PREDICTION OF HARVEST IN CAULIFLOWER BASED ON METEOROLOGICAL OBSERVATIONS
A model for planning planting/harvest sequences could be the multi linear regression between accumulated day-degree > 5°C from planting to harvest and the effects of cultivar, planting day number and the square of planting day number:
y = C1-6 - 11.24(x) + 0.0423(X2)
where y is the predicted accumulated day-degree > 5°C, C1–6 are estimates for the six cultivars and x is the planting day number in the year (1 January = day no 1). This model accounted for 84 % of the variation in observed accumulated day-degree > 5°C.
The duration of the period from transplanting to curd initiation showed more variation than the duration of the period from curd initiation to harvest.
A model for predicting harvest time in the season and after the curd is initiated could be the linear regression between the natural logarithm of the curd diameter and the accumulated day-degree > 5°C from curd initiation:
ln CD = -4.07 + 0.0114(acc. dd.)
where CD is the curd diameter and acc. dd. are the accumulated day-degrees > 5°C. This model accounted for 89 % of the variation in observed in curd diameter. The model can be used to predict the time when a certain curd diameter is reached. The prediction is based on plant samples taken 4 to 5 weeks before main harvest.
The models must be combined with data for actual weather, long term average weather and weather forecasts for the area.