A Diagnostic Analysis on the Effectof the Residual Layer in Convective Boundary Layer Development near M ongolia Using 20th Century Reanalysis Data

HAN Bo,ZHAO Cailing,L¨U Shihua,and WANG Xin

Key Laboratory of Land Surface Process and Climate Change in Cold and Arid Regions, Cold and Arid Regions Environmentaland Engineering Research Institute, Chinese Academy of Sciences,Lanzhou730000

A Diagnostic Analysis on the Effectof the Residual Layer in Convective Boundary Layer Development near M ongolia Using 20th Century Reanalysis Data

HAN Bo?,ZHAO Cailing,L¨U Shihua,and WANG Xin

Key Laboratory of Land Surface Process and Climate Change in Cold and Arid Regions, Cold and Arid Regions Environmentaland Engineering Research Institute, Chinese Academy of Sciences,Lanzhou730000

A lthough the residual layer has already been noted in the classical diurnal cycle of the atmospheric boundary layer, its effect on the development of the convective boundary layer has not been well studied.In this study,based on 3-hourly 20th century reanalysis data,the residual layer is considered as a common layer capping the convective boundary layer.It is identified daily by investigating the development of the convective boundary layer.The region of interest is bounded by (30?–60?N,80?–120?E),where a residual layer deeper than 2000 m has been reported using radiosondes.The lapse rate and w ind shear w ithin the residual layer are compared w ith the surface sensible heat flux by investigating their climatological means,interannual variations and daily variations.The lapse rate of the residual layer and the convective boundary layer depth correspond well in their seasonalvariations and climatologicalmean patterns.On the interannualscale,the correlation coefficientbetween their regionalaveraged(40?–50?N,90?–110?E)variationsishigher than thatbetween the surface sensible heat flux and convective boundary layer depth.On the daily scale,the correlation between the lapse rate and the convective boundary layer depth in most months is still statistically significant during 1970–2012.Therefore,we suggest that the existence of a deep neutral residual layer is crucial to the formation of a deep convective boundary layernear the Mongolian regions.

convective boundary layer,residual layer,lapse rate,surface sensible heat flux,w ind shear

1.Introduction

The convective boundary layer(CBL)is the main manifestation of the atmospheric boundary layer(ABL)during the daytime(Stull,1988;Garratt,1994;Zilitinkevich,2012). Its development and maintenance have a direct influence on many atmospheric phenomena,such as cloud formation (Barthlott et al.,2010;Thornton et al.,2011)and pollutant distribution(Lin and M cElroy,2010;Banta et al.,2011). Therefore,the factors thataffect the grow th of the CBL have been discussed for many years(Stull,1988;Garratt,1994; Moeng and Sullivan,1994;Bianco etal.,2011;Maronga and Raasch,2013).

The buoyancy caused by the underlying heating is often suggested to be the most important factor(Moeng and Sullivan,1994;Maronga and Raasch,2013),which is easily accepted because itexhibits a sim ilardiurnaland annualvariation as the CBL depth,especially over continents.Furthermore,its global spatial distribution supports this theory because a deep CBL is often reported overarid or semi-arid regions.The dom inance of the surface sensible heat flux on the CBL has also been confi rmed by using numericalsimulation results(such as large eddy models).However,other factors such as the heterogeneity scale and the coherentstructure of turbulence can also significantly impact the localCBL development(Avissar and Schm idt,1998;Roy and Avissar,2000; Maronga and Raasch,2013).

Many factors can influence the depth of a CBL(Pan and Mahrt,1987;Bianco etal.,2011).Mostof these factors affect the atmospheric thermal or dynamic processes near the land surface(i.e.by changing the energy source of the thermals w ithin the CBL)(Lenschow and Stephens,1980).The thermals w ithin the CBL w ill continue to rise until reaching the inversion layer(or entrainmentzone),when their kinetic energy in the verticaldirection is totally consumed.Becausethe entrainmentprocess raisesthe top of the CBL,the thermal structure w ithin the inversion layer can affect the CBL evolution by modifying the entrainmentrate(Gentine etal.,2013b; Gentine et al.,2013a).However,such an impact may be no less important than that of the surface sensible flux because some early studies justdefine the inversion layeras a 0-order jump of potential temperature and the m ixed layer top buoyancy flux is a constant fraction of the surface buoyancy flux (Tennekes,1973;Garratt,1994).Moeng and Sullivan(1994) compared the development of a shear-driven and buoyancydriven planetary boundary layer(PBL)in a large eddy model; the effectofsurface heating overwhelmed thatofw ind shear. However,because the actualatmosphere ismore complicated than that in an idealized numerical simulation,the potential contributions of other factors on CBL development need to be reinvestigated.

The atmospheric layerabove the inversion layer isusually called the free atmosphere layer,w ithin which the turbulent movement can be neglected.When the stratification of the lower free atmosphere layer is neutral,it can be referred to as a residual layer because its characteristics(mean state and concentration variables)are generally observed to be initially the same as those of the recently decayed m ixed layer(Stull, 1988;Marsham et al.,2008;Huang et al.,2010;Freire and Dias,2013).In this type of CBL,when the potential temperature in the m ixed layer approaches the value in the residual layer,the inversion layer disappears and the thermals in the mixed layermove freely upward into the residual layer.Some of the literature refers to such a processasa coupling between the residual layerand CBL(Stensrud,1993;Fochesatto etal., 2001).Related works can be traced back to Stull(1976)and his classic schematic figure for the planetary boundary layer cycle(Stull,1988).

In recentdecades,a stable residual layercapping the CBL over the Sahara region(i.e.Sahara residual layer)has been reported via coordinated research fl ights over the Saharan heat low(Parkeretal.,2005;Marsham etal.,2008;Messager etal.,2010).These studies are more concerned w ith the effectof the residual layeron dusttransport;the effectofsuch a large-scale residual layer on the local CBL developmenthas not been well studied.Recently,Han et al.(2012)reported that when a neutral residual layer caps the CBL,the grow th of the CBL w ill be mainly determined by the lapse rate of the residual layer rather than by the intensity of the surface heat flux.However,their results are not conclusive because the number of observation cases used was rather small.Generally speaking,although the existence and the potential impactof the residual layeron the CBL developmenthave been mentioned,these results have notbeen verified in long-term studies oroverw idespread regions.

In this study,we focus mainly on the CBL development over the region of(40?–50?N,90?–110?E)near Mongolia (30?–60?N,70?–130?E),where most of the arid and sem iarid regions of East Asia are located.We choose this region mainly because of its deep CBL during the summer,butalso because a deep and neutral residual layerhas been observed at (39?28′N,102?22′E).The reanalysis data used and ourdefi nition of the residual layer for the reanalysisdata are introduced in section 2.The climatologicalmean features of the residual layer and their co-variations w ith the CBL at different time scales are presented in section 3.The major conclusions and a discussion are provided in section 4.

2.Data and method

2.1.The reanalysis data

By using radiosondes in fi eld observation experiments, the detailed vertical profi le of atmospheric variables can be obtained.The top of the CBL is usually identified from a jump of(virtual)potential temperature or the mass ratio of water vapor.However,such types of observation data are temporally and spatially lim ited.For example,global radiosonde data,such as the Integrated Global Radiosonde Archive(Elliottand Gaffen,1991;Durre etal.,2006),may have a coarse resolution in the vertical direction and vary among sites;thus,these datasets are not suitable for CBL research.Therefore,the datasets used in this study should presenta continuous CBL depth over a long period and cover a w ide region.Considering these criteria,we choose the 20th century reanalysis(20R)dataset.Its atmospheric boundary layer depth(and other variables in the surface layer)is provided every 3 hours,which is comparable to the time interval of a radiosonde observation.The variables on multi-pressure levels in 20R are produced every 6 hours.Details of the 20R datasetcan be referred to in the worksofCompo etal.(2011) and Saha etal.(2010).

The depth of an ABL(hhereafter)from 20R is calculated follow ing the non-local PBL diffusion scheme(Troen and Mahrt,1986;Holtslag etal.,1990;Holtslag and Boville, 1993;Hong and Pan,1996):

whereRibcis the criticalbulk Richardson number,U(h)and θv(h)are the horizontalw ind speed and virtualpotential temperature at the top of the ABL,respectively,andθvais the virtualpotential temperature at the lowest levelof the model. θsis the appropriate temperature near the surface:

where

bis an experimental constant,is the virtualheat flux at the surface,uuu?is the friction velocity,andLis the Monin–Obukhov length scale.Equation(3)is suitable only forneutralorunstable conditions[which is a condition for the formation of the CBL.The details and parameters used in the calculation ofhin 20R are provided by Hong and Pan (1996).

A clear relationship betweenhand othervariables cannot easily be identified from Eqs.(1)to(3),but it is clear thathis determ ined mainly by three factors:the sensible heat flux at the surface,the verticalshear of horizontalw ind speed and the profi le of the virtual potential temperature.Therefore, if 20R can generally describe the 3D thermodynam ic structure of the atmosphere and the surface turbulentheat fluxes in the study region,then the depth of the ABL should be close to reality,and the potential cause of the variation ofhcan be discussed by using statistical analysis.However,it is important to remember that the reanalysis data are based on a combination of real observations and model results,so the uncertainty of the ABL depth given by 20R should not be neglected.Here,we compare the observed ABL depths in Badain Jaran desert(39?28′N,102?22′E)during 3–7 July 2012 w ith thatgiven by 20R(Fig.1).Clear differences can be found.However,itshould be noted thatmoments of observation and reanalysis are not identical.Even so,the 20R gives a shallow CBL on 6 July and deep ones on 4 and 5 July. Moreover,the potential temperature profi les also show sim ilar features w ith observations.A long-term intercomparison of ABL depth between observation and reanalysis is difficult to apply because the observation data are shortand limited. If the surface turbulentheat fluxes,3D atmospheric fields and the depth of the ABL given by 20R are reasonable,then itcan be used to discuss the importance of the residual layer for the CBL development.

The area of interest in this study is the arid and sem i-arid regions nearMongolia,overwhich distinguishable depth differences exist between a nocturnal boundary layer and the CBL,and a deep and neutral residual layer has been monitored using radiosondes(Han etal.,2012).The daily maximum ABL layer(hmax),which primarily represents the final status of the CBL development,is the focus of this study.As such,defining the residual layer is the most crucial step and w illbe given next.

2.2.Identifying the residual layer in reanalysis data

Because a near-neutral layer above the CBL has been suggested to be referred to as the residual layer in previous works,e.g.Stull(1988)and Freire and Dias(2013),the name of the residual layer w illalso be used in this study.Itshould be noted that,based on many observational studies,a(neutrally stratified)residual layer capping the CBL is notalways apparent,which makes discussion of the connection between a residual layer and the CBL difficult.Therefore,the residual layer considered in this study does notneed to be neutral; instead,it is considered as a common layer above the CBL w ith its characteristics varying w ith time.In other words, the residual layer represents the upper externalenvironment thatcan significantly affectthe localCBL developmentin this study.

The classical theory for the diurnalcycle of the ABL suggests that a residual layer should cap the CBL at the early stage of the CBL’s development(Stull,1988).However,because thermodynam ic processes due to atmospheric advection or radiation transm ission can alter the stratification status in the mid-layerof the atmosphere,a distinguishable neutrally stratified layercapping on the CBL isnotalwaysapparent.Thus,many studies considered the layer above the CBL as the free atmosphere layer rather than the residual layer(Fedorovich et al.,2004;Zilitinkevich et al.,2012).Recently, Freire and Dias(2013)suggested thatwhen the lapse rate of the layer above the CBL is close to zero,then it should be called the residual layer;otherw ise,it should be called the free atmosphere layer.Furthermore,they even suggested a two-residual layer structure exists above the CBL.This approach is valuable for site observations w ith dense vertical resolutions,but it is certainly notappropriate for reanalysis such as20R,which hasa vertical resolution ofapproximately 50 hPa(approximately 500 m).

Severalstudies based on observations reportthatthe CBL top tends to jump when a deep and neutral stratified residual layer caps the CBL(Han etal.,2012;Freire and Dias,2013). Although the ABL development is differentbetween observation and 20R,both can reproduce the jump of the CBL top at least over Badain Jaran desert(39?28′N,102?22′E),except for on 6 July 2012(Fig.1).Moreover,the maximum grow th of the observed CBL depth is approximately equal to the depth of the near-neutral layer.The suppression of CBL development on 6 July is due to the precipitation over the observation site before noon.It should be noted that during summer,the cloud cover w ithin the boundary layer is nearly 10%over the region concerned(figure omitted);therefore, the CBL there is mainly developed on a clear day.Forother seasons,the cloud cover north of 50?N can be greater than 50%.Because the potential coupling between the residual layer and CBL is most significant in summer(sections 3.2 and 3.3),the cloud effectw illnotbe considered in this study.

Based on the discussions above,the residual layer is defined in 20R as follows(Fig.2):

(1)The layercapping the CBL is called the residual layer, regardless of the stratification status of the atmospheric column.Therefore,a stable stratified residual layer is allowed in this study.However,itshould be noted thata residual layer mustbe eroded by a mixed layeron the day.

(2)The depth of the residual layer is equal to the fastest grow th of the CBL for each 3-h period during the day. Its lower and upper boundaries(see Fig.2)are defined as the height of the CBL before and after the CBL maximum grow th,respectively.

(3)The features of a residual layer can be calculated by interpolating the pressure levelvariable to the location of the residual layer boundaries.The moments for the reanalysis data used to describe the residual layer should be in frontof that when the CBL erodes into the residual layer(e.g.the residual layeron 5 July as in Fig.2).

Fig.1.Verticalprofi les ofpotential temperature from 3 to 7 July(a–e)over Badain Jaran desert(39?28′N,102?22′E) obtained using radiosondes(IMET-AB).The observations at three times,1000 LST(solid line),1300 LST(dashed line)and 1600 LST(dotted line)are given.The top of the CBL is indicated as fi lled black dots.The potential temperature profi les given by 20R(for the times of0800 and 1400 LST)using linear interpolation are given as thick gray lines.The ABL heights given by 20R are given as short-dash–long-dash horizontal lines(for the times of0800, 1100 and 1400 LST).

Fig.2.An illustration to show how a local residual layer is defined using 20R.Black dots are 3-hourly boundary layer heights;contours show the variation of the potential temperature profi le(intervals:2 K).The results for the location (39?28′N,102?22′E)from 4 to 5 Jul are shown.The residual layerdefined in this study is indicated by gray boxes.Note that the time interval for potential temperature is 6 hours,while that for ABL height is 3 hours.

Table 1.An intercomparison of the lapse rate of the residual layer (γR,units:10-3K m-1)between observation and reanalysis in 2012.Here,the reanalysis data have been linearly interpolated to the location of(39?28′N,102?22′E).The observed lapse rate is derived from the profi le at 1000 LST,while that from reanalysis is calculated using the method given in section 2.2(Fig.1).

W ith these considerations,the residual layer is not only determ ined from the profi le of the atmosphere but also from the feature of CBL development.An intercomparison of residual layer structure over the Badain Jaran desert (39?28′N,102?22′E)between observation and reanalysis is given in Table 1.A lthough there are some differences, 20R generally captures the stratification variation w ithin the residual layer during the observation period.The applicability of this definition over other regions can also be discussed if there are sufficient reliable observations.Follow ing the residual layer definition above,the daily features of the residual layer can be obtained from 20R.The daily maximum ABL depth,which ishighly correlated w ith the daytime mean depth of CBL,is used to representthe daily depth of the CBL. With these considerations,the connections between the residual layer and the CBL are of particular interest in this study.

Using reanalysis data w ith coarse vertical resolution to discuss the detailed structure of sub-layers w ithin ornear the ABL should always be caution.Here,the depth of the residual layer is usually greater than 1000 m over the region concerned(Fig.3),which means there are at least two pressure levels involved in a residual layer for a general situation.If the stratification of the residual layerchanges little in the verticaldirection,the vertical resolution of20R should be appropriate to describe the residual layer structure.By using this definition,the depth of the residual layerw illbe quite closely correlated w ith the CBL depth on the same day.However,the features of the defined residual layer,such as the lapse rate of the potential temperature(γR)and horizontal w ind shear in the vertical direction(Wsh),are independent of the CBL structure and can represent the main upper environmentof a developing CBL.Therefore,the residual layer we define here is the atmospheric layer thatis most likely to be coupled w ith the CBL.This is different from the residual layer capping on a nocturnalboundary layer,although they are suggested to be the same in Stull’s ABL diurnal cycle(Stull,1988).

Other reanalysis data,such as ERA-40(Uppala et al., 2005),also provide the ABL depth(by using different parameterizations),butata larger time intervalof 6 hours.By using these 6-hourly data,the calculated residual layerw illbe too deep and stable.We also calculate the residual layer by using the 3-hourly ERA-interim data(Dee et al.,2011),and find that there is no fundamentaldifference in its given residual layer compared w ith 20R for the period ofobservation in 2012 overBadain Jaran(figure omitted).However,the differences of ABL height in different reanalyses,and their effects on the description ofa residual layershould be noted and w ill be investigated in future work.

3.Results

3.1.Climatologicalmean

Based on the definition in section 2,the climatological mean depth(hR)and the lapse rate w ithin the residual layer (γR)in East Asia are provided in Fig.3.The maximumhRis located near the border line between China and Mongolia, overwhich the maximumhRis even greater than half of the mean maximum CBL depth(notshown).γRexhibits the opposite seasonal variation tohR.Because the residual layer is more neutral,less energy is consumed when the thermals enter the residual layer(Han etal.,2012),a deeper CBL may be stimulated by a neutral residual layer in summer(Fig.3c). There are also differences in the climatologicalmean pattern betweenhRandγR.The m inimumγR(～1.7 K km-1)appears in centralwestMongolia in summer,while the deepest residual layerhas a more stable stratification at～3 K km-1.Such a difference is attributed to the effectof the surface sensible heat flux(Hs).

The climatologicalmean surface sensible heat flux is provided in Fig.4.The pattern ofHsthroughout the year is similar to thatof the residual layer depth(and also the CBL depth),which indicates its importance to the climatological mean distribution of the CBL.However,the surface sensible heat flux in summer is concentrated mainly westof 90?E, which is slightly different from the pattern of the residual layer(also for the CBL)depth(Fig.3c).The neutral residual layereastof90?E may have enlarged the CBL depth there.

The w ind shear w ithin the residual layer(Wsh)may also affect the mean pattern of the CBL depth.However,because the climatologicalmeanWshshows an opposite seasonalvariation to that of the CBL depth,its effectmay be less important.Sim ilar results were derived from a numerical simulation by Moeng and Sullivan(1994).

Fig.3.The climatologicalmean depth of the residual layer(hR,color shading,units:103m)and its lapse rate (γR,contours,units:K km-1)in the four seasons from 1970 to 2010.The contour line is dashed for the range of(0,4)atan intervalof 0.5 and solid for the range of(4,+∞)atan intervalof 1.The line w ith a value of 4 is thickened.The regionalaveraged hRandγRover the entire area is provided at the top of each figure.The location and value for the local maximum hR(purple H)and the m inimumγR(blue N)are also indicated in each figure,along w ith their values in parentheses.

After comparing the climatologicalmeans,the potential effects ofHs,WshandγRon the climatological mean CBL depth exhibitsignificant regional dependence.For example, in the summer,the maximumhRnear(41?N,82?E)corresponds wellw ith the m inimumγR;however,theHsis weaker than its surroundings in that area.At(42?N,100?E),the much strongerHsandWshare likely more responsible for the deep CBL rather thanγR.Therefore,although theHsm ight have dom inated the main seasonal variations and patterns of CBL depth,the effects due to residual layer cannot be neglected.

3.2.Co-variations on an interannual scale

Compared w ith the climatologicalmean patterns,the covariation ofvariablespresentsbetterevidence for theirpotential connections.In this partof the discussion,the seasonal mean variables w ill fi rstbe calculated;then their regionalaverages in summer can be obtained.To obtain the seasonal mean depth of the CBL,we fi rst calculate thehmaxon each day,and then the dailyhmaxcan be used to derive the seasonal mean CBL depth in each year.The seasonalmeanγRandWshare also obtained follow ing this method.

The linearly regressed anomaly forhmaxin differentseasons at each grid based on the local variations ofγRis presented in Fig.5.It is evident that the simultaneous variations ofγRandhmaxare the most significant in summer,followed by the spring(MAM),and weakestin the autumn(SON)and w inter(DJF).In the summer,γRappears to control a large portion of thehmaxanomaly over the regions between 42?N and 48?N.

Compared w ithγR,the region w ith significant positive correlation coefficients betweenhmaxandWshis the largest in spring,from the centerof Mongolia to the south of Russia (Fig.6).The maximumhmaxanomaly regressed byWshis approximately 127 m.In summer,meanwhile,the regressedhmaxaccording toWshshows a much more complicated structure:Over Mongolia,north of 45?N,the regressedhmaxis positive,w ith a maximum value of 181 m;whereas south of 45?N,the regressedhmaxis negative and the minimum value is-184 m.A lthough a stronger w ind shear seems to be beneficial for the development of the CBL from Eq.(1),largeeddy simulation results given by Han et al.(2012)suggest that itcan also suppress the CBL developmentby sustaining the inversion layer,which may induce the negative correla-tion betweenWshandhmaxto the south of 45?N.

Fig.4.As in Fig.3 except for the surface sensible heat flux(Hs,color shading,units:W m-2)and the w ind shear w ithin the residual layer(Wsh,contour intervals:10-3s-1).The localmaximum Hs(blue H)and maximum Wsh(dark green W)are also indicated in each figure,along w ith their values in parentheses.

IfγRandWshare independentofeach other,then their regressedhmaxanomalies in the summer can be greater than 550 m to the north of 45?N(figure not shown),which is greater than that regressed by the surface sensible heat flux there(Fig.7).Therefore,the potential contribution of the residual layer to the development of the CBL is comparable to that from surface sensible heat flux over certain locations. To give a clear representation of the relation between the CBL development and other factors that may have influenced it, the regionally averaged summer means ofhmax,Hs,γRandWshover(40?–50?N,90?–110?E)from 1970 to 2010 are provided in Fig.8.All series are normalized to facilitate the comparison.Surprisingly,the correlation coefficientbetweenhmaxandHs(0.75)is less significant than thatbetweenhmaxandγR(-0.92),although both of them are statistically significantat the 95%confidence level.Itis difficult to conclude if the lapse rate of the residual layer is more important for CBL development than thatofHsbased on their correlation coefficients alone;however,γRseems to be more crucial forhmaxthanHsat times.Forexample,in 1972 and 1988,althoughHsis stronger in the summer,hmaxis notdeeper,and the stratification of the residual layer is stable;while in 1982 and 1985, even thoughHsis notstrong,a more neutrally stratified residual layer(presented as a negative normalizedγR)may have caused a higherhmax.The correlation betweenhmaxandWshis poor,which may be caused by the spatial differences in their local correlations(Fig.5c).

During mostperiods,a stronger(weaker)Hsand a more neutral(stable)residual layer tends to appear simultaneously in summer.This mightbe attributed to the fact thata deeper CBL caused by a strongerHstends to induce a deeper residual layer.To highlight the independent contribution of the residual layer to the CBL,regardless ofHs,the ratio of the regionally averagedhmaxandHs(hmax/Hs)and its correlation coefficientw ithγRare given in Fig.8.γRcan stillexplain approximately two-thirds of the residualvariation ofhmaxwhen the contribution fromHsis removed.Therefore,the contribution from a neutrally stratified residual layer may be more important than that from an intense buoyancy fl ux of a welldeveloped CBL over the region(40?–50?N,90?–110?E)in summer.

3.3.Daily variations

Fig.5.The linearly regressed anomaly for the annual hmaxin differentseasons according to the localvariation ofγR(contour intervals:60 m).The 95%confidence levels are stippled.The 0 lines are thickened.

Because the residual layer is available for every day,it is necessary to check whether the good correspondence between the residual layer structure and CBL developmentcan be verified on the daily scale.A field experimentwas conducted in the Badain Jaran desert in late August2009(Han etal.,2012)and early July 2012(Fig.1).Therefore,we fi rst select two periods to investigate the daily variations of the residual layer and CBL.The fi rst period is from 1 August to 31 September 2009,and the second is from 1 June to 31 July 2012.Follow ing previousdiscussions,the regionally averageddaily maximumγR,andover (40?–50?N,90?–110?E)on each day are presented in Fig. 9;the associated correlation coefficients are listed in Table 2.Because of the significant auto-correlation,the effective number of degrees of freedom[see von Storch and Zw iers (2001)for details]for all of the series is smaller than their sample numbers,which reduces the statisticalsignifi cance of the correlation coefficients.

Surprisingly,although positive correlation coefficients existbetween the originaltheir de-trended series have negative correlations in Augustand Septemberof 2009.The integration of positiveduring a dayexhibits an even more negative correlation w ithThe correlation coefficientbetweenduring June and July in 2012 is positive but insignificant.Therefore,an intense surface sensible heat flux seems to affectthe variation of the CBL depth over a long period butnotw ithin the daily scale.

The correlation coefficients betweenare more statistically significant during both periods.This confi rms the importance of the residual layer on CBL development. Furthermore,the one day lead/lag correlation coefficients betweenγRandhmaxare also significant,at least for the 95% confidence level.This may indicate thata residual layer can retain characteristics of the CBL for two continuous days in this region,which partly confi rms the typicaldiurnalcycle of the ABL given by Stull(1988).

Compared w ith observations,the development of the regionalmean CBL from reanalysis data on 30 and 31 August (Fig.9)is sim ilar w ith the observation in Badain Jaran(Han etal.,2012).From the observation experiment,the maximum surface sensible heat flux is about150 W m-2on 30 August, and the maximum CBL depth is over 3000 m;while on 31 August,although the maximum surface sensible heat flux is approximately 250 W m-2,the maximum CBL depth is just 1800 m.Significant differences occur in the observed lapse rates of the residual layeron these two days.Observed phenomena on these two days is quite similar to the regionalaveraged results given by 20R.However,the regionalaveraged maximum CBL depth shows a continuous increase from 3 to 7 July 2012 from 20R,which is not in common w ith the site observation(Fig.1).Therefore,even though the representativeness of 20R needs to be further investigated,the relationship among its CBL depth,the lapse rate of the residual layerand the surface sensible heat flux is quite sim ilaras that derived from site observations.

Fig.6.Sim ilar to Fig.5 except for the anomaly of hmaxaccording to the localvariation of the verticalshear of horizontalw ind(contour intervals:40 m).

Table 2.The correlation coefficients of the daily regionally averaged variables w ith hmaxfrom 1 August to 30 September2009 and from 1 June to 31 July 2012.The results for the original series(O)and de-trended(D)series are listed.The+1(-1)indicates that the variable listed in the table is leading(lagging)hmaxby 1 day.The correlation coefficients thatare statistically significantat the 95%confidence level are in bold font,and those at the 99%confidence levelare in bold-italic font.

The correlation coefficient between the daily regionally averagedhmaxandγRfor each month during 1970–2012 is displayed in Fig.10a;the results from the de-trended series are displayed in Fig.10b.The correlation coefficient betweenhmaxandγRis negative and significant most of the time,especially for the de-trended series.The correlation is the strongest in Apriland Septemberand weakest in January and July.Considering the seasonalvariation of the CBL(Fig. 3),the potential effect ofγRis strongest when the monthly mean CBL is grow ing(i.e.April)or decaying(i.e.September).When the monthly mean CBL is at its maximum or minimum,the effectofγRbecomes much weaker.

Fig.7.Similar to Fig.5 except for the linearly regressed anomaly of hmaxaccording to the local variation of sensible heat flux(contour intervals:100 m).

Fig.8.The time series ofnormalized summer(JJA)mean hmax,Hs,γR,Wshand hmax/Hsfrom 1970 to 2012,all of which are averaged over(40?–50?N,90?–110?E)(a).The correlation coefficients of Hs,γR,Wshw ith hmaxare provided in parentheses.The correlation coefficientof hmax/Hsw ithγRis also provided.A star means the coefficient is statistically significantat the 95%confidence level.The normalized Hs,γRand Wshversus hmaxis given(b).The colors of the scatters are the same as those of the lines in(a).

As per the discussions above,itseems that the effects ofHsandγRon the developmentof the CBL are different.From October to March,when the monthly mean CBL depth is the shallowest in a year(data not shown,but the value can be referred to by the depth of the residual layer in Fig.3),its daily depth is mainly controlled byHs.During this period, the residual layer may be too shallow and stably stratified (Fig.3a)to affect the CBL.From March to May,a nearly neutral residual layer seems to have transm itted the features of the CBL developmentbetween days.This is a type of accumulative grow th of the CBL:a deep CBL helps maintain a deep and neutral residual layeruntilanothernew CBL begins to develop the nextmorning.Such an effectof the residual layerbecomes weak in m id-summer,when the monthly mean CBL depth is close to its maximum in a year.A fterwards, when the monthly mean CBL depth begins to decay,accumulative grow th of the CBL becomes significantagain.This is evidenced by the fact that the surface sensible heat flux in the autumn isonly abouttwo-fi fths of thatin the spring(Figs. 4b,d),but the CBL depth in autumn is approximately two to three times that in spring.Therefore,the residual layer may have accelerated the increase of the CBL depth in spring,and slowed the decay in autumn.

4.Summary and discussion

Fig.9.The daily normalized series of regionally averagedγR,and Wshover(40?–50?N,90?–110?E)in(a)2009 and(c)2012.The vertical dot-dot-dash lines indicate the period when radiosondes were used in the Badain Jaran observation experiment in 2009 and 2012.The scatter plots for the daily normalized variables versus hmaxare given in(b)and(d).The color of the scatters is the same as the lines in(a).

This paper analyzes the effectof the residual layer on the CBL development near Mongolia.Based on observations, the depth of the residual layer is subjectively assumed to be equal to the largestgrow th of the CBL depth in 3 hours.The effectof the residual layer on the developmentof the CBL is investigated over a large area during 1970–2012.The main conclusions are as follows:

(1)The climatological mean distribution of the CBL depth in each season may be notonly determined by the surface sensible heat flux,butalso affected by the stratification of the residual layer.

(2)The interannual variations of the seasonalmean CBL depth are significantly correlated w ith the w ind shear and stratification w ithin the residual layerover the region of(40?–50?N,90?–110?E)in summer.The correlation of the regionally averaged CBL depth w ith the lapse rate of the residual layer is even stronger than thatw ith the surface sensible heat flux.

(3)On the daily scale,the correlation between the surface sensible heat flux and CBL depth is no longersignificantand even becomes negative in Apriland September.Meanwhile, the lapse rate w ithin the residual layer is still significantly correlated w ith the CBL depth;thus,the daily variation of the CBL depth may be more influenced by the stratification status w ithin the residual layer than by the buoyancy flux originated from the heating below.

Considering that the definition of the residual layer used in this study assumes a close connection to the grow th of the CBL,a discussion of the relationship between the depths of the residual layer and CBL is meaningless.However,if the residual layer isdeep enough,then the impactof itsstratification from large-scale circulation can be easily observed.Such a process represents the impactof the large-scale circulation on the localCBL development.Forexample,the contribution ofadvection and the synoptic-scale circulation on CBL developmenthas been suggested by Bianco etal.(2011);however, the mechanisms responsible have not yet been determined, which may be attributed to the unknown changes in the layer above the CBL.Based on this study,the large-scale circulation can impact the CBL development through changing the structure of the residual layer.Specifically,the study region experiences frequently synoptic activities(Ren etal.,2010). Whether these large-scale processes are connected w ith the maintenance and variation of the residual layer,and further can affect the localCBL development,requires investigation.

Fig.10.The daily correlation coefficients between the regionally averagedandand between(c,d)in every month from 1970 to 2012.The left column is from the originalseries and the rightcolumn is from the de-trended ones.The correlation coefficients thatare statistically significantat the 95%confidence level(t-test)are dotted.

From this study,the correlation between CBL depth and γRorWshover the study region shows significantspatial differences(see Figs.5 and 6).Such spatial dependency may be caused by the terrain effect.On the continental scale,the main circulation over East Asia is modulated by the thermal and dynam ic effect of the Tibetan Plateau,which means the thermal structure of the atmosphere at m id-level height w ill also be impacted.While on a smaller spatial scale,the coupling between the CBL and the slope–valley w ind system in mountain regions has been noted for decades(Banta,1984, 1986;Stensrud,1993),and the deep residual layer there is quite sim ilar to whatwe observed in Badain Jaran(Fig.1).

If the structure above the CBL is mainly impacted by factors such as the large-scale circulation,then the label of“residual layer”should be reconsidered,as its characteristics may no longer be the resultor residualof the local CBL developmentthe previous day.This problem has also been identified in the Badain Jaran experiment(Han etal.,2012);the layer above the CBL is called the neutral layer in thatstudy. In this study,the lapse rate of the residual layer is still significantly correlated w ith the CBL depth on the previous day (Table 2),which indicates that the characteristics of the CBL developmentcan be stored in the residual layeruntil the follow ing day.Therefore,the capping layer over the CBL can still be considered the residual layer,but its features are not completely determ ined by the CBL of the previous day.

Acknow ledgements.This research was funded by the National Natural Science Foundation of China(Grant No.41205005), the National Basic Research Program of China(Grant No.2010CB950503),and the West Light Foundation of the Chinese Academy of Sciences to HAN Bo.The Twentieth Century Reanalysis Project dataset is provided by the U.S.Department of Energy, Office of Science Innovative and Novel Computational Impact on Theory and Experiment(DOE INCITE)program,and Office of Biological and Environmental Research(BER),and by the National Oceanic and Atmospheric Administration Climate Program Office.

REFERENCES

Avissar,R.,and T.Schmidt,1998:An evaluation of the scale at which ground-surface heat flux patchiness affects the convective boundary layer using large-eddy simulations.J.Atmos. Sci.,55,2666–2689.

Banta,R.M.,1984:Daytime boundary-layerevolution overmountainous terrain.Part 1.Observations of the dry circulations. Mon.Wea.Rev.,112,340–356.

Banta,R.M.,1986:Daytime boundary-layerevolution overmountainous terrain.Part II.Numerical-studies ofupslope flow duration.Mon.Wea.Rev.,114,1112–1130.

Banta,R.M.,and Coauthors,2011:Dependence of daily peak O-3 concentrationsnearHouston,Texas on environmental factors: Wind speed,temperature,and boundary-layer depth.Atmos. Environ.,45,162–173.

Barthlott,C.,J.W.Schipper,N.Kalthoff,B.Adler,C.Kottmeier, A.Blyth,and S.Mobbs,2010:Model representation of boundary-layer convergence triggering deep convection over complex terrain:A case study from COPS.Atmospheric Research,95,172–185.

Bianco,L.,I.V.Djalalova,C.W.King,and J.M.Wilczak,2011: Diurnal evolution and annual variability of boundary-layer heightand its correlation to other meteorologicalvariables in California’s Central Valley.Bound.-Layer Meteor.,140,491–511.

Compo,G.P.,and Coauthors,2011:The twentieth century reanalysis project.Quart.J.Roy.Meteor.Soc.,137,1–28.

Dee,D.P.,and Coauthors,2011:The ERA-Interim reanalysis: configuration and performance of the data assim ilation system.Quart.J.Roy.Meteor.Soc.,137,553–597.

Durre,I.,R.S.Vose,and D.B.Wuertz,2006:Overview of the integrated global radiosonde archive.J.Climate,19,53–68.

Elliott,W.P.,and D.J.Gaffen,1991:On the utility of radiosonde humidity archives for climate studies.Bull.Amer.Meteor. Soc.,72,1507–1520.

Fedorovich,E.,R.Conzem ius,and D.M ironov,2004:Convective entrainment into a shear-free,linearly stratified atmosphere: Bulk models reevaluated through large eddy simulations.J. Atmos.Sci,61,281–295.

Fochesatto,G.J.,P.Drobinski,C.Flamant,D.Guedalia,C.Sarrat,P.H.Flamant,and J.Pelon,2001:Evidence of dynam ical coupling between the residual layer and the developing convective boundary layer.Bound.-Layer Meteor.,99,451–464.

Freire,L.S.,and N.L.Dias,2013:Residual layer effects on the modeling of convective boundary layer grow th rates w ith a slab model using FIFE data.J.Geophys.Res.,118,12 869–12 878.

Garratt,J.R.,1994:The Atmospheric Boundary Layer.Cambridge University Press,316 pp.

Gentine,P.,A.K.Betts,B.R.Lintner,K.L.Findell,C.C.van Heerwaarden,A.Tzella,and F.D’Andrea,2013a:A probabilistic bulk model of coupled mixed layer and convection. Part I:Clear-sky case.J.Atmos.Sci.,70,1543–1556.

Gentine,P.,A.K.Betts,B.R.Lintner,K.L.Findell,C.C.van Heerwaarden,and F.D’Andrea,2013b:A probabilistic bulk model of coupled mixed layer and convection.Part II:Shallow convection case.J.Atmos.Sci.,70,1557–1576.

Han,B.,S.H.Lu,and Y.H.Ao,2012:Development of the convective boundary layer capping w ith a thick neutral layer in Badanjilin:Observations and simulations.Adv.Atmos.Sci., 29,177–192,doi:10.1007/s00376-011-0207-4.

Holtslag,A.A.M.,and B.A.Boville,1993:Local versus nonlocal boundary-layer diffusion in a global climate model.J. Climate,6,1825–1842.

Holtslag,A.A.M.,E.I.F.Debruijn,and H.L.Pan,1990: A high-resolution air-mass transformation model for Shortrange weather forecasting.Mon.Wea.Rev.,118,1561–1575.

Hong,S.Y.,and H.L.Pan,1996:Nonlocalboundary layervertical diffusion in a medium-range forecastmodel.Mon.Wea.Rev., 124,2322–2339.

Huang,Q.,J.H.Marsham,D.J.Parker,W.S.Tian,and C. M.Grams,2010:Simulations of the effects of surface heat flux anomalies on stratification,convective grow th,and vertical transportw ithin the Saharan boundary layer.J.Geophys. Res.-Atmos.,115,doi:10.1029/2009JD012689.

Lenschow,D.H.,and P.L.Stephens,1980:The role of thermals in the convective boundary-layer.Bound.-Layer Meteor.,19, 509–532.

Lin,J.T.,and M.B.M cElroy,2010:Impacts of boundary layer mixing on pollutantverticalprofi les in the lower troposphere: Implications to satellite remote sensing.Atmos.Environ.,44, 1726–1739.

Maronga,B.,and S.Raasch,2013:Large-eddy simulations of surface heterogeneity effects on the convective boundary layer during the LITFASS-2003 experiment.Bound.-Layer Meteor.,146,17–44.

Marsham,J.H.,D.J.Parker,C.M.Grams,B.T.Johnson,W.M.F. Grey,and A.N.Ross,2008:Observations of mesoscale and boundary-layer scale circulations affecting dust transportand upliftover the Sahara.Atmos.Chem.Phys.,8,6979–6993.

Messager,C.,D.J.Parker,O.Reitebuch,A.Agusti-Panareda,C. M.Taylor,and J.Cuesta,2010:Structure and dynamicsof the Saharan atmospheric boundary layer during the West A frican monsoon onset:Observations and analyses from the research fl ights of 14 and 17 July 2006.Quart.J.Roy.Meteor.Soc., 136,107–124.

Moeng,C.H.,and P.P.Sullivan,1994:A comparison of sheardriven and buoyancy-driven planetary boundary-layer flows. J.Atmos.Sci.,51,999–1022.

Pan,H.L.,and L.Mahrt,1987:Interaction between soilhydrology and boundary-layer development.Bound.-Layer Meteor.,38, 185–202.

Parker,D.J.,C.D.Thorncroft,R.R.Burton,and A.Diongue-Niang,2005:Analysis of the A frican easterly jet,using aircraft observations from the JET2000 experiment.Quart.J. Roy.Meteor.Soc.,131,1461–1482.

Ren,X.J.,X.Q.Yang,and C.J.Chu,2010:Seasonalvariations of the synoptic-scale transient eddy activity and polar front jet over EastAsia.J.Climate,23,3222–3233.

Roy,S.B.,and R.Avissar,2000:Scales of response of the convective boundary layer to land-surface heterogeneity.Geophys. Res.Lett.,27,533–536.

Saha,S.,and Coauthors,2010:The NCEP climate forecastsystem reanalysis.Bull.Amer.Meteor.Soc.,91,1015–1057.

Stensrud,D.J.,1993:Elevated residual layers and their influence on surface boundary-layerevolution.J.Atmos.Sci.,50,2284–2293.

Stull,R.B.,1976:M ixed-layer depth model based on turbulent energetics.J.Atmos.Sci.,33,1268–1278.

Stull,R.B.,1988:An Introduction to Boundary Layer Meteorology.Kluwer Academic Publishers,670 pp.

Tennekes,H.,1973:Model fordynam ics of inversion above a convective boundary-layer.J.Atmos.Sci.,30,558–567.

Thornton,D.C.,A.R.Bandy,and J.G.Hudson,2011:Fast sulfurdioxide measurements correlated w ith cloud condensation nuclei spectra in the marine boundary layer.Atmos.Chem. Phys.,11,11 511–11 519.

Troen,I.,and L.Mahrt,1986:A simple-modelof the atmospheric boundary-layer-sensitivity to surface evaporation.Bound.-Layer Meteor.,37,129–148.

Uppala,S.M.,and Coauthors,2005:The ERA-40 re-analysis. Quart.J.Roy.Meteor.Soc.,131,2961–3012.

von Storch,H.,and F.W.Zw iers,2001:Statistical Analysis in Climate Research.Cambridge University Press,494 pp.

Zilitinkevich,S.,2012:The height of the atmospheric planetary boundary layer:State of the art and new development.National Security and Human Health Implications of Climate Change,H.J.S.Fernando,Z.K lai′c,and J.L.M cCulley,Eds., Springer Netherlands,147–161.

Zilitinkevich,S.S.,S.A.Tyuryakov,Y.I.Troitskaya,and E.A. Mareev,2012:Theoreticalmodels of the heightof the atmospheric boundary layerand turbulententrainmentat its upper boundary.Izvestiya,Atmospheric and Oceanic Physics,48, 133–142.

:Han,B.,C.L.Zhao,S.H.L¨u,and X.Wang,2015:A diagnostic analysis on the effectof the residual layer in convective boundary layerdevelopmentnear Mongolia using 20th century reanalysis data.Adv.Atmos.Sci.,32(6),807–820,

10.1007/s00376-014-4164-6.

(Received 23 July 2014;revised 29 September2014;accepted 29 October 2014)

?Corresponding author:HAN Bo

Email:hanbo@lzb.ac.cn

?Institute of Atm ospheric Physics/Chinese Academ y of Sciences,and Science Press and Springer-Verlag Berlin Heidelberg 2015