the occurrence frequency of foehn, asindicated in Fig. 2. Although we pointed outthat a combined effect of the dominance of PJand associated synoptic-scale disturbancesfavors the extraordinary persistence of foehn,we have given no definitive answers to theabove question.

According to Kawamura et al. (1998), thePJ-type response to anomalous convective heat-ing over the Philippine Sea is crucial to thefrequent occurrence of extreme summers inEast Asia, Japan particularly, in the past twodecades. They extracted six typical hot sum-mers (1981, 1984, 1985, 1988, 1989 and 1990)in terms of how large-scale anomalous circu-lations that cause severe hot summers arestrongly regulated by tropical forcing. Based ontheir classification with respect to the typicalhot summer, we thus present, in Fig. 14, the

Fig. 14. Composite anomaly patterns of850-hPa height and wind vector andOLR fields for July in a typical hotsummer, based on Kawamura et al’s(1998) classification. Contour intervalis 10 m, and negative height values aredenoted by dashed lines. Heavy andlight shadings denote significant nega-tive and positive OLR anomalies, re-spectively. Reference arrow is 3 m s�1.


composite anomaly map of 850-hPa height andwind vector and outgoing longwave radiation(OLR) fields for July. Comparing this figurewith Fig. 11, the most notable difference is theconfiguration of the PJ pattern. A salient cy-clonic circulation anomaly, which is part of PJ,is centered on 25�N, 145�E, while an anoma-lous anticyclonic circulation is established to itsnortheast. A north-south pair of these two vor-texes generates easterly wind anomalies be-tween the two. We also see that the anomalouseasterly winds expand westward up to Japan.As already shown in Fig. 11, on the other hand,the PJ pattern that appeared in July 1999 ischaracterized by an east-west pair, rather thana north-south pair of the cells, and the geo-graphical location of its overall pattern is dis-placed to the west. This clearly results from thewestward displacement of an enhanced convec-tion area in the subtropics. The center of nega-tive TBB anomaly area (Fig. 11) is shifted about20� westward and 2–3� northward, compared tothat of the corresponding negative OLR anoma-lies. Thus, it can be interpreted that the west-ward displacement of intense convection areachanges the geographical location and configu-ration of PJ, eventually leading to the intensi-fication of a strong east-west pressure gradientacross Japan.

Another question may be raised: why did thewestward displacement of enhanced convectionarea occur in July 1999? The most likely causeis the extraordinary feature of SST anomaliesover the warm pool region of the western NorthPacific. Looking at Fig. 15, which shows thecomposite July SST anomaly map for the typi-cal hot summer, it turns out that significantwarm SST anomalies at low latitudes are in-dicated over the warm pool region around140�–160�E. It should be noted, however, thatwarm anomalies observed in July 1999 arepronounced to the west of about 150�E (see Fig.12 again). We anticipate that the westwardshift of the warm anomalies over the tropicalwestern North Pacific brings about a similarshift of the enhanced tropical convection in thatvicinity.

Major findings of this study are summarizedas follows. The occurrence of extreme foehnsobserved at Toyama in the Hokuriku districtreaches 7 days in midsummer of 1999, which isthe largest number of the foehn days during the

25-year period 1975–1999, using four objectivecriteria characteristic of foehns. A sequence of5 foehn days from July 30 to August 3 makesthe 1999 summer quite exceptional. The extra-ordinary prolonged foehn prevails, not only atthe Toyama Plain, but also along the coast ofthe Sea of Japan of central Japan, and asso-ciated local wind patterns show that there is aremarkable difference between the early andlater periods of the foehn sequence of 5 days. Atthe early period (July 30–August 1), the pres-sure gradient becomes pronounced between arapidly developing extratropical cyclone oversoutheastern Siberia and an intensified anti-cyclone east of Japan, whereas at the later pe-riod (August 2–3), a combination of the sameanticyclone and a typhoon migrating north-ward along its western periphery induces anew strong east-west pressure gradient aroundJapan. These synoptic-scale conditions, whichlead to definite intensification of southerly geo-strophic winds over central Japan, provide afavorable condition for the extraordinary per-sistence of foehn.

We also find from the simulation and obser-vation that on a monthly mean basis, a notice-able east-west pressure gradient across Japanis induced due to the predominance of PJ in re-sponse to tropical convective heating, and thepersistent PJ pattern influences the develop-

Fig. 15. As in Fig. 14 but for SSTanomalies. Contour interval is 0.1�C,and shading denotes regions of positivevalues.


ment and movement of adjacent synoptic-scaledisturbances, such as a growing typhoon andan enhanced anticyclonic circulation east ofJapan seen at both periods. We anticipate thata combined effect of the excitation of the PJpattern and associated synoptic-scale distur-bances is crucial for the occurrence of excep-tionally prolonged foehn. The PJ pattern thatappeared in July 1999 is unusual in terms of itsgeographical location and configuration. This isattributed to the westward displacement of anenhanced convection area over the warm poolregion of the western North Pacific; that is, thatarea is shifted about 20� westward and 2–3�

northward, compared to the typical hot sum-mer that Japan underwent. It is presumed thatthe westward shift of marked warm SSTanomalies over the warm pool region leads to asimilar shift of the intensified convection inthat vicinity.

Japan, its northern and eastern parts partic-ularly, again experienced a severe hot summerin 2000. Localization and enhancement of trop-ical convection are observed in July over thewarm pool region of the western North Pacificas well as July 1999. It seems that both summ-ers exhibit very similar anomalous states ofthe tropical western Pacific atmosphere-oceansystem. Associated with this, we also see thefrequent occurrence of foehn in the Hokurikudistrict in the midsummer of 2000. Furtheranalysis should also be made with respect tothe case of 2000.

Acknowledgments

Comments by two anonymous reviewerswere extremely helpful. This research was sup-ported by the Japan Science and TechnologyAgency, through the project study of disasterpredictions in global hydrologic processes; byGrants-in-Aids (11640430) of the JapaneseMinistry of Education, Science, Sports andCulture; and by the Mitsubishi Foundation forthe Promotion of Science.

References

Arakawa, S., 1988: An introductory survey of localcirculation. Note on Meteorological Research,163, 1–22, in Japanese.

Gill, A.E., 1980: Some simple solutions for heat-induced tropical circulation. Quart. J. Roy. Me-teor. Soc., 106, 447–462.

Ikawa, M. and Y. Nagasawa, 1989: A numericalstudy of a dynamically induced foehn observedin the Abashiri-Ohmu area. J. Meteor. Soc.Japan, 67, 429–458.

JMA, 1993: Outline of operational numerical weath-er prediction at Japan Meteorological Agency.NPD/JMA, 128pp.

Kalnay, E., M. Kanamitsu, R. Kistler, W. Collins,D. Deaven, L. Gandin, M. Iredell, S. Saha,G. White, J. Woollen, Y. Zhu, M. Chelliah, W.Ebisuzaki, W. Higgins, J. Janowiak, K.C. Mo,C. Ropelewski, J. Wang, A. Leetmaa, R. Rey-nolds, Roy Jenne and Dennis Joseph, 1996:The NCEP/NCAR 40-year reanalysis project.Bull. Amer. Meteor. Soc., 77, 437–471.

Kawamura, R., M. Sugi, T. Kayahara and N. Sato,1998: Recent extraordinary cool and hotsummers in East Asia simulated by an ensem-ble climate experiment. J. Meteor. Soc. Japan,76, 597–617.

———, T. Matsuura and S. Iizuka, 2001: Inter-annual atmosphere-ocean variations in thetropical western North Pacific relevant tothe Asian summer monsoon-ENSO coupling. J.Meteor. Soc. Japan, 79, 883–898.

Kurihara, K. and T. Tsuyuki, 1987: Development ofthe barotropic high around Japan and its asso-ciation with Rossby wave—like propagationsover the North Pacific: analysis of August1984. J. Meteor. Soc. Japan, 65, 237–246.

Matsuno, T., 1966: Quasi-geostrophic motions in theequatorial area. J. Meteor. Soc. Japan, 44, 25–43.

Murakami, T. and J. Matsumoto, 1994: Summermonsoon over the Asian continent and thewestern North Pacific. J. Meteor. Soc. Japan,72, 719–745.

Nitta, T., 1987: Convective activities in the tropicalwestern Pacific and their impact on the North-ern Hemisphere summer circulation. J. Meteor.Soc. Japan, 65, 373–390.

Parker, D.E., C.K. Folland and M. Jackson, 1995:Marine surface temperature: Observed varia-tions and data requirements. Clim. Change,31, 559–600.

Saito, K. and M. Ikawa, 1991: A numerical study ofthe local downslope wind ‘‘Yamaji-kaze’’ inJapan. J. Meteor. Soc. Japan, 69, 31–56.

———, 1993: A numerical study of the local down-slope wind ‘‘Yamaji-kaze’’ in Japan. Part 2:Non-linear aspect of the 3-D flow over a moun-tain range with a col. J. Meteor. Soc. Japan,71, 247–272.

Scorer, R.S. and H. Klieforth, 1959: Theory of moun-tain waves of large amplitude. Quart. J. Roy.Meteor. Soc., 85, 131–143.


Sugi, M., K. Kuma, K. Tada, K. Tamiya, N. Hase-gawa, T. Iwasaki, S. Yamada and T. Kitade,1990: Description and performance of theJMA operational global spectral model (JMA-GSM89). Geophys. Mag., 43, 105–130.

Sugi, M., R. Kawamura and N. Sato (1995): The cli-

mate simulated by the JMA global model. PartI: Global feature. Report of the National Re-search Institute for Earth Science and DisasterPrevention, 54, 155–180.

Yoshino, M., 1976: Local wind bora. University ofTokyo Press, 289pp.


Documents

Extraordinary Persistence of Foehn Observed in the