The Study on Evapotranspiration Model of Intelligent Irrigation Management for Horticultural Crop in Protected Culture

• Main Authors: Ling-Hsi Chen, 陳令錫
• Other Authors: 陳加忠
• Format: Others
• Language: zh-TW
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/de3k6n

博士 === 國立中興大學 === 生物產業機電工程學系所 === 106 === The irrigation technology with evapotranspiration (ET), base on the water balance, is the global tendency for water-saving. This study adapted the lysimeter theory and utilized the weighing balance, data logger and weather sensors for monitoring of seasonal horticulture. The weighing raw data were constituted with two components: evapotranspiration and irrigation operation. The irrigation component was replaced by nearby average values with data smoothing technique to become the target evapotranspiration samples. The target evapotranspiration samples were divided into training data sets and testing data sets. The regression of R2 for estimating dependent variable ET of five independent variables (Rn, VPD, T, RH, Ws) or two independent variables (Rn, VPD) or single independent variables either Rn or VPD are all R2 above 80%. However, the R2 value of the single independent variables either T or RH was lower than 70%, that indicated the estimating ET performance was unstable. The wind speed inside greenhouse was low, so that its effect of estimating ET was small. Using a single VPD parameter to estimat ET has good linear correlation, the R2 was 88.4%. The effect of different regions are not significant. However, the single Rn to estimat ET was affected with different regions. The reason could be explained that the weak Rn for temper region and strong Rn for subtropical weather of Taiwan. The plants’ ET data are the basic information for rationally irrigation decision strategy. The evapotranspiration of single plant of eustoma, tomato and cucumber were conducted for several years’ experiments. ET of tomato and cucumber plant are increase with accumulation of solar radiation intensity and VPD under light saturation point on sunny days. ET will decrease with the change of cloud thickness. Mean square error (MSE) value was used to estimate the performance of artificial intelligent (AI). In training data set, the RMSE of AI is 0.597(MSE=0.356). The value was bigger than 0.380 of the RMSE of five independent variables with linear regression. In testing data set, the RMSE of AI was 0.530(MSE=0.281). The value was bigger than 0.281 of the five independent variables with linear regression. It shows AI has excellent eatimation ET performance. According to the experimental results, development of the Light Accumulation Irrigation Trigger Device (LAITD) was suitable for Taiwan farming. The features of LAITD are low installation and maintenance cost, approach to the ET technology, sufficient irrigation on sunny day, automatic reduced irrigation on raining day, and intensively irrigation around midday when high light shining. The LAITD enhance to a timely, appropriate and rational irrigation technique.