Tanny, J., H.I.T. - Holon Institute of Technology, P.O.B. 305, Holon 58102, Israel
Cohen, S., Institute of Soil, Water and Environmental Sciences, Agricultural Research Organization, Volcani Center, P.O.B. 6, Bet Dagan 50250, Israel
Grava, A., Institute of Soil, Water and Environmental Sciences, Agricultural Research Organization, Volcani Center, P.O.B. 6, Bet Dagan 50250, Israel
Haijun, L., China Agricultural University (East Campus), P.O. Box 151 Qinghua Donglu 17#, Beijing 100083, China

This paper presents measurements and analysis of airflow patterns, turbulence characteristics and ventilation in a commercial flat-roof screenhouse in which a banana crop was grown. The screenhouse was a rectangle, 352 x 228 x 6 m high, covered with transparent, 15% woven shading screen. An eddy covariance (EC) system was deployed within the screenhouse, 186 m south and 128 m east of its northwest corner to measure evapotranspiration. A three-dimensional sonic anemometer was deployed within the screenhouse near the EC system. Canopy height during the measurement period averaged 4.2 m and the anemometer was installed 5 m above ground. Internal temperature and humidity were measured near the anemometer by three aspirated psychrometers, shielded from direct solar radiation, at heights of 1.5, 3 and 5 m above ground level. External meteorological conditions were measured by a standard meteorological station located outside the screenhouse, about 150 m to the east. A good correlation was found between inside air velocity (uint) and outside wind speed (uext). As expected, the inside air velocity was significantly lower than outside wind speed, the relation being uint=0.27uext, - 0.11 (R2 = 0.89). Most of the time, airflow directions inside and outside were similar but during a few events, for southern external wind the airflow inside was northern. Friction velocity calculated using the measured wind components inside the screenhouse, u*meas, is compared with the one modelled through the logarithmic wind profile, under neutral stability, u*mod. The relation between the two was u*mod=1.02xu*meas+ 0.11 (R2 = 0.77), suggesting that the logarithmic wind profile model is approximately valid. Average turbulence intensity is 0.49 with a standard deviation of 0.12 suggesting that on average, Taylor's hypothesis is marginally satisfied. Air exchange rate, Xs, estimated through the vapour balance technique, increased with external wind speed, as expected, and followed the relation: Xs = 2.39uext + 17.82 (R2 = 0.36).