2006 a Reanalysis of the Atmospheric Boundary

DOI 10.1393/ncc/i2006-10019-9

IL NUOVO CIMENTO Vol. 29 C, N. 5 Settembre-Ottobre 2006

A reanalysis of the atmospheric boundary layer field experiment(SPCFLUX93) at San Pietro Capofiume (Italy)()

C. Cassardo(1), S. Ferrarese(1), A. Longhetto(1), M. G. Morselli(2)and G. Brusasca(2)

(1) Dipartimento di Fisica Generale, Universita` di Torino - Via P. Giuria 110125 Turin, Italy

(2) ARIANET S.r.l. - Via Gilino 9, 20128 Milan, Italy

(ricevuto il 22 Maggio 2006; approvato il 10 Luglio 2006)

Summary. A fortnight eld experiment was carried out at San Pietro Capoume(Po Valley, Italy) during the month of June, 1993, and was named SPCFLUX93.This location was chosen as representative of the Po Valley. The SPCFLUX93experiment was devised according to the results of some previous measurementscarried out in mountainous areas of South Europe (i.e. ALPEXALPine EXper-iment, PYREXPYRenean Experiment), and aimed to represent a prototype forfurther eld observations. The dataset of the SPCFLUX93 experiment consistedof: i) meteorological and chemical data collected continuously with slow-responsesensors in the atmospheric surface layer and into the soil; ii) data coming from fast-response instrumentation (sonic anemometers and uxmeter); iii) radiosoundingscarried out with free and tethered balloons; iv) continuous vertical wind soundingswith a Mini-Sodar. The aim of the SPCFLUX93 eld experiment was to investigatethe following topics: atmospheric turbulence, dry and wet atmospheric total deposi-tion, energy balance, thermal wave propagation in the soil. Few years later, the at-mospheric and hydrological scientic community conduced an extensive programme,the Mesoscale Alpine Programme (MAP), on weather and climate in mountainousregions. This programme considered many aspects of alpine meteorology, rangingfrom high-resolution numerical modelling to experimental campaigns performed onboth sides of the Alps, with the aim to better understand the interaction processes ofatmospheric uxes with the orography. Many puzzling problems were posed by thecomplexity of these interactions; among them, the perturbations on the boundarylayer structure caused by the airows that cross the Alps and reach the Po Valleywould still require more experimental observations and theoretical studies. Theseconsiderations prompted us to reanalyze the SPCFLUX93 dataset.

() The authors of this paper have agreed to not receive the proofs for correction.

c Societa` Italiana di Fisica 565

566 C. CASSARDO, S. FERRARESE, A. LONGHETTO, ETC.

In this paper, a layout of the eld experiment (including the instrumentation details,the experimental relevant dataset and database composed by meteorological stan-dard data, vertical prole data, ultrasonic anemometer data, and chemical data) ispresented; the collected data are described; the details of the mesoscale meteoro-logical situation over San Pietro Capoume during the experiment are presented;nally, some analyses on the data are shown, and the main results coming from theseveral applications carried out using the dataset are illustrated or summarized. Inparticular, the most interesting results are related to the following topics: the char-acteristics of the turbulence in the surface layer (using the fast-response data), thevalidation of land surface schemes (using the surface observations), the evaluationof mixed layer depth (using radon ux data) and the estimate of deposition velocity.

PACS 92.60.-e Properties and dynamics of the atmosphere; meteorology.

1. Introduction

A renewed interest in the meteorology around and downwind of large mountain rangeslike the Alps prompted the atmospheric and hydrological scientic community to promotethe comprehensive programme MAP (Mesoscale Alpine Programme [1, 2]). MAP wasdedicated to the eld observation and model simulation on the Alpine Meteorology, andintended to lighten many aspects of the interaction between synoptic atmospheric owand the Alps. In particular, the relevance of mesoscale circulation patterns induced in thePo Valley by the Alpine chain on the occasion of specic directions of the atmosphericow was recognised and studied. For this reason, a number of xed target areas forground-based observing systems were envisaged on the southern side of the Alps (the PoValley).

The Po Valley is a at region located in northern Italy around the 45th parallel andstretched along the Po River, bounded on the North and the West by the Alps (a compacttopographic barrier extended along parallels on its eastern range and C shaped on thewestern one, and with its greatest vertical extension in the north-western sector) and onthe South by the Apennines (a smoother and lower barrier disposed along an axis NW-SEfrom northern to southern Italy). Due to this particular conguration, the climate of thePo Valley can be assumed to belong to the sub-littoral continental type [3], characterisedby a typical low wind regime, with frequent occurrence of nocturnal thermal inversionsand diurnal convective conditions. In fact, the Po valley peculiar geographic character-istics allows to assimilate this area to a large valley bounded by two mountainous ridges(Alps and Apennines) sheltering the lowland from northerly and westerly synoptic circu-lation and favouring mesoscale, thermally induced breeze circulation. Thus, observationsof vertical proles of wind, humidity and temperature in the lower troposphere and ofmomentum, sensible and latent heat uxes in the surface layer could be used to check orimprove the turbulence parameterisations in the atmospheric boundary layer of the leeside ow.

In conformity with previously performed experiments over mountainous areas of SouthEurope (ALPEXALPine Experiment: [4]; PYREXPYRenean Experiment: [5]), aeld observation campaign on the atmospheric boundary layer was organised in 1993in the middle eastern region of the Po Valley. A joint Italian team participated in thepreliminary and in the active phases of the campaign. The team was composed of: the

A REANALYSIS OF THE ATMOSPHERIC BOUNDARY LAYER FIELD EXPERIMENT ETC. 567

Department of General Physics of University (DFG) of Turin, the Institutes of Scienceof Atmosphere and Climate (ISAC) of Turin and Bologna, the former Centre of theEnvironmental and Material Research of the Electric Power Board (ENEL) of Milan, theCentre of Information, Studies and Experiences (CISE) of Segrate (Milan) and the JointResearch Centre (JRC) of Ispra (Varese).

This campaign aimed to investigate the following objectives: atmospheric turbulence,energy balance, thermal wave propagation in the soil, dry and wet atmospheric totaldeposition, and in the selection of the most suitable site for the experiment. The chosenlocation was San Pietro Capoume (hereafter referenced as SPC and described in detailin sect. 2), and the campaign was named SPCFLUX93.

The period of the year chosen for the active phase was the month of June. As theclimatology of the region suggested that usually the spring rainy period ends in therst weeks of June, in order to have a good chance to measure well-established diurnalconvective and nocturnal stable episodes, it was decided to start the measurements inthe second part of the month of June.

In this paper, a description of the site can be found in sect. 2, while an account of thedeployed instrumentation is given in sect. 3. The meteorological conditions during theactive phase of the campaign are discussed in sect. 4. A summary of the data collectedduring SPCFLUX93 is shown in sect. 5. Some analyses of the data are resumed in sect. 6.

2. The site

San Pietro Capoume is located at about 30 km east of Bologna (northern Italy),in the southern central part of the Po Valley, at a height of about 10m above the sealevel, at a distance of about 100 km from the Apennines chain, and about 60 km from theAdriatic sea (g. 1). Near the station, the soil type is loam, and the dominant vegetationtype is grass, regularly cut by the farmers. This site was selected for several reasons.Firstly, the vegetation cover can be considered uniform at the mesoscale range, althoughthere were dierent kinds of canopy in the neighbourhood of the station. Secondly, SPCwas selected as a suitable site for a WMO (Word Meteorological Organisation) station(code 16144) because it is located in a at region (horizontally homogeneous at localscale), thus its in situ data could be also considered as representative of the surroundingwider region stretched from the Apennines chain to the Adriatic sea. Finally, ENEL andCISE already performed conventional atmospheric measurements at SPC as far back asthe year 1992. It was then decided to use the existing station, and to equip it with newsensors.

3. The instrumentation and the dataset

The instruments used in SPCFLUX93 experiment were installed at 18 dierent loca-tions around the SPC meteorological station. All general information regarding the 18locations are summarised in table I. The locations could be split into 5 groups, accordingto the measurements carried out:

a) Standard meteorological observations:locations T1A, T1B, T2, T3 and T4 were equipped with slow-response (response time 1 s) instruments (table II).b) Advanced meteorological observations:


Fig. 1. Detailed map of the San Pietro Capoume area; the box in the right-bottom sectorindicates the location of the area.

locations S1, S2, S3 and S4 were equipped with fast-response ( < 1 s) instruments(table III).

c) Vertical prole observations:4 locations were equipped to measure vertical proles (table IV): P1 (Mini-Sodar, work-ing almost continuously during the experiment, gives vertical proles of horizontal windvelocity and direction with a vertical resolution of 10m); P2 (Airsonde balloons, givevertical proles in the layer 03000m, with a resolution of about 10m); P3 (Tethersondeballoons, give vertical proles in the layer 0500m, with a resolution of about 510m);P4 (WMO radiosounding station, giving four vertical proles a day in the troposphere,at 00, 06, 12 and 18 UTC with mandatory and signicant levels).

d) Deposition measurements:3 locations providing integrated concentration data of some atmospheric trace gases sub-divided according to the deposition type were included in this group: I1 (dry deposition),I2 (wet deposition) and I3 (fog deposition); they are reported in table V.

e) Radon concentration measurements:2 stations were included in this group: R1 (atmospheric concentration) and R2 (soilconcentration); they are reported in table VI.

The SPCFLUX93 eld experiment dataset was made up of the instantaneous datarecorded by the instruments installed in location S1 and S2 (fast-response sensors), P1,


Table I. General information about the locations and the database.

Group Location Owner Acquisition Averaging Site Availability of(*) time time instantaneous

data

a

T1A ENEL - SRI 2 s 30min shelter noENVIRONMENTAL AREA

ICG/CNR DFG

T1B ENEL - SRI 2 s 30min mast noENVIRONMENTAL AREA

T2 FISBAT/CNR 1min 30min tower no

T3 SMR/ER (**) 1min 1 h shelter no

T4 IMS (***) 3 h 3 h shelter no

b

S1 ENEL S.P.A. - SRI - 1021Hz 30min mast yesENVIRONMENTAL AREA,

DFG

S2 DFG 21Hz 30min tower yes

S3 JRC 1021Hz 15min tower no

S4 JRC 1021Hz 30min mast no

c

P1 ENEL S.P.A. - SRI - 10min 30min profile NoENVIRONMENTAL AREA,

CISE

P2 ENEL - SRI 23 h - profile YesENVIRONMENTAL AREA

P3 ENEL - SRI 23 h - profile YesENVIRONMENTAL AREA

P4 IMS (***) 12 h - profile Yes

d

I1 ENEL S.P.A. - SRI - 34 days - shelter YesENVIRONMENTAL AREA,

CISE

I2 FISBAT/CNR 1 week - ground Yes

I3 ENEL S.P.A. - SRI - event - ground YesENVIRONMENTAL AREA,

FISBAT/CNR

eR1 CISE 1min 30min shelter No

R2 CISE 1min 30min ground No

(*) As specified in sect. 3.(**) Regional Meteorological Service of Emilia-Romagna Region.(***) Italian Meteorological Service.

P2, P3 and P4 (vertical prolers), and I1, I2 and I3 (chemical measurements), and ofthe averaged data recorded by the other instruments, whose software directly calculatedthe output. For sake of uniformity, in the reanalysis the data coming from the locationsS1 and S2 were averaged and the database was created according to the specications oftable I.


Table II. List of standard meteorological observations (group a, see table I and sect. 3).

Location Physical quantity measured Height of measurement Unit

T1A Air pressure 2m hPa

T1A Precipitation 2m mm

T1A Concentrations of SO2,NO,NO2,O3 2m ppb

T1B Air temperature 2,10m C

T1B Soil temperature 0,5,10,20,30,50,100 cm CT1B Relative humidity 2m %

T1B Soil moisture 10 cm %T1B Leaf wetness 0.5m 02 (Dry/Wet)

T1B Solar global radiation 2m mW/cm2

T1B Net radiation 2m mW/cm2

T1B Soil heat flux 3,8,15 cm mW/m2

T1B Horizontal wind velocity and direction 10m m/s, deg

T2 Air temperature 1,5,10,30,50m C

T2 Relative humidity 1,50m %

T2 Horizontal wind velocity and direction 1,5,10,30,50m m/s, deg

T3 Air temperature 0.5,1.5m C

T3 Relative humidity 1.5m %

T3 Solar global radiation 1.5m W/m2

T3 Precipitation 1.5m cumulated mm

T3 Horizontal wind velocity and direction 10m m/s, deg

T4 Air and dew point temp. 2m C

T4 Atmospheric pressure 2m hPa

T4 Precipitation 2m cumulated mm

T4 Horizontal wind velocity and direction 10m m/s, deg

T4 Cloudiness 2m eighths

4. Mesoscale meteorological situation over SPC during the experiment

The main meteorological characteristics, analyzed using the European MeteorologicalBulletin [6] maps, are here summarized.

On the 15th of June, a weak depression centred on the Po Valley at 00 UTC (i.e. 02a.m. local time) was lling, with winds at 850, 700 and 500 hPa coming from NW. Onthe 16th, the pressure and geopotential gradients at all levels were weak. On the 17th,a frontal cold system approached the eastern Alps, driven by NW winds at 500 hPa. Inthe evening of the 17th (between 8.00 p.m. and 8.30 p.m., as we can see in g. 6), thiscold front crossed SPC station during its motion toward SE. Following it, a wedge ofrelatively high pressure established on the Po Valley, with post-frontal surface currentstoward west and at 700 hPa toward east. In the following period, from the 19th to the22nd, the pressure eld was high and at. All frontal systems kept bounded North of


Table III. List of advanced meteorological observations (group b, see table I and sect. 3).

Location Physical quantity measured Height of Unitmeasurement

S1 Sonic 3D wind and sound velocity 10m m/s

S1 Sonic vertical wind velocity 10m m/s

S1 Temperature uctuations 10m C

S1 Moisture uctuations 10m g/m3



S3 O3 uctuations 10m ppb

S3 NO2 uctuations 10m ppb


S4 Moisture uctuations 10m g/m3

S4 O3 uctuations 10m ppb

S4 Net radiation 2m W/m2

S4 Soil heat ux 3 cm W/m2

Table IV. List of vertical prole observations (group c, see table I and sect. 3).

Location Physical quantity measured Height of Unit Verticalmeasurement resolution

P1 Horizontal wind velocity and direction prole 0350m m/s, deg 10m

P2 Dry and wet bulb temperature prole 03000m C 10mP2 Atmospheric pressure prole 03000m hPa 10mP3 Dry and wet bulb temperature prole 0500m C 510m

P3 Atmospheric pressure prole 0500m hPa 510m

P3 Horizontal wind velocity and direction prole 0500m m/s, deg 510m

P3 O3 concentration prole 0500m ppb 510m

P4 Dry and wet bulb temperature prole 0100 hPa C (*)

P4 Atmospheric pressure prole 0100 hPa hPa (*)

P4 Horizontal wind velocity and direction prole 0100 hPa m/s, deg (*)

(*) Mandatory and significant levels.


Table V. List of deposition measurements (group d, see table I and sect. 3).


I1 Air concentrations of SO2, SO4=, NO2, NO3

, NH3, 1m g/m3

NH4+, HNO2 and HNO3

I2 Precipitation 1m mm

I2 Acidity of precipitation 1m PH

I2 Conducibility of precipitation 1m s/cm

I2 Concentration of NH4+, Na, K, Ca, Mg, Cl, NO3, 1m mg/l

SO4= and PO4=, in precipitation

I2 Alkalinity of precipitation 1m meq/l

I3 Quantity of droplets 1.5m mm

I3 Acidity of droplets 1.5m pH

I3 Conducibility of droplets 1.5m s/cm

I3 Concentrations of NH4+, Na, K, Ca, Mg, Cl, NO3, 1.5m mg/l

SO4=, PO4=, Fe, Mn, Pb, Br, Fl and Sulphite in

droplets

I3 Alkalinity of droplets 1.5m meq/l

I3 Sum of anions and cations in droplets 1.5m -

the Alps, and 500 hPa ow was slowly rotating counter-clockwise, blowing from West(on the 20th), SW (on the 21st) and nally from South (on the 22nd). On the 23rd,a SW circulation driven by the deep minimum over Finland created the conditions foran episode of surface orographic cyclogenesis over the Ligurian Sea, with the onset ofa frontal S-shaped system at the surface (over the Atlantic Ocean) and a strong SWow at 500 hPa. In the morning of the 25th, a wedge of the Azores anticyclone movedrapidly toward the Po Valley lling the pre-existing depression and causing a 500 hPaow from NW.

On the 26th and the 27th, the surface high-pressure eld over the Po Valley remainedsubstantially levelled, while the airow came from NW at 700 and 500 hPa. On the 28th,a frontal cold system approached the northern side of the Alps, originating pre-frontalows from SW at the surface, and from WNW at 700 and 500 hPa. Finally, on the 29th,

Table VI. List of radon concentration measurements (group e, see table I and sect. 3).


R1 Air concentration of radon 2m Bq/m3

R2 Soil radon ux 50 cm mBq/m2s


Fig. 2. Time trend of global radiation (location T1B, solid line) and net radiation (locationT1B, dashed line), in Wm2, during the whole campaign.

the low pressure associated with this system drew SW currents at all levels.From the above analysis it might be summarized that, in the whole observation period,

the meteorological conditions were stable, with only 3 exceptions: the passage of a coldfront during the 18th, the cyclogenetic episode on 23rd-24th and a cyclonic situation onthe 28th.

5. Overall view of data

During the campaign, an enormous amount of data was collected. In this paper,we only describe the most meaningful patterns, showing a signicant subset. For thecomplete record of data, the reader is referred to [7]. While otherwise stated, hours areindicated in local time, i.e. 2 hours later than the UTC or GMT.

5.1. Meteorological standard data. The solar global radiation (at location T1B)showed peak values slightly larger than 900Wm2 (g. 2, solid line), and on the 16th,18th, 19th, 22nd, 26th, 27th and 28th the daily trend was typical of a cloudless day.Net radiation (at the same location) reached as far as 550600Wm2 during daytime(g. 2, dashed line) while the night time values generally did not exceed the threshold of50Wm2, with the minimums recorded immediately after the sunset.

A unique severe rainfall event was recorded (at T1A location) during the campaign,in the morning of the 25th, and an episode of weak rain was observed near the noonof the 23rd. The relative humidity (measured in location T1B) showed a regular dailycycle, with a maximum during night time (the 100% level was reached on every night buton the 21st, 23rd and 24th) and a minimum of about 3040% at 2 p.m. The nocturnalvalues of relative humidity and net radiation supported the possibility of the formationof a thin layer of haze in the rst hours of the morning, typical of the climate of the PoValley.

Air temperature measured at the meteorological shelter height showed a typical daily


Fig. 3. Time trend of air temperature gathered at T2 location at dierent heights(1,5,10,30,50m) and indicated respectively as T1, T5, T10, T30 and T50 on the 26th of June1993, in Celsius degrees.

trend with the highest values (about 30 C) on the 23rd and 28th and the lowest nocturnalvalues on the 16th (about 12.5 C). The maximum daily excursion (about 12 C) wasobserved on the cloudless days. Soil temperature measured 20 cm underground (locationT1B) showed a behaviour similar to the one of air temperature with a time lag of 4-5hours and with the maximum excursion (about 3-4 C) during the cloudless days. Thepeaks (about 25 C) were observed on the 23rd and 28th, while the minimum value (about20.5 C) was recorded on the 16th.

Figure 3 reports, as an example of a typical clean-sky day, the daily trend of airtemperatures measured at 1, 5, 10, 30 and 50m in the station T2 on the day 26th.In the central hours of the day, the atmosphere was vertically almost isothermal, whileduring night time a stable thermal stratication developed and grew, with temperaturedierences of 3 C or more. On the contrary, the relative humidity (location T2) presentedthe largest dierences in the daytime (more than 20% between the measurements at 50mand at 1m).

Air pressure (location T1A) reached its highest value on the 25th (1024 hPa), withtwo other relative maxima on the 16th and 18th, while the lowest value was recorded onthe 23rd (1010 hPa, just before the cyclogenesis occurrence). Two other relative minimawere present on the 17th and 28th, according with the meteorological analysis of sect. 4.

Horizontal wind speed (at T2 location) was always lower than 6ms1, excepting onthe late evening of the 17th and on the 23rd, on the occasion of the fronts passage.Especially in the sunny days, a marked daily cycle was present, with nearly zero valuesduring night time (on the 18th and 19th) and peaks of about 4-6ms1 around noon.Concerning the wind proles gathered in station T2, g. 4 reports the daily trend ofhorizontal wind speeds measured at 1, 5, 10, 30 and 50m, gathered during the day 26th.At all levels, the wind speed proles showed a similar behaviour. In particular, duringthis day, due to the high-pressure circulation, wind velocity remained unusually low until6.00 p.m., while in the evening the sea breeze circulation was restored.


Fig. 4. Time trend of horizontal wind speed gathered at T2 location at dierent heights(1,5,10,30,50m) and indicated, respectively, as V1, V5, V10, V30 and V50 on the 26th of June1993, in ms1.

5.1.1. The wind in the surface layer. Comparing the horizontal wind speeds v1 and v10measured at 1m and 10m, respectively (location T2), the following regression line wasfound:

(1a) v1 = Av10,

with A = 1.410.01. This equation is consistent with the assumption of the logarithmicwind prole for a neutral atmosphere over vegetation:

(1b) v(z) =ukln(z dz0

),

where u is the friction velocity, k the von Karman constant, d the zero-plane displace-ment length, z the vertical height and z0 the roughness length. In fact, when the typicalexpressions for z0 and d (z0 = 0.10h, d = 0.67h, where h is the vegetation height) areconsidered, in the case of short grass (h = 1050 cm) the ratio v10m/v1m varies in therange 1.331.44.

5.1.2. The calibration of soil misture sensor. The sensor type MC1, manufactured byLastem and installed during SPCFLUX93 at location T1B, is a conducimetric probe forthe measure of the soil water content. The dimensions of the probe, roughly assimilable toa parallelepiped, are approximately 23 12 3mm3. This sensor evaluates soil moisturemeasuring the value of the electrical resistance in alternate current between two electrodesseparated by a hygroscopic dielectric. The electrodes are shielded by a cover made ofstainless steel with holes in order to assure the equilibrium with the outside water.

The correlation between conductivity and water content depends on the compositionand the degree of soil compaction, and must be obtained experimentally, as well as thedependence on the temperature of the sensor. The factory indicates in the operational


manual some general functions which can be used for generic sandy and loamy soils. Inorder to get the true correlation function for the SPC soil, an experimental set-up wasprepared.

Due to the dierent speed of propagation of the water in the land according to itshumidity and the possibility of the evaporation of the water from the soil volume sample,a small container, whose dimensions were slightly larger than those of the sensor, wasused. A plastic cylinder with its top open was chosen. The soil volume in the cylinderwas Vrs = 33.6 0.6 cm3. All following measurements were repeated ten times, andtheir average was chosen. To avoid the evaporation from the container, the recipient wasenveloped in a wool rag.

The cylinder was gradually lled with water and the current output was measured witha Fluke multimeter for a time of 2 hours, in order to have a stable current output. Therewere strong uctuations in the current, thus the arithmetic mean between the minimumand the maximum currents recorded was chosen. As error, the largest dierence betweenthe minimum and the maximum currents recorded was chosen, i.e. 0.1mA, larger thanthe instrumental precision (0.01mA). Finally, to determine the water content, the totalweight of probe, water and recipient was measured.

Two experiments were performed. The type of soil was: in the rst experiment, asandy soil sample taken from the Po river beach in Turin; in the second one, the SPCsoil of the measurement site described in sect. 2.

In table VII, the consecutive measurements carried out with sandy soil are reported.The error associated to the water weights was assumed for all measurements as thegreatest average error, i.e. 0.4 g. The last column of table VII reports the volumetric soilmoisture content ().

Looking at the table, in correspondence of the sample test number 7 a discontinuityis evident, due to the impossibility to execute consecutive measures. These values wereleft with the purpose to evidence eventual phenomena of hysteresis, which, as can beseen, in sandy soil is very small.

Also for the SPC soil measurement, the procedure was the same described above. Themeasurements are reported in table VIII. Even in this case, a strong discontinuity (tests4-5) is present in the measurements, but, in the case of loam soil, the hysteresis is mostremarkable. Two could be the reasons for this behaviour, perhaps caused by the presenceof a vertical gradient in the container: i) the gravitational drainage; ii) the evaporationof surface soil in the container. Thus, the test numbers 4 and 5 were excluded from thecalculation of the regression curve.

For the calculation of the calibration curves, two dierent kinds of curves were se-lected: linear and exponential regression. For both soils, the curves with the best corre-lation coecient were the exponential ones.

For the sandy soil, the regression curve between the volumetric soil content (inm3m3) and the current i (expressed in mA) was

(2) SAND = A exp[Bi], A = 2.86 0.06m3m3, B = 0.30 0.06.

For the loamy soil, the regression curve was

(3) LOAM = A exp[Bi], A = 1.266 0.013m3m3, B = 0.14 0.02.

To obtain the saturation ratio (q), equal to the ratio between the volumetric soil content() and the porosity (s), it is sucient to multiply the A coecients in eqs. (2) and (3) by


Table VII. List of measurements relative to sandy soil of Po river beach at Torino. Secondcolumn reports the quantity of water in the cylinder. Third column reports the current circulatingthrough the sensor. Fourth column reports the volumetric soil content evaluated as the ratiobetween the water content (2nd column) and the total volume.

Test Water Current number weight (0.1mA) (0.02m3/m3)

(0.4 g)1 0.0 20.8 0.00

2 2.4 12.9 0.07

3 3.8 10.4 0.11

4 4.8 9.8 0.14

5 5.4 9.5 0.16

6 8.0 9.2 0.24

7 2.8 11.8 0.08

8 4.6 9.6 0.14

9 6.2 9.3 0.18

10 6.8 9.0 0.20

11 8.4 8.9 0.25

12 10.4 8.5 0.31

13 10.8 7.3 0.32

14 12.2 6.6 0.36

Table VIII. Same of table VII but for loamy soil (SPC station).

Test Water Current number weight (0.1mA) (0.02m3/m3)

(0.4 g)1 0.0 20.7 0.00

2 3.4 18.3 0.10

3 5.2 14.0 0.15

4 1.8 14.7 0.05

5 2.0 14.6 0.06

6 5.8 14.1 0.17

7 7.0 11.3 0.21

8 10.8 10.1 0.32

9 13.0 8.8 0.39

10 15.0 7.4 0.45


Fig. 5. Graphic of the behaviour of the relationship = (i) given by eq. (4) for loam soilcompared with the measurements performed in this work (squares) and with the experimentalpoints (diamonds) fournished by the probe manifacturer.

the porosities for sandy soil (s = 0.395m3m3) and for loamy soil (s = 0.451m3m3),according to [8].

In g. 5, the diagram of the relation = (i) for the loamy soil (eq. (3)) is comparedwith the measurements carried out in this work (squares) and with the measurementssupplied from the manufacturer in the manual (diamonds). As can be seen, there is a fullcorrespondence between the values supplied from the manufacturer for a generic loamsoil (composed of 74% of sand, 15% of lime and 11% of clay) and the curve of eq. (3).

Regarding the dependence of the measured current (i) from the temperature (T ) of thesensor, two measurements were carried out using the same loamy soil. On this occasion,the same quantity of soil moisture was used. The two measurements were performed atthe temperatures T1 = 20 C and T2 = 1 C, giving current values i1 = 12.0 0.1mAand i2 = 12.9 0.1mA, respectively. Using these two points, the angular coecient forthe temperature dependence was evaluated as = 0.0474 0.0008mA/C. Therefore,assuming as reference temperature the value T0 = 20 C, the corrected current icorrreferred to the reference temperature T0 could be evaluated from the current i measuredat the generic temperature T by means of the relationship

(4) icorr = i+ (T0 T ).

5.2. Vertical profile data. Figure 6 shows the time trend of the wind velocity prolesrecorded by the Mini-Sodar during the passage of the cold front on the evening of the17th at the SPC station. From these plots, it is evident the clockwise rotation of thewind direction from NW to NE and its strengthening after 8.30 p.m., thus it is possibleto track the front passage at surface between 8.00 p.m. and 8.30 p.m.

In g. 7, the prole of dry- and wet-bulb temperatures measured by Airsonde onthe 21st of June 1993 at 2.54 p.m. are shown. Under these anticyclonic conditions, the


Fig. 6. Time trend of minisodar horizontal half-hourly wind proles between 6 p.m. and 12 p.m.(local time) of the 17th of June; the up-down oriented arrows indicate northern provenience.

atmosphere was nearly neutral, with a weak inversion at about 20002200m capping themixed boundary layer, whose depth was located at about 2000m. Above this height,the thermal gradient was roughly constant and adiabatic, indicating the presence ofconvective motions. In the air layer immediately below the 2000m inversion (from 1600to 1900m), dry- and wet-bulb temperature nearly coincided, indicating the presence ofsome stratocumulus and fair-weather cumulus, eectively observed in that day at SPCand whose evidence was indirectly conrmed also by the scattered values of global andnet radiation on the 21st (g. 2).

In g. 8, the prole of dry- and wet-bulb temperatures measured with the higher-resolution system Tethersonde on the 17th June 1993 at 6.47 a.m. are shown. The airlayer under 200m, moister than the upper one, was marked by a weak thermal inversion,whose depth (200m) was in good agreement with the experimental law of [9] and [10]:

(5) zi = At,

where A = 70 if t is the time (in hours) elapsed from sunset and zi the inversion layer


Fig. 7. Airsonde prole of the 21st of June at 2.54 p.m. (local time) of dry- (solid line) andwet-bulb (crosses) temperatures, in degrees.

depth (in m). In the lower 50m there was a second, moister and strongest inversionlayer, probably caused by the solar radiation, which favoured the dew evaporation andthe consequent cooling of air, and a weak air mixing near ground.

These two plots show an example of the quality and quantity of information that couldbe inferred from such kind of proles. During the entire SPCFLUX93 eld experiment,38 proles with Airsonde and 48 proles with Tethersonde were carried out. In the 92%of cases Airsonde proles reached 2500m, and Tethersonde proles reached 450m.

5.3. Chemical data. Figures 9 and 10 show the time trend of SO2, NO and NO2atmospheric concentrations: the measured values were typical of rural site, with meanvalues of 1, 2 and 8 ppb, respectively and peaks of 13 ppb for SO2 and of 26 ppb for NOand NO2.

The O3 concentration, reported in g. 11, showed a typical photochemical daily cycle,with the exception of the days 23rd-25th, when the diurnal insulation was scarce, dueto the cloudiness related to the fronts. The peak-averaged values were 80 and 90 ppb


Fig. 8. Tethersonde prole of the 17th of June 1993 at 6.47 a.m. (local time) of dry- (solidline) and wet-bulb (dashed line) temperatures, in degrees.

during the overall campaign.The deposition velocity of O3 was evaluated by means of the eddy correlation tech-

nique, applied to uctuations of the vertical wind component measured by a sonicanemometer and the ozone concentration measured with a high-frequency analyser: thevalues ranged between about 15 and 85 cm s1.

Dry deposition was evaluated with the inferential technique [11], using meteorologicalstandard data and concentrations of some gaseous species. The concentrations of SO2 andNO2 were measured with an automatic analyser, those of HNO2, HNO3 and NH3 withan anular denuder, while ne particulate was measured with a lter pack. Sulphur andnitrogen dry depositions, integrated over the entire SPCFLUX93 period, were evaluatedto 26.0 and 47.5mgm2, respectively.

A dry-wet sampler performed the measurements of the wet deposition; the computedvalues resulted 5.5 and 4.5mgm2, respectively. Thus, the sulphur and nitrogen to-tal (i.e. wet plus dry) depositions during the campaign were estimated to be 31.5 and52.0mg/m2, respectively.


Fig. 9. Time trend of SO2 (location I1), in ppb, during the whole campaign.

5.4. Ultrasonic anemometer data analysis and processing . The data of two SolentResearch sonic anemometers (manufactured by Gill Instruments Ltd) installed at 10m(location S1) and 25m (S2) were processed with the SONELA (SONic anemometer dataELAboration) model, described in [12] and [13], and more extensively in [14]. Accordingto this model, three consecutive rotations are imposed to the reference frame:

Fig. 10. Time trends of NO and NO2 (locations I1-I3), in ppb, during the whole campaign.


Fig. 11. Time trends of O3 (location T1A), in ppb, and global radiation (location T1B), inWm2, during the whole campaign.

a) the x coordinate of the reference system is aligned with the mean horizontal wind(v = 0), and the rotation angle between the mean horizontal wind direction andthe North direction is then computed;

b) the x coordinate of the reference system is aligned with the mean 3D wind vector(w = 0), and the angle between the mean wind vector and the mean horizontalwind velocity is then calculated;

c) a rotation around the x-axis is performed, to ensure that vw = 0; the angle related to this rotation is then calculated [15].

These numerical operations make the anemometer set up with its x-axis along the stream-lines. They were necessary to avoid the large errors which could occur in the calculationsof turbulent uxes when the wind sensor is not perfectly vertical (misalignment problem)or when the mean streamline is not perfectly horizontal.

The SONELA model calculated some 30 minute averaged quantities such as hori-zontal and vertical wind speeds, sonic temperatures, their standard deviations, winddirections, rotation angles ( and ), and the second-, third- and fourth-order crossedstatistical moments of the uctuating quantities, including momentum, sensible, andlatent heat uxes.

The mean horizontal wind speeds measured by sonic anemometers at locations S1 andS2 were compared with the ones measured by the slow-response anemometers (locationT2, at 10 and 30m). The intercomparisons showed a good correlation between the twodierent sensors, with only a little o-set lower than 0.5ms1. The intercomparison ofthe sonic temperature Ts (dened as the temperature calculated from the sound velocity)measured with sonic anemometer was also in good agreement with the sonic temperatureevaluated using the observed values of temperature T and humidity detected by the slowresponse instruments installed in location T2 at the nearest level using the expression

(6) Ts = T (1 + 0.51q),


even if, in this case, an evident systematic error (perhaps due to the incorrect precisionof sonic path of the instrument), which was quantied in 1.5 C, was present. Despiteits magnitude, this systematic error did not aect the turbulent ux evaluation as it wasrelated only to the mean temperature and not to its uctuations.

5.5. Turbulent heat fluxes assessment . When data of net radiation Rn and soil-atmosphere conductive ux G0 (the interface heat ux between soil and atmosphere) areavailable, it is possible [12] to balance the heat energy budget equation at the air-soilinterface:

(7) Rn = H + LE +G0,

where H = cpwT is the sensible heat ux, is the atmospheric pressure, cp is thespecic heat at constant pressure, LE = wq is the latent heat ux, the latent heatof evaporation (and/or fusion), and q the specic humidity of air. Turbulent heat uxescan be evaluated starting from the measurements of the value wT s , which includes thecontributions of both sensible and latent heat uxes, according to the expression

(8) wT s = wT (1 + 0.51q) + 0.51Twq.

It is dicult to obtain a reliable quantitative estimate of G0 by instruments, becauseG0 is the ux at the interface between atmosphere and soil surface, two dierent medi-ums with dierent properties. A possible estimation could be performed by using theobservations of soil heat ux G and of soil temperature T (both measured at the samelevel underground) with the formula [16]

(9) G = G0 +(cT

z

)z,

where the values for the soil heat capacity c were evaluated according to the soil typeand moisture. In SPC soil was loam, and during SPCFLUX93 experiment, soil moisture(location T1B) was approximately constant during the experiment and equal to about41% of porosity (see also subsect. 6.4).

Sensible and latent heat uxes evaluated using eqs. (7), (8) by ultrasonic anemometermeasurements were compared with the ones directly measured by the uxmeter (bothinstalled at S1 location). In gs. 12 and 13 the scatter diagrams of these two turbulentheat uxes show that the uxes evaluated by the ultrasonic anemometer measurementswith SONELA model reproduced quite well the measured ones.

6. Discussion on few main interesting results

After the conclusion of the eld experiment, some analyses were carried out. Themain arguments were

to analyse the characteristics of atmospheric turbulence in the surface layer throughthe examination of sonic anemometers and other fast response data;

to determine the soil moisture value according to the SPC soil type; to validate some numerical schemes (solar radiation and soil temperature) through

the use of synoptic observation and ground-based and shelter data;


Fig. 12. Scatter diagram of sensible heat uxes respectively evaluated with ultrasonic anemome-ter (x-axis, location S2) and SONELA model and measured by using Campbell uxmeter (y-axis,location S1), in Wm2, during the whole campaign.

to develop and validate an algorithm to establish meaningful initial conditions forbiospheric models starting from synoptic observations;

to validate the energy budget in biospheric models; to evaluate mixed layer depth using radon ux data; to estimate deposition velocity using dry and wet deposition measurements.

Fig. 13. Scatter diagram of latent heat uxes respectively evaluated with ultrasonic anemome-ter (x-axis, location S2) and SONELA model and measured by using Campbell uxmeter (y-axis,location S1), in Wm2, during the whole campaign.


Fig. 14. Spectra of U , V and W wind components on 16th June at 2.00 p.m. during unstableconditions (z/L = 1.07 and u = 0.36m/s). Lines without markers represent the energy|(f)|2 (Su, Sv and Sw), and lines with markers the power f |(f)|2 of spectra (fSu, fSv andfSw).

In the following subsections, the most interesting results will be summarised.

6.1. Spectra of wind velocities. Spectra of the three wind velocity components (U ,V and W ) were generated for the whole observation period of SPCFLUX93.The spectrashowed behaviours similar to the expected ones for a at and rural region like the PoValley. Namely

U and V components showed similar spectra, and this result was consistent withthe observed similarity of their standard deviations;

spectra of W component were dierent from the U and V ones, both in stable andconvective conditions; the similarity between U , V and W spectra was observed onlyduring strong wind episodes, when turbulence was mainly mechanical;

for all wind velocity components, the peak frequencies in stable conditions werehigher (and their associated amplitudes lower) than the corresponding ones in convectiveconditions;

during convective conditions, spectra of U and V components sometimes showedtwo maxima: the greatest one had the lowest frequency and was related to the buoyancy,while the smallest one, induced by the shear stress, had the highest frequency.

Nevertheless, some new peculiarities were found with respect to the spectra normallyreferenced in the literature. In fact, since the Po Valley is usually characterised bylow wind conditions, stable and convective conditions were often observed, and theirintensities were greater than in the Kansas (1968) and Minnesota (1973) experiments.In particular, the most interesting observations were:

During unstable conditions, in U and V spectra, the peak amplitudes (|(f)|2 forthe energy and f |(f)|2 for the power) in the low-frequency region were more pronouncedthan the one relative to the high frequencies, which sometimes disappeared (g. 14) or


Fig. 15. Same of g. 14 but at 10.00 p.m. during stable conditions (z/L = 0.52 and u =0.16m/s).

was masked. This phenomenon was associated to free-convection cases, where buoyancyeects were dominant over shear stress.

In the nocturnal stable cases, turbulence spectra of all wind components exhibitedtwo maxima (g. 15); the lowest frequency peak was due to gravity waves, while thehighest was due to mechanical turbulence.

In extremely stable cases (with the ratio z/L > 10, where z is the quote and L theMonin-Obukhov length), the mechanic turbulence was negligible, friction velocity wasabout 102ms1, and small values of w were observed. In these cases, W spectra werenot signicant because the data were near the lower threshold of the sonic anemometerrange (102m/s). Also spectra of U and V , in the high-frequency region, exhibited anunrealistic positive slope, and only the peak due to gravity waves was evident.

6.2. Radiation parameterisation. The knowledge of the solar incoming radiation isimportant because this is the main term, during daytime, entering in the net radiationformulation. The correct representation of the net radiation in numerical models isfundamental because it represents the energetic input that is partitioned into sensible andlatent turbulent heat ux, and conductive ux, according to the surface characteristics.A check of the radiation input is then a crucial factor to test the correct behaviour ofthe surface parameterisations of a numerical circulation model.

Using the information coming from the synoptic observation (location T4) relative tothe cloud cover, an algorithm (taken from [17]) calculating the solar incoming radiationstarting from the observations of high and low-middle cloudiness was checked. Thecomparison with the observations showed that the proposed parameterisation [16] workswell in clear-sky conditions, while in cloudy days the calculated solar radiation wasslightly higher than the observations.

As an example, in g. 16 the time series of observed (points) and simulated (solidline) net radiation during the whole period of SPCFLUX93 campaign are reported. The


Fig. 16. Time trends of net radiation observations carried out at location T1B (dots) and ofLSPM output (solid line), in Wm2, during the whole campaign.

agreement between data and model predictions is good, even if some little overestimatesor underestimates of maximum values at noon are present. These disagreements areperhaps caused by errors in the interpolation of cloudiness from synoptic observations(used to calculate the solar radiation) or by the inaccuracy of the global radiation pa-rameterisation in cloudy days.

6.3. Initialisation of SVAT models. The proximity of the WMO station (locationsP4 and T4) to the SPC site allowed to test a method (described in [16] and in [18]) toderive from synoptic observations (which are widespread) the necessary data to driveSVAT (Soil Vegetation Atmosphere Transfer) schemes.

As known, a SVAT (Soil-Vegetation-Atmosphere Transfer) scheme needs some bound-ary conditions: air pressure, temperature, humidity information (relative, specic, dew-point, etc.), wind vector, precipitation, solar radiation or cloudiness.

This method was tested using the SPCFLUX93 data. The results [19,20] showed thatair temperature and relative humidity data are well reproduced, wind velocity is correctlyreproduced (at least for the daily trend), while precipitation is somehow smoothed, be-cause synoptic data only provided the information of cumulated precipitation in the last6 hours. From these conclusions, we can state that this methodology can be safely usedfor climatological purposes.

6.4. Initialization of soil temperature and soil moisture in numerical models. Fre-quently soil parameters are required in SVAT schemes or LAMs (Limited Area Models)in order to initialize the values at the beginning of the simulations. The soil values arerecognized as very important parameters able to aect in a substantial way the surfaceturbulent heat uxes and, thus, the atmospheric stability.

However, measurements of soil temperature and moisture on a global or mesoscalearea are not available. There are some locations in which those data are measured,but they are too sparse and sometimes not representative of the surrounding area. Thesatellite images can be also used to provide the surface values of soil temperature and


moisture, but they cannot provide the values for the deepest soil layers. For a simulationof medium-range weather forecast (710 days), the knowledge of parameters in the rst2050 cm of soil is required.

For this reason, in the recent years, the Meteorological Services are trying to ndsome methods to infer the values of soil parameters on a global or mesoscale region.

In this subsection, two simple methods were derived for the initialization of soil tem-perature and moisture, respectively. These algorithms contained some empirical param-eters to be determined through a calibration over the experimental site. Nevertheless, asthe equations were based on physical processes, these methods could be generalised andused for many dierent sites.

For soil temperature, it is useful to remember that the heat conduction equation canbe written as

(10)T

t=

2T

z2,

where T = T (z, t) is the soil temperature, z is the axis directed downward into soil and isthe soil thermal diusivity. A solution of this equation can be found under the hypothesisof constant and with the following boundary condition at the soil-atmosphere interface:

(11) T (0, t) = T0 +T sin(t+ ),

where T0 is the average soil temperature in all soil layers during the cycle = 2/, is the angular frequency, T is the amplitude of the thermal wave and the phase. Thesolution of eq. (10) with the boundary condition (11) is

(12) T (z, t) = T0 +T exp[ zD

]sin

(t z

D+

),

where D is the depth of exponential decay for the temperature, and D = .Based on eq. (12), we proposed the following empirical equation:

(13) Temp(z, t) = Tair +Texc exp[ zD

]sin

(2365

zD

+ )

for the evaluation of soil temperature starting from the following data:

the air temperature Tair;

the observed yearly excursion of mean air temperature Texc = TJuly TJanuary; the Julian day of the year G.

This formula was tested for SPC, using the decadal values and the following parameters:Texc = 14 C (taken from the climatology of the site), D = 2.3m (typical for a loamysoil) and = 1.64 rad. Figure 17 shows the comparison among the decadal meansof the data measured at SPC from June 1993 to February 2005 (the soil temperatureand moisture sensors were left on the measurement site until end of February 2005) andthe corresponding ones evaluated using eq. (13) for the two reference levels of 5 cm and100 cm. The temperatures evaluated with eq. (13) seemed to give a realistic estimate ofthe observed soil temperature.


Fig. 17. Soil temperatures (decadal values) measured at SPC at the depths of 5 (Tobs-5, squares)and 100 cm (Tobs-100, diamonds) compared with the corresponding ones evaluated by eq. (13)(Tsoi-5, solid line, and Tsoi-100, dashed line) for the period June 1993-February 1995.

The square correlation coecient for the temperatures at 5 and 100 cm were 0.93and 0.95, respectively, while the biases (observed minus evaluated) were 0.6 C and 0.2 C,respectively.

Concerning soil moisture, in this case the propagation of moisture into soil obeys tothe following equation [21]:

(14)q

t=

1s

z

[K

z(z + )

],

where q is the saturation ratio, s the soil porosity, K the hydraulic conductivity, and the moisture potential (eq. (14) do not consider the eventual input-output of waterdue to evapotranspiration and precipitation). An analytical solution of eq. (14) cannotbe derived due to the strong dependence of the hydraulic conductivity and moisturepotential on soil moisture itself. Nevertheless, we must consider that

i) during normal conditions, surface soil shows larger soil moisture uctuations thandeepest soil, with a yearly cycle showing its minimum during the warmest months(when evaporation is generally larger) and conversely its maximum during thecoldest months;

ii) deepest soil shows the lowest variations and its soil moisture content approachesthe eld capacity;

iii) excluding arid conditions, wintertime soil moisture approaches the eld capacity;

iv) on the occasion of very strong precipitation events, soil moisture can exceed tem-porarily the eld capacity, but when rainfall conditions ended, due to the very highhydraulic conductivity, soil moisture rapidly decreases to the eld capacity;


Fig. 18. Soil moisture (decadal values) measured at SPC at the depth of 10 cm (Qobs-10,squares) compared with the corresponding one evaluated by eq. (15) (Qsoi-10, solid line) for theperiod June 1993-February 1995.

v) periods characterised by precipitations above the normal are generally also charac-terised by relative humidity values above the normal, and vice versa.

Based on these considerations, we postulated the following relationship:

(15) qemp(z, t) = qfc (qfc qwi)(RHmax RHairRHmax RHmin

)exp

[ zD

],

where qemp is the empirical soil moisture (expressed in units of saturation ratio = /s),qfc the eld capacity, qwi the wilting point (both expressed in units of saturation ratio),RHmin and RHmax are the air relative humidity thresholds, and RHair is the air relativehumidity. The fraction including relative humidities must range between 0 and 1.

Equation (15) was tested for SPC, using the decadal values and the following param-eters: qfc = 0.761, qwi = 0.343 (typical of loamy soil), RHmin = 65%, RHmax = 90%(taken from the climatology of the site), and D = 2.3m (typical for a loamy soil). Fig-ure 18 shows the comparison among the decadal means of the data measured at SPCfrom June 1993 to February 2005 and the corresponding ones evaluated using eq. (15)for the reference level of 10 cm. Generally speaking, there is a qualitatively good agree-ment between data predicted by eq. (15) and observations, even if some extreme values,particularly the minimums, are sometimes not well captured by eq. (15).

The square correlation coecient was 0.82 and the bias (observed minus evaluated)was 0.03.

6.5. Energy budget . Sensible, latent and ground-atmosphere heat uxes, evaluatedas explained in sect. 5, and the net radiation data measured in location T1B, were usedto validate the output of a simulation performed with the SVAT scheme LSPM (LandSurface Process Model [22]). The assumed vegetation type was short grass. In orderto avoid the so-called spin-up problem (the climatic system can require a few monthsbefore the initial values of the soil moisture elds are forgotten), initial values of soiltemperature and moisture were estimated using eqs. (13) and (15). It has been alsoveried that these values were similar to those predicted by LSPM after a 6-months runstarted on January 1st, 1993, i.e. six months before the beginning of the SPCFLUX93


Fig. 19. Time trends of sensible heat ux observations carried out at location S1 (dots) andof LSPM output (solid line), in Wm2, during the whole campaign.

experiment (the period of 6 months was chosen because [16] demonstrated that LSPMwas able to reach its equilibrium state in this period).

The boundary conditions necessary for the LSPM run (screen-level temperature, rela-tive humidity, wind, sea level pressure, precipitation and cloud coverage) were extractedand arranged from the data of the close synoptic station by using the algorithm de-scribed in subsect. 6.3. The necessary radiation input (long and short wave) for the twomodels were simulated using the empirical package of LSPM, based on [17], as told insubsect. 6.2.

Fig. 20. Time trends of latent heat ux observations carried out at location S1 (dots) and ofLSPM output (solid line), in Wm2, during the whole campaign.


Fig. 21. Time trends of air-soil heat ux data evaluated with eq. (6) by using observationscarried out at locations S1 and T1B (dots) and of LSPM output (solid line), in Wm2, duringthe whole campaign.

Figures 19 and 20 show sensible and latent heat uxes evaluated using ultrasonicanemometer (points) and calculated by LSPM (solid line). LSPM predictions of bothheat uxes were close to the observed data in their whole range. Only few maximumvalues of the latent heat ux were underestimated, particularly in concurrence with largeerrors in the evaluation of net radiation or in cloudy-sky conditions (g. 17).

Figure 21 shows that the heat ux G0 at the ground-atmosphere interface evaluatedfrom observations using eq. (4) are in good agreement with the predictions of LSPM,with the exception of a little overestimate during daytime around noon and the presenceof strong negative minima (calculated) at sunrise.

6.6. Thermal wave propagation into soil . The data referring to temperature measure-ments at 7 levels in the rst meter of soil allowed us to test dierent soil parameterisationschemes. In particular, we compared two widely used soil schemes: the (5-level) multi-layer scheme, used for instance in the LSPM [22], and the (3-level) force-restore scheme,used for instance in the BATS [23]. This comparison was performed running BATS andLSPM (initialised in the same way as explained in subsect. 6.4) for a 6 months simulation(starting from 1st January 1993). Synoptic data arranged with the method describedin subsect. 6.3 were used as initial and boundary conditions for both models. As shownin [16], the multilayer scheme produced a more accurate estimate of the thermal wavepropagation with respect to the results coming from the use of the classic force-restoremethod.

6.7. Evaluation of mixed layer depth using radon flux . The vertical proles performedwith Airsonde (P2 location), Tethersonde (P3 location) and with the radiosoundings(P4 location) were used to calculate the observed Mixed Layer Depth (MLD) duringSPCFLUX93. These data allowed to test the capability of 4 dierent methods to predictMLD: the Holzworth method [24,25], the EPA-METPRO pre-processor [26], the Gryning-Batchvarova model [27] and the 222Rn box-model [28].


Fig. 22. Comparison of O3 deposition velocity calculated with Multi-Layer (solid line) and Big-Leaf (dashed line) schemes and evaluated from eddy correlation measurements (points) during15-23 June.

The Holzworth method predicts the MLD using the vertical temperature prole at 06UTC (i.e. 08 a.m. in local time) and the 2m temperature time trend. The MLD is thenevaluated as the interception of the observed morning prole and of the adiabatic lapserate starting from the daily 2-m temperature maximum value.

The EPA-METPRO pre-processor predicts separately the MLD for stable, neutraland unstable conditions using the Nieuwstadt and Van Dop method [29], the Zilitinkevicformulation [30] and the Carson model [31] modied byWeil and Brower [32], respectively.

The Gryning-Batchvarova model evaluates the MLD from the assessment of sensibleheat ux vertical prole during daytime, while a dierent formulation is used for nighttime hours.

The 222Rn box-model considers the atmosphere as a box of unit area and whose heightis the MLD. Under the following four hypotheses (the unique Rn sources and sinks arethe soil Rn ux and the radioactive decay, respectively; in the mixed layer the mixingis uniform; the horizontal Rn advections are negligible), the MLD evolution is evaluatedon the basis of a mass-balance equation.

The comparison of the four methods with the SPCFLUX93 observations [33] showedthat the three conventional methods provided a good approximation of the observedMLD, while the box-model underestimated the MLD, also during strongly convectiveconditions.

Due to the fact that the assumption of uniform mixing in the mixed layer was notstrictly satised even if the convection was well developed (as in the case of thermalconvection), the box-model could not give a realistic estimate of the MLD. Nevertheless,in some cases, the box-model took as a MLD the level where a discontinuity in thetemperature and humidity vertical proles is present. This model can then be useful toinfer some information related to the diusive characteristics of low atmosphere.


6.8. Estimate of O3 total deposition in the Po Valley . A short report is given hereon the O3 total deposition evaluation (dry, wet and fog), in the rural test site, accordingto the long-term project estimation.

To quantify dry deposition, direct measurements of meteorological and chemical pa-rameters were used to run inferential technique models [34]. Firstly, the inferentialmethod by using the Multi-Layer [35] and Big-Leaf [36] approaches was tested. Thesemodels were compared against the eddy correlation technique [37] and the results duringSPCFLUX93 are displayed in g. 22.

Day-night cycle strongly aects deposition processes: the maximum values (0.5 cm/s)were reached at noon and the minimum values during night time (0.05 cm/s). Deposi-tion velocity from Eddy Correlation was compared with Multi-Layer (ML) and Big-Leaf(BL) formulations. Both BL and ML models showed a good agreement, even if the BLexhibited larger peaks during the day. These dierences could be due to the fact that MLscheme calculates the stomatal resistance using the information based on temperaturevertical prole (for the stability) and solar radiation (for the stomata activity).

7. Conclusions

A eld experiment was carried out at San Pietro Capoume (Po Valley, Italy) duringJune 1993. The location was selected as representative of the Po Valley, in order tosupply a suitable dataset with the aim to investigate the following topics: atmosphericturbulence, dry and wet atmospheric total deposition, energy balance, thermal wavepropagation in the soil.

This paper presented a detailed discussion on instrumentation set-up, data collectionand test of their reliability in the climatic conditions of the Po Valley, in view of providinguseful observational information to the scientic community. Moreover, considering themore recent MAP experiment, the SPCFLUX93 dataset, referring to the lee side of theAlps, was reanalysed for allowing a comparison with the observations carried out duringMAP around the mountainous regions, in order to hopefully improve the understandingof the perturbations on the PBLs of synoptic and mesoscale airows crossing the Alps.

The instruments installed during SPCFLUX93 eld experiment were: low-responsesensors for standard meteorological observations, fast-response sensors (ultrasonic anemo-meters, hygrometers and thermometers), in-situ vertical prolers (Airsonde and Teth-ersonde balloons), remote-sensing instruments (Mini-Sodar) and some instrumentationused to measure physic and chemical data (dry and wet atmospheric deposition).

The database contains the mean values of all data and the instantaneous readings ofultrasonic fast-response sensors.

The preliminary and most interesting result gathered by reanalysing SPCFLUX93data were related to the following topics:

analysis of the characteristics of turbulence in the surface layer through the pro-cessing of ultrasonic anemometers and other fast-response data;

validation of some numerical schemes (solar radiation and soil temperature) throughthe use of synoptic observation, and ground-based and shelter data;

validation of the energy budget of biospheric models; development and validation of an algorithm for initialising biospheric models based

on synoptic observations; evaluation of mixed layer depth using radon ux data; estimate of deposition velocity using dry and wet deposition measurements.


All participants to the SPCFLUX93 eld experiments are acknowledged for the in-

strumental assistance, and namely Mr. D. Bertoni (Department of General Physicsof the Torino University), V. Colombo and F. Rocchetti (of the ENEL S.p.A.-Environmental Area), and M. Catenacci of CESI, for the skilled assistance providedduring the executive phase of the campaign SPCFLUX93. Last but not least, RegioneEmilia Romagna and Italian Meteorological Service are acknowledged for allowing theuse of the data coming from their instrumentation, and ECMWF is acknowledged forthe use of synop observations.

REFERENCES

[1] Bougeault P., Binder P., Buzzi A., Dirks R., Houze R., Kuettner J., Smith R. B.,Steinacker R. and Volkert H., Bull. Am. Meteorol. Soc., 82 (2001) 433.

[2] Bougeault P. and Binder P., WMO Bulletin, 51 (2002) 14.[3] Wallen C. C., in World Survey of Climatology, edited by CHIEF: Landsberg H. E.,

Vol. 6 (1970), pp. 127-183.[4] Egger J. and Tafferner A., J. Atmos. Sci., 47 (1990) 2417.[5] Lott F., Q. J. R. Meteorol. Soc., 121 (1995) 1323.[6] EMB (European Meteorological Bulletin), Druck und Verlag: Deutscher

WetterdienstOenbach, June 1993.[7] Manzi G., Galofaro D., Morselli M. G. and Brusasca G., Campaign of S. Pietro

Capoume, June 1993, Internal Report No. E1/94/04 of ENEL S.p.A. (Milano)Italy(1994).

[8] Clapp R. B. and Hornberger G. M., Water Resour. Res., 14 (1978) 601.[9] Anfossi D., Bacci P. and Longhetto A., Atmos. Environ., 8 (1976) 537.[10] Anfossi D., Bacci P. and Longhetto A., Q. J. R. Meteorol. Soc., 102 (1976) 173.[11] Catenacci G., Cavicchioli C., Manzi G. and Brusasca G., Models for the estimate

of dry deposition uxes of gaseous and particle pollutants, Internal Report No. CISE-SAA-94-133 of ENEL S.p.A. (Milano)-Italy (1994).

[12] Cassardo C., Sacchetti D., Morselli M. G., Anfossi D., Brusasca G. andLonghetto A., Nuovo Cimento C, 18 (1995) 419.

[13] Ferrarese S., Bacci P., Bertoni D., Cassardo C., Forza R., Giraud C.,Longhetto A., Morselli M. G., Pangia M. and Purini R., Nuovo Cimento C, 21(1998) 357.

[14] Anfossi D., Cassardo C., Ferrero E., Longhetto A., Sacchetti D., BrusascaG., Colombo V., Marzorati A., Morselli M. G., Rocchetti F. and Tinarelli G.,On the analysis of ultrasonic anemometer data, Internal Report No. 282/93 of Istituto diCosmogeosica, CNR (National Research Council) 1993.

[15] Mc Millen R. T., Boundary-Layer Meteorol., 43 (1988) 231.[16] Ruti P. M., Cassardo C., Cacciamani C., Paccagnella T., Longhetto A. and

Bargagli A., Contrib. Atmos. Phys., 70 (1997) 201.[17] Page J. K., Prediction of Solar Radiation on Inclined Surfaces, edited by Page J. K. (D.

Reidel Publishing Company) 1986.[18] Cassardo C., Carena E. and Longhetto A., Nuovo Cimento C, 21 (1998) 189.[19] Cassardo C., Ruti P. M., Cacciamani C., Longhetto A., Paccagnella T. and

Bargagli A., MAP Newsletters, 7 (1997) 24.[20] Cassardo C., Balsamo G. P., Pelosini R., Cacciamani C., Cesari D., Paccagnella

T. and Longhetto A., MAP Newsletters, 11 (1999) 26.[21] Pielke R. A.,Mesoscale Meteorological Modelling, 1st edition (Academic Press, New York,

N.Y.) 1984.[22] Cassardo C., JI J. J. and Longhetto A., Boundary-Layer Meteorol., 72 (1995) 87.


[23] Dickinson R. E., Henderson-Sellers A., Kennedy P. J. and Wilson M. F.,Biosphere-Atmosphere Transfer Scheme (BATS) for the NCAR Community ClimateModel, NCAR Technical Note (December 1986).

[24] Holzworth G. C., Mon. Weather Rev., 92 (1964) 235.[25] Holzworth G. C., J. Appl. Meteorol., 6 (1964) 1039.[26] Paine R. J., Users guide to the CTDM meteorological preprocessor (METPRO) program,

US-EPA Report EPA/600/8-88/004 (1988).[27] Gryning S. and Batchvarova E., Boundary-Layer Meteorol., 56 (1990) 261.[28] Fujinami N. and Esaka S., Radiation Protection Dosimetry, 24 (1988) 89.[29] Nieuwstadt F. T. M. and Van Dop H., Atmospheric Turbulence and Air Pollution

Modelling (Reidel Publ. Co., Dordrecht) 1982.[30] Zilitinkevic S. S., Boundary-Layer Meteorol., 3 (1972) 141.[31] Carson D. J., Q. J. R. Meteorol. Soc., 99 (1973) 450.[32] Weil J. C. and Brower R. P., Estimating convective boundary layer parameters for

diusion applications, draft report prepared by Environmental Center, Martin MariettaCorp., for Maryland Dept. of Natural Resources (1983).

[33] Faggian P., Riva M., Capra D., Brusasca G. and Queirazza G., Acqua-Aria, 6(1996) 601.

[34] Cavicchioli C., Manzi G., Brusasca G. and Catenacci G., in Emissions, Deposition,Laboratory Work and Instrumentation, edited by Borrell P. M., Borrell P., CvitasT. and Seiler W., Vol. 2 (1997), pp. 297-300.

[35] Meyers T. P. and Baldocchi D. D., Tellus, 40b (1988) 270.[36] Erisman J., Van Pul A. and Wyers P., CEC Air pollution, Research Report, 47 (1993)

215.[37] Baldocchi D. D., Hicks B. B. and Meyers T. P., Ecology, 69 (1988) 1331.

2006 a Reanalysis of the Atmospheric Boundary

Documents