Study On The Characteristics Of Energy Balance And Evapotranspiration Models Of Rice Paddy In The Low Hilly Region Of Red Soil

It has very important and scientific significance and practical value to study the evaporation and energy balance in different growth periods of paddy system of red soil area in low hilly, due to the planning water layout and improving water efficiency. Based on the observational data, this thesis analyzed the energy balance’s variation and timescale effect of paddy system, discussed the influence of weather conditions in different timescales on common models’estimating evaporation and distribution, established a neural networks model of evaporation and distribution to provide reference to water management of rice production in red soil area of low hill. The following were the main results:(1) Each component of paddy energy balance usually appeared one peak, and the maximum appeared at noon. The diurnal variation curves appeared apparent feature because each component had high sensitivity to different weather conditions. The latent heat and sensible heat flux appeared apparent difference in different growth stages, and the latent heat flux and net radiation had simultaneous changes:jointing period (in August) had the maximum, and the milk stage had less value. Sensible heat flux was higher in jointing period (in late July and early August) and after milk stage(in late September), lower in other period.(2) On the hour-scale and day-scale, the impact factor of the paddy’s latent heat flux (net solar radiation Rn, soil heat flux G.relative humidity RH, temperature T, wind velocity WS, actual vapor pressure Ea, soil temperature Tc, soil moisture SW) existed apparent time scale effect:the dominant factory of hour-scale were Rn、G、RH、T, while dominant factories of day-scale were Rn and G. The factors had extremely significant correlation with latent heat flux on day-scale and hour-scale except that WS and SW had little correlation with latent heat flux. On month-scale, G had significant correlation with latent heat while Rn、Tc、T、Ea had extremely significant correlation.(3) The main effect factors of paddy’s sensible heat flux had large differences on hour-scale and day-scale. The main effect factors on hour-scale were Rn and G while they had large changes in different period on day-scale, which showed that each factory had different contribution rate to sensible heat in different periods, single factory had little contribution rate to sensible heat. On hour-scale, each factory had extremely significant correlation with sensible heat flux; on day-scale, only Ea、RH and Rn had extremely significant correlation with sensible heat flux; on month-scale, Ea and RH had extremely significant correlation with sensible heat flux.(4)The fitting efficiency of each model’s estimating evaporation and distribution had large differences with actual measured values on different timescale. On day-scale, the fitting efficiency of FAO-PM and Priestley-Taylor was the highest, FAO-17-PM was the second, Hargreaves was the lowest. On hour-scale, the fitting efficiency of Priestley-Taylor was the highest, which was the most suitable one for evaporation and distribution real-time forecasting in this area; FAO-PM and FAO-17-PM was second, which overestimated the evaporation and distribution while Hargreaves underestimated. Building a MLP neural network model between actual measured values and the five meteorological factors on day-scale and hour-scale could get good simulation results.