A comparison of seeing values measured simultaneously by ESO DIMM and GSM shows that most of the time both instruments are in excellent mutual agreement. An example of such agreement is given in Fig. 3.

$\begin{figure} \par\includegraphics[scale=0.75]{ds1836f2.eps} \end{figure}$

Figure 2: Summary of the GSM data at Paranal: evolution of a) wavefront outer scale ${\cal {L}}_0$ , b) seeing $\varepsilon$ and c) isoplanatic angle $\theta _0$ during the 19 nights of the mission. For each night the median value of the considered parameter is plotted as a dot and the rectangle indicates the range in which 50% of the values are included. The histograms of these 3 parameters d), e) and f) correspond closely to a log-normal distribution

$\begin{figure} \par\includegraphics{1836f3.eps}\end{figure}$

Figure 3: The seeing measured by GSM (full circles) and ESO DIMM (dotted line) on the night of November 29 is plotted in the upper panel, showing an excellent agreement between both instruments. Simultaneous recordings of the C_n² with temperature sensors at the heights of 3, 7 and 21 m are plotted in the lower panel

$\begin{figure} \par\includegraphics[scale=0.65,clip]{ds1836f4.eps} \end{figure}$

Figure 4: Same as Fig. 3 for the night of December 2. The local turbulence around midnight caused the difference of the seeing measured by GSM and DIMM. It matches the increase of C_n² at the height of 3 m

However, for some periods GSM measured a worse seeing than DIMM because of the lower height of GSM, which makes it more sensitive to the SL turbulence. This situation appeared soon after the beginning of our mission as an enigmatic effect: for several consecutive nights, starting from December 1 and until December 5, GSM registered a much worse seeing than DIMM for a period of about one hour, around local midnight. For the rest of the time, the agreement was good. A most clear example of this phenomenon (which we call here midnight local turbulence, or MLT) is given in Fig. 4. On December 2, starting from 03h00 UT, GSM gave a seeing systematically worse than that given by DIMM. This particular local phenomenon (MLT) is explained in the next subsection.

4.2 Statistical contribution of the surface layer to the global seeing

The micro-thermal data, also plotted in Figs. 3 and 4, confirm the idea that the divergence between GSM and DIMM seeing values is caused by the strong turbulence in the first few meters above the ground. Indeed, the period of disagreement exactly matches the turbulence increase at 3 m. Similar behavior was observed on the nights of December 1-5 corresponding to the period of stable atmospheric conditions, already noted in Sect. 3. Ground wind of 5 m/s was prevailing during this period, with a direction slowly changing from East to North during the night, and with moderate velocity variations.

$\begin{figure} \par\includegraphics{1836f5.eps}\end{figure}$

Figure 5: Histograms of the C_n² values computed from the temperature sensors at 3 altitudes, a) 3 m , b) 7 m and c) 21 m

A simple statistical processing of the temperature micro-fluctuation data permits to give a global behavior of the C_n² measured at the three heights during 9 nights. For each height, about 45000 values of the structure constant have been recorded and processed. The results can be seen in Fig. 5 which shows clearly that statistically the C_n² mean value near the ground (at 3 m) is greater than the one at 21 m. The ratio of the corresponding values of C_n² is approximately 16. Conversely, the dispersion grows with height.

To estimate the relative contribution of the SL to the total seeing degradation, a turbulent energy ratio (TER) can be defined as the ratio of the turbulent energies in the (3 m - 21 m) and (3 m - infinity) layers:

$\displaystyle TER = {\int\limits_{3~{\rm m}}^{21~{\rm m}}C_n^2{\rm d}h \over \int\limits_{3~{\rm m}}^{+\infty}C_n^2{\rm d}h}\cdot$

(1)

For all the nights during which temperature data where available, this ratio has been computed each four minutes in order to match the sampling rate of the GSM which provides the turbulent energy from 3 m to infinity. Numerator of the above expression is deduced from the integration of rough C_n² profile given by the temperature sensors between 3 m and 21 m. Night by night mean values of the TER can be found in Fig. 6. From the table one can see that nightly values are in a ratio of 1 to 2.4.

In comparison with the seeing produced by turbulence layers above 21 m, a degradation would occur if the turbulent effects in the SL were added. The value of the seeing degradation (%) is given by (1-[1-TER]^0.6).

Histogram of the seeing degradation values and its cumulative distribution are plotted in Fig. 6. The median value is 7.2%. Fifty percent of the values centered on the median are between 4.0% and 11.0%.

It is thus evident that the contribution of the SL to the global seeing cannot always be considered as negligible and, although GSM is more affected by SL turbulence, DIMM is not always entirely free of it either. Ground turbulence, a part of which appears above DIMM but below VLT, may partially explain the difference sometimes detected between UT telescopes and DIMM seeings.

$\begin{figure}{$\vcenter{\begin{tabular}{\vert c\vert c\vert} \hline Date & TE... ...vcenter{\includegraphics[width=5cm, height=5cm]{ds1836f6.eps} } $ } \end{figure}$

Figure 6: Values of the turbulent energy ratio for the 9 nights for which temperature data are available. Histogram of the seeing degradation values and corresponding cumulative distribution

4 Contribution of the surface layer to the overall seeing

4.1 Surface layer detection from GSM and DIMM measurements

4.2 Statistical contribution of the surface layer to the global seeing