Abstract: By the correlation analysis, we determined biological indices such as “bud initiation”, “high temperature acceleration”, “low temperature retardation”, and the climate trend background of spring warming. Taking Shandong Province as an example, we established forecast models with biological implications by stepwise regression. The results showed that the first day when daily average temperature steadily passed 5 ℃ is more suitable as the temperature index of “flower bud sprouting” by comprehensive considerations of the correlation coefficient and the number of days required for apple flowering. The critical temperature of “high temperature acceleration” and “low temperature hysteresis” was “daily highest temperature ≥20 ℃” and “daily lowest temperature ≤1 ℃”respectively as found by correlation analysis. Dynamic prediction models were established with different starting point and ending point of accumulated temperature. The prediction effect of the model with accumulated temperature starting point of March 1 and the daily average temperature steadily passed through 5 ℃ is fairly good, but without significant advantage in terms of real prediction effectiveness. With the help of numerical prediction products, the restriction of real data can be overcome, which would improve the initiative and flexibility of forecast services.

Key words: apple, flowering forecast, biological index, numerical prediction.