Construction and Validation of a Model for Estimating Surface Water Evaporation in Greenhouse Based on Principal Component Analysis

Abstract

【Objective】 Surface water evaporation (Ep) in greenhouse is often used as a reference in irrigation management. The purpose of this paper is to present a model to estimate it. 【Method】 The model was derived based on meteorological data and pan-evaporation measured from March to July in 2020 and 2022. The relationship between Ep and meteorological data was analyzed using the principal component method, from which a multiple linear regression model was developed to estimate Ep. 【Result】 Ep increased as time elapsed, with its average increasing from the range of 1.84~1.94 mm in March to the range of 3.77~5.15 mm in June in both 2020 and 2022. Ep was influenced by radiation and relative humidity the most. Photosynthetically active radiation had the highest correlation with Ep, with their correlation coefficient being 0.852 (P<0.01), followed by solar radiation and relative humidity with their associated correlation coefficient being 0.811 and -0.770, respectively (P<0.01). The first principal component of solar radiation, photosynthetically active radiation, and relative humidity has a great effect on Ep, with the eigenvalue being 4.44. The solar radiation affected Ep significantly, with the highest score coefficient (0.328), followed by relative humidity and photosynthetically active radiation, whose score coefficients were 0.311 and -0.321, respectively. Principal component analysis and verification showed that the estimated Ep using the proposed model agreed well with measured data, with P<0.01, R2=0.908, MBE=0.10, RMSE=0.48 mm/d, and the consistency index=0.94. 【Conclusion】 Solar radiation, photosynthetically active radiation, and relative humidity are the main factors influencing Ep in the greenhouse. The multiple linear regression model derived from the principal component analysis is accurate and can provide real-time estimates of surface water evaporation in the greenhouse.

Keywords

• greenhouse
• pan evaporation
• meteorological factors
• multiple collinearity
• principal component analysis