[Vertical temperature distribution and its forecast for two tree structures of apple orchard during the blooming period in the Loess Plateau].

Temperature is the most sensitive environment factor for the blooming period of apple. Temperatures at different levels were measured by automatic micro-climatic gradient system in the blooming periods from 2011 to 2014, in two Fuji apple orchards with two different tree ages and structures [small canopy open center shape (SMCOCS) and freedom spindle shape (FSS)], respectively, which were typical in the Loess Plateau. Variations of the temperature gradient in both canopy and tree body were analyzed in sunny, overcast, cloudy, and rainy weather conditions, and a predicting model was established that could predict the temperature of the canopy (TL) according to the temperature observed in nearby meteorological station (TM). The results showed that the vertical distribution of canopy temperature and its difference to the outside of orchard was mainly due to the tree structure, rather than the weather condition. The average temperature and daily minimum temperature increased while the daily maximum temperature and the diurnal temperature range decreased from the bottom to the upper of the canopy. For SMCOCS, the diurnal temperature range reached its peak under the canopy in the clear days, and the diurnal temperature range was less than that for FSS in the middle and upper canopy in cloudy or overcast conditions. The daily variation of temperature difference between inside and outside the orchard behaved as a single peak-valley-peak for FSS but as a single peak for SMCOCS. The minimum temperature outside the orchard was closer to that in the middle of canopy, but higher than that in the bottom of the canopy. For SMCOCS, the minimum temperature in the bottom of its canopy was rather lower than that in the orchard outside, especially in cloudy or overcast day, while in the middle or upper canopy, the minimum temperature difference with the orchard outside was smaller than that for the FSS. The linear model was found to be able to predict the TL with absolute errors below 1 °C, and the best prediction was found for the FSS in rainy days.