| V. Review Paper: WHAT CAN METEOROLOGICAL NUMERICAL MODELS DO FOR THE ASTRONOMER Elena Masciadri |
The recent developments in High Angular Resolution techniques
and the high performances that new generation telescopes
are planned to attain in the next century have changed, in the last years,
the definition and the methods of the 'site testing' science. A set of astroclimatic parameters are to be estimated to define the quality of an astronomical site and optimize the flexible-scheduling of the scientific programs and instruments placed at the telescope's focus. Two different approaches can be conceived in order to attain this goals: experimental measurements and the numerical simulations. The latter's principal characteristics is the ability to provide 3D spatial characterizations of the astroclimatic parameters and to supply forecasts. This contribution will aim to answer the following questions: (1) which are the astroclimatic parameters that we can characterize with 3D or 2D maps with a numerical model (2) which are the different kinds of models that can supply such maps (3) what is the forecasting reliability that we can hope to attain. Finally, we will discuss the recent progress and future perspectives on the simulation of the optical turbulence. |
| V.1 ON THE PREDICTION OF ASTRONOMICAL SEEING FLUCTUATIONS WITH NEURAL NETWORKS Alex Aussem & Marc Sarazin |
The European Southern Observatory's Astronomical Site Monitor (ASM)
for the Very Large Telescope at Cerro Paranal in Chile aims at
delivering short-term predictions of the Seeing variability - from 10
up to 60 minutes ahead - to allow optimization of the observing
strategy (observing block selection, scheduling of calibration tasks).
Years of seeing monitoring at various astronomical sites have shown
that the seeing is not stationary and motivated the definition of a
variability measure, the FEFSC, i.e., the "Finite Exposure Fractional
Seeing Change" (Racine96, Sarazin97), the impact of which on the
performances of adaptive optics systems has been clearly illustrated
(Rigaut98). Extensive data collected at Paranal have been used to
appraise a forecasting methodology based on neural networks (NN). We
show that NN achieve a success rate up to 70% in predicting the FEFSC
trend. The forecasting tools which have been prototyped, are a
contribution to a future telescope decision support system. |
| V.2 UTILISING WEATHER SATELLITE DATA FOR EVALUATION AND FORECASTING APPLICATIONS AT ASTRONOMICAL SITES Andre Erasmus |
Existing and potential astronomical sites tend to be located in remote areas where
conventional atmospheric (meteorological) observations are absent or scarce. Weather satellites,
particularly geostationary satellites, provide an important data source for astronomical site
characterisation since they observe atmospheric parameters such as cloud cover and water vapour
over large sections of the globe with a spatial and radiometric resolution suitable for monitoring
and/or forecasting applications at astronomical sites. Satellite data archives now comprise 10
or more years of data coverage for many areas, providing an adequate climatological database
for site evaluation and comparison.
Using observations at wavelengths in the atmospheric window (11.5 microns) the amount of cloud in
successive atmospheric layers can be measured. Therefore, cloud occurring below the altitude of a
site can be distinguished from cloud cover above the site and the cloudiness at sites with
different altitudes can be accurately compared. Imaging at 6.75 microns gives a measure of water
vapour in the middle and upper troposphere and, for many high altitude sites (above 3500m), this
closely approximates the vertically integrated water vapour above the site.
This paper provides an overview of weather satellite data coverage for different
areas of the globe and describes the characteristics of these data relevant to monitoring and
forecasting applications at astronomical sites. The methodology for cloud cover and water vapour
measurement is described. To demonstrate the successful application of these data, results from a
current project to identify potential telescope sites in Northern Chile for NOAO-CTIO are
presented in brief. In line with the stated scientific goals the workshop, the use of weather
satellite data to identify potential astronomical sites around the globe is proposed.
|
| V.3 COMPARISON OF CLOUD COVER OBSERVATIONS AND FORECASTS BASED ON SATELLITE IMAGERY TO ATMOSPHERIC ABSORBTION MEASURED AT THE VLT OBSERVATORY Andre Erasmus & Marc Sarazin |
Tools for monitoring and forecasting local conditions (wind,
temperature, pressure, humidity) as well as line of sight
conditions such as atmospheric transparency (precipitable water vapour
and cloud cover and astronomical seeing) have been developed for the
operation of the ESO-VLT Observatory in Northern Chile.
Because the observatory is located well above the inversion layer,
nebulosity is mainly due to high altitude cirrus clouds. To forecast the
occurrence of these clouds a model was developed which uses 3-hourly GOES8
Water Vapour (6.7 micron, 8x8km resolution) channel satellite imagery. In order
to also detect and forecast low and middle level clouds, an upgraded
version of the model employing, additionally, thermal IR (10.7 micron,
4x4km resolution) imagery is under development.
The "current observatory sky" is defined by a circular area of about 32km
in radius centered on the observatory site. For each 3-hour period out to
24 hours, a forecast is made based on the satellite image "area" predicted to pass
over the site. The "areas" are centred on a predicted "most likely path" which is
determined using a motion algorithm that takes into account both the translation
(500mb wind) and the development(total geopotential thickness change of the
700mb-300mb layer) of pressure systems. For each 3-hour forecast period, the pixels
within the image areas are weighted according to their proximity to the most
likely path and hence their likelihood to pass over the site.
Each pixel is classified as clear or cloudy based on either the
computed upper tropospheric humidity (UTH)(6.7 micron channel) or brightness
temperature (10.7 micron channel). In addition, using the UTH values, an intermediate
class of 'transparent cirrus' is defined to account for generally clear sky situations
but where the thickness of cirrus clouds is such that they have a variable impact
on extinction. In such cases, a "transparency index' is computed by the model.
In order to verify cloud observations and forecasts produced by the model and to
"calibrate" the transparency index, the output of the model over one year of operation
is compared to simultaneous photometric measurements made at Paranal with the Seeing Monitor telescope.
The economic value of these forecasts for telescope flexible scheduling
is discussed. |
| V.4 PROGRESS IN THE VALIDATION OF THE ATMOSPHERICAL MODEL MESO-NH ADAPTED TO PROVIDE 3D MAPS OF OPTICAL TURBULENCE Elena Masciadri |
The forecasting and the 3D mapping of the optical turbulence (OT) is fundamental in the most High
Angular Resolution observational techniques. The retrieval
of the volumetric distribution of the Cn2, that is the intensity of the OT, in a
region around the telescope is a necessary requisite to apply the best techniques of
the wavefront correction such as the MCAO (Multiconjugate Adaptive Optics (Beckers88, Tallon92))
and the Adaptative Optics with LGS (Laser Guide Stars (LeLouarn98, Chester90)). Moreover,
the ability of forecasting the optical turbulence is a cornerstone for a real application of the
flexible-scheduling (Masciadri99a, 99b). Some years ago we thought to adapt an atmospherical non-hydrostatic model (Meso-Nh) conceived to provide 3D maps of classical meteorological parameters p, T and V in order to provide 3D maps of Cn2. The idea was to initialize the model with vertical (p, T and V) profiles provided by analysis from European Center for Medium Weather Forecasts and/or radiosounding from Meteorologic Stations. Starting the simulation, when the flow was adapted to the orographic model, the validation of the model would be realized comparing the vertical Cn2 profiles simulated and measured by optical instruments such as the Generalized Scidar and in situ measurements such as balloons. In these last years we realized three feasibility studies over different sites: Cerro Paranal (Chile), Roque de Los Muchachos (Canaries Islands) and San Pedro Martir (Mexico). The principal difficulties to validate the model are In this paper we will sketch the principal results and progress obtained. We will trait problems such as: the quantitative estimation of the OT (Masciadri99a) the global coherence of the integrated parameters characterizing the wavefront perturbations (the seeing, the isoplanatic angle, the wavefront coherence time, the spatial coherence outer scale) simulated by the model (Masciadri99a), the statistical estimation (Masciadri99b), the influence of the radiation from the ground on the OT simulated (Masciadri2000, Bougeault98), the simulation of the temperature near the ground in order to adjust the primary mirror temperature and eliminate the dome seeing (Masciadri2000). Moreover we will show the interesting consequences obtained integrating the simulated Cn2 respect lines of sight different from zenith. We will deal about a new model calibration techniques and we will finish reporting the last modifications added in the model: the isoplanetic angle with a partial correction and the structure constant of the specific humidity Cq2. This last will permit to simulate turbulence perturbations for the millimetric observations. |
| P.V.1 (poster) PRINCIPAL COMPONENT ANALYSIS APPLIED TO SEEING PREDICTION A. Garcia de Gurtubai, A. M. Varela, C. Muñoz-Tuñon & T.J. Mahoney |
A principal component study is presented of mean temporal values
corresponding to seeing, air temperature, relative humidity, barometric
pressure, wind speeed and wind direction parameters obtained at the Roque de los
Muchachos Observatory (La Palma) during 90 nights in 1999. The main goal
of this analysis is perform a least-squares fit in a six-dimensional space
in order to generate a routine to allow us establish reliable seeing
prediction from the meteorological input data.
In this contribution we present the techniques and the preliminary
results. |