Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://site2010.sai.msu.ru/static/doc/AOtarola_site2010.pdf Äàòà èçìåíåíèÿ: Mon Oct 18 13:30:52 2010 Äàòà èíäåêñèðîâàíèÿ: Mon Oct 1 20:09:08 2012 Êîäèðîâêà:

Characterization of Precipitable Water Vapor at the TMT Candidate Sites in Northern Chile: The Importance of Surface Data
A. OtÀrola T. Travouillon, M. SchÆck, S. Els, R. Riddle W. Skidmore, R. Dahl, D. Naylor, R. Querel

Kislovodsk, October 4-9, 2010

1


Outline
Location of sites tested by TMT in northern Chile Synoptic-scale meteorological conditions Monitoring of PWV at the TMT candidate sites
­ ­ ­ ­ Satellite studies: Andre Erasmus and collaborators Infrared radiometers (20 mm) Radiosonde soundings from the Antofagasta station Surface weather data: assuming an underlying PWVvertical profile model

Results
­ Overall, seasonal, PWV: night time results

Final remarks
­ PWV from surface data & the water vapor scale height ­ Possible methods to infer the water vapor scale height
Kislovodsk, October 4-9, 2010 2


Location of the TMT candidate sites in northern Chile
Tolar 2.290m ALMA

Armazones Paranal 3.064m

100 km

Tolonchar 4.480m

Kislovodsk, October 4-9, 2010

3


Synoptic-scale meteorological conditions in northern Chile

[1] [2] [3]

Downward phase of the Hadley Cell (relatively dry air) The High Pressure Center with subsidence of dry air, creates a temperature inversion layer over the Pacific The Andes mountain deflects the zonal mean flow northward and helps maintaining the south Pacific subtropical anticyclone
Kislovodsk, October 4-9, 2010 4


Monitoring of PWV at the TMT candidate sites
Preliminary Studies (Erasmus & van Staden 2001, 2002, 2003)
­ Uses atmospheric emission at 6.7 mm observed from orbital platform (GOES-8)

IRMA, Infrared Radiometry (data series too short)
­ Receivers designed and built by scientists at the University Of Lethbridge. 3rd generation of the IRMA receiver was supported by TMT ­ Detects atmospheric radiances at 20 mm from surface ­ Needs a Line-by-Line Radiative Transfer Model ­ Needs information on the atmospheric temperature lapse rate and water vapor scale height. Obtained from analysis of radiosonde data

Weather station data (helpful for determination of long-term stats)
­ Uses atmospheric temperature and relative humidity at surface level ­ Needs information on water vapor scale height. Obtained from analysis of radiosonde data
Kislovodsk, October 4-9, 2010 5


Preliminary PWV studies Erasmus & van Staden (2001, 2002, 2003) PWV in range 300hPa-600hPa (10hPa increments) Clear sky conditions only
Uses atmospheric radiances observed from NOAA's GOES-8 satellite in the 6.7 mm band.
R

X

b m



True Radiance Brightness Temperature Calibrated temperature Upper Troposhere Humidity

Teff Plank Function ( R) T b Teff a
Reference: Soden & Bretherton (1993), JGR 96, 16669:16688 Soden & Bretherton (1996), JGR 101, 9333-9343 D. A. Erasmus and M. Sarazin, SPIE-4168 (2000)

UTH f (T , )

x UTH x s ( AirTemperatur e) Water vapor mixing ratio 1 PWV g


0

p

x dp

Integrated water vapor
6

Kislovodsk, October 4-9, 2010


Results from the preliminary PWV studies of Erasmus & van Staden
Caveats: [1] Satellite studies emphasize the 300hPa-600hPa region [2] The methodology produces sensible results for clear days; which might imply a bias towards relatively drier conditions. Site Cerro Tolar Cerro Armazones Cerro Tolonchar San Pedro MÀrtir Mauna Kea 13N Elevation (m) 2290 3064 4480 2830 4050 PWV Median (mm) 4.02 2.87 1.70 2.63 1.86 Tolar

Tolonchar Armazones PWV 10% (mm) 1.59 1.15 0.70 1.06 0.72
7

Kislovodsk, October 4-9, 2010


IRMA (20 mm)
University Of Lethbridge and Herzberg Institute of Astrophysics
-G= 6.95 K/km H = 1.4 km Surface Press. Surface Temp.

IRMA Calibrated Radiances PWV
References: G. J. Smith (2001), Master Thesis I. Chapman (2003), Master Thesis R. R. Querel (2007), Master Thesis Chapman & Naylor (2005)

BTRAM LBL-ML RTM

DT ±0.1% DPWV ±10% DP ±0.1% DPWV ±1% DH ±21% DPWV ±23%
Kislovodsk, October 4-9, 2010 8


The Use of Surface Weather Data to Infer PWV

Caveats: [1] The IRMA radiometers produced limited data: 172 days only Armazones [2] The observations missed a good fraction of the Fall and Spring seasons

Reason: The remoteness of the sites conspired against keeping track of the calibration of these radiometers and keeping them operating at all times.
Kislovodsk, October 4-9, 2010 9


Use of surface weather data to infer PWV at the sites in Chile
The larger dataset that we could use to learn about PWV in the atmospheric column at each of the sites in northern Chile consists on surface measurements of Temperature and Relative Humidity The limited amount of IRMA data in 2007 could be used to compare and validate results It was considered appropriate derive PWV from radiosonde soundings from the Antofagasta station. This is located ~100 km from Armazones and from Tolar, but further away west from the Tolonchar site.

Kislovodsk, October 4-9, 2010

10


How to use surface weather data to infer the PWV in the atmospheric column
Assumptions:
­ In areas dominated by high pressure systems (such as the northern part of Chile), the weather is dominated by subduction of relatively dry air (the mean vertical profile of RH is ~20%). The vertical distribution of water vapor tends to decrease exponentially with a given scale height ­ The strongest temperature inversion layer induced by the subsiding air, forms at an altitude lower than that of the sites under study. Consequently, the surface water vapor at the sites is coupled with the higher levels of the troposphere

Kislovodsk, October 4-9, 2010

11


Verification of the Assumptions using Radiosonde data from Antofagasta

JULY (27 profiles) : Temperature Inversion is strong: 10 K 600 m thick, from 0.6 km to 1.2 km a.s.l. JANUARY (30 profiles): Temperature Inversion is weaker: 2 K 600 m thick, from 1.0 km to 1.6 km a.s.l.
Kislovodsk, October 4-9, 2010 12


Verification of the Assumptions using Radiosonde data from Antofagasta
A Summer Case The real profile has plenty more structure (larger fluctuations) The low-frequency decay is still exponential, and the scale height plays an important role. Error% (H=1.55 km) ~ -4% Error% (H=1.74 km) ~ +20% Uncertainty in the scale height leads to over determination and underdetermination of the PWV

Kislovodsk, October 4-9, 2010

13


Model for the vertical profile of water vapor density and PWV
wv p0 wv pS RH 100
T C 18 .678 234.5 T C 257.14 T C

wpv S 611.21 e wv p0 v 0 RV T K

v z v 0 e
PWV z 0



( z h0 ) H

z z

z z

z
v
0

MAX

dz

PWV z 0

z z

z z



MAX

v 0 e



( z z0 ) H

dz

0

PWV z 0 v 0 H e 1

z 0 z

MAX



H



Kislovodsk, October 4-9, 2010


14


The Water Vapor Scale Height (H)

V

0

PWV





Tolar and Armazones: H=1.55 km (median of summer) H=1.74 km (median other seasons) Radiosondes launched from Antofagasta Tolonchar: H=1.13 km Radiosondes launched from Chajnantor
Giovanelli et al. (2001), PASP 113, 803

Small

H

Large
Kislovodsk, October 4-9, 2010 15


The TMT weather stations & data analysis procedure

T, RH V
0
PWV z0 v 0 H e 1
z
0

z

MAX



H



The PWV data series (with 2-minutes time resolution) was averaged in periods of 24 hours before used To compute the global statistics.
Kislovodsk, October 4-9, 2010 16


Results: Overall & Seasonal
Site Alt. (m) PWV (mm) GOES PWV (mm) Surf. D. DPWV DPWV (mm) (%) PWV Night Time (mm)

Otarola et al. (2010), PASP, 122, 470-484

Tolar
Armazones Tolonchar

2290
3064 4480

4.02
2.87 1.70

4.73
3.19 1.83

+0.71
+0.32 +0.13

18%
11% 8%

4.2
2.9 1.7

The global median values obtained from surface weather data agree within 20% with the results from the preliminary studies of Erasmus & van Staden
The difference decreases with altitude. This might be to the fact that the Satellite studies are more sensitive to the 300 hPa ­ 600 hPa region 300 hPa ~ 10 km a.s.l. and 600 hPa ~ 4 km a.s.l.
Kislovodsk, October 4-9, 2010 17


Comparison of PWV from Surface-data Radiosonde soundings and IRMA Radiometers
Florian Kerber Results showed No Bias (Site-2010) Reason: He used In-situ & night time radiosondes
Less of an offset between IRMA and PWV from surface data But more scattering Offset with radiosondes can be explained by a combination of factors: 1)There is a known dry bias in the radiosonde humidity sensors 2)Radiosondes are launched from about 100 km away from Armazones 3)Collocated data is very limited 4)Noise level in IRMA receivers ~ 0.25 ­ 0.5 mm Obvious offset between IRMA and Radiosondes (AFTA)
Kislovodsk, October 4-9, 2010 18

Bias of about 1mm


Radiosonde Humidity Bias
Comparison shows: Dry bias in RS-92 humidity sensor (9 % at surface level & 50% at 15 km altitude) -Originates in a solar radiation illuminating the relative humidity sensor A solution: protective caps, but they limit the sensor ventilation A temperature-dependent calibration error (bias-increasing with decreasing temperature) -The response time of the sensor gets slower with lower temperature. This introduces a lag effect. ~20% dry-bias In Atacama Desert RH ~20% absolute Then bias is 4% of RH
19

DRH, % Vaisala RS-92 & Cryogenic Frost-point Hygrometer (CFH) From VÆmel et al. (2007)

1-mm PWV bias

Kislovodsk, October 4-9, 2010


Radiosonde Humidity Bias

0.5 mm dry bias

Blue: Original sounding data Red : After correcting dry-bias (Vome

l et al (2007), J. Atmos. & Ocean. Tech., 24, 953-963)

Kislovodsk, October 4-9, 2010

20


Variability of atmospheric absolute humidity with distance
Use of aircraft data to study the variability atmospheric absolute humidity as a function of spatial scales OtÀrola et al. (ongoing research Using NSF/NCAR HIAPER aircraft And Lockeed ELECTRA Aircraft data) Distance, km
Kislovodsk, October 4-9, 2010

~20%



q

q

~5%

20% of 2.3mm is 0.46 mm
21


The importance of the water vapor scale height (H)
An error/uncertainty in H translates 1:1 into a Fractional error in PWV
PWV z0 v 0 H e 1 dPWV dH ~ PWV H
z
0

z MAX H







Kislovodsk, October 4-9, 2010

22


Conclusions
Under the particular conditions of humidity profiles in the Atacama desert. TMT was able to use surface measurements of temperature and relative humidity to infer the PWV in the atmospheric column. The PWV derived from surface data in the period 2004-2007, were used to compute the global PWV statistics The results compare within 20% with the preliminary studies of Erasmus & van Staden. In the process we have learned that radiosonde data has to use with care. Daytime humidity profiles from radiosonde soundings are affected by a dry bias. The relative humidity profiles need to be corrected before attempting to derived PWV from them. Spatial variability of the specific humidity field needs to evaluated when comparing insitu results of PWV with PWV derived from soundings launched far from the location of interest. In computing PWV from surface data (or from mm/IR-wavelength radiometers is very important to take into account the uncertainty in the water vapor scale height

Kislovodsk, October 4-9, 2010

23


Suggestions for monitoring the local water vapor scale height
Install weather stations, at different geographic height, to monitor temperature and relative humidity. The PWV scale height can be estimated from the water vapor densities at the two stations and their known altitude difference. Use of a tethersonde equipped with temperature and relative humidity sensors. There are models with 500m range and more. Height difference between sensors gets affected by the pressure exerted in the tethersonde balloon. Launch a radiosonde from site. This has the advantage to provide also information about the height of the boundary layer. This can be take into account to improve the model that uses surface water vapor density to estimate the integrated water vapor in the atmospheric column.

Kislovodsk, October 4-9, 2010

24


Acknowledgements
The TMT Project gratefully acknowledges the support of the TMT partner institutions. They are the Association of Canadian Universities for Research in Astronomy (ACURA), the California Institute of Technology and the University of California. This work was supported as well by the Gordon and Betty Moore Foundation, the Canada Foundation for Innovation, the Ontario Ministry of Research and Innovation, the National Research Council of Canada, the Natural Sciences and Engineering Research Council of Canada, the British Columbia Knowledge Development Fund, the Association of Universities for Research in Astronomy (AURA) and the U.S. National Science Foundation.
Kislovodsk, October 4-9, 2010 25