Modelling of Greenhouse Microclimate for Rose Production in Zimbabwe.

Abstract

The main objective of the project was to adapt and validate the Gembloux Dynamic Greenhouse Climate Model for the prediction and control of the microclimate of an operational greenhouse in Zimbabwe. The use of modelling as a microclimate control strategy is aimed at achieving optimal climate control through the provision of a more affordable alternative. Measurements of climate variables were taken in a central position within as well as outside a commercial rose greenhouse at Floraline Pvt. Ltd over the period November 2004 to February 2005. The first stage of the project involved assessing the homogeneity of the greenhouse air by comparing temperature and humidity measurements taken at five positions within the greenhouse structure. It was found that the greenhouse air was not homogeneous and the placement of air temperature and humidity sensors in the greenhouse was important for correct decision making. The model performance was analyzed by comparing the measured climate variables (air temperature and humidity) inside the greenhouse with those from the simulation using the Gembloux Dynamic Greenhouse Climate Model. This was done for two data sets, 3 to 11 November 2004 and 25 November 2004 to 01 December 2004; these were used as the calibration and validation periods respectively. In the study, both the direct and diffuse solar radiation had to be simulated as only the global radiation was measured whilst the stomatal resistance values used for the transpiration sub-model were 1800 s m-1 (maximum) and 200 s m-1(minimum).

Overall, the model underestimated the air temperature by up to 6 °C whilst the humidity was overestimated by up to 60 %. The root mean square errors (RMSE) for the calibration period were in the range 1.0 - 2.0 °C (daytime) and 0.2 - 0.8 °C (nighttime) for the air temperature. The vapour pressure had RMSE in the range 0.6 - 1.0 kPa (daytime) and 0.1 - 0.2 kPa (nighttime). The model deviated more from the actual measurements during the daytime because of increased temperature and vapour pressure differences between the inside and outside of the greenhouse. The differences between the simulated and measured air temperature was attributed to their volumetric heat capacities, arising from those of pressure (1013 hPa and 850 hPa respectively). There was a notable improvement in the nighttime simulation of the air temperature and vapour pressure during the validation period. The differences between the calibration and validation periods indicated errors in the simulation programme that were not picked up during the runs. Based on the air temperature results, the model has shown its potential for use in greenhouse microclimate prediction in Zimbabwe. The high correlation (r2=0.96) between the simulated and measured air temperature shows that the model has the potential for use in decision making. The model can be reliably used to determine the weekly greenhouse air temperatures because of the observed maximum difference of 2 °C between the simulation and actual measurements. However the model cannot be implemented without further calibration of the model parameters to improve the simulation of humidity, which was significantly different from the actual measurements. Humidity is an important variable in