Estimating reference evapotranspiration using Penman-Monteith equation integrated with optimized solar radiation models

Abstract

Accurate estimates of reference evapotranspiration (ET0) are of great significance to water resources planning and management, but the actual solar radiation (Rs), as the primary parameter for ET0 estimation, is difficult to obtain directly in most areas. Thus, studying the impacts of locally calibrated empirical solar radiation (Rs) models to improve the accuracy of Penman-Monteith (PM) is significant. Meanwhile, swarm intelligence algorithms have proved their potential in the domains of agriculture and hydrology, but few studies applied them in optimizing Rs models to improve ET0 estimation. This study used the particle swarm optimization (PSO), the gravitational search algorithm (GSA), and the mind evolutionary algorithm (MEA), respectively, to optimize the nine most common empirical Rs models, comprising three sunshine-based models (Angstrom, Ögelman, Bahel), three temperature-based models (Hargreaves, Bristow-Campbell, Hunt), and three combined-based models (Fan, Chen, El-Sebaii), and then integrated them into the PM for ET0 estimation at four climatic zones of China. The results showed that the Fan model obtained the most accurate Rs estimates in china, while the sunshine-based and temperature-based exerted significantly different applicability at different climatic zones. Regarding optimization algorithm, this study found that GSA performed better for the Ögelman model, Fan model, and Chen model when integrating into the PM equation for ET0 estimation, whereas MEA performed better for the Angstrom model, Hunt model, and El-Sebaii model. After optimization, PMGSA–Fan obtained the most accurate estimates of ET0 at four climatic zones, with a regional spatial gradient in estimates accuracy of Rs from north to south of China. In terms of sunshine-based models, the PMMEA–Angstrom performed better in TMZ, TCZ, and MPZ, whereas the PMGSA–Ögelman performed better in SMZ, respectively; in terms of temperature-based models, the PMMEA–Hunt performed better in TMZ and SMZ, whereas the PMLSM–Hargreaves performed better in TCZ and MPZ, respectively. Overall, this study identifies the optimal Rs estimation model and optimization algorithm for ET0 estimation at four climatic zones, which provide regionally and nationally accurate water consumption information without actual measured Rs in China.