Abstract:

The agricultural industrial parks in Qingdao are distributed in various terrain areas such as inland, coastal and mountainous areas with large temperature differences. In order to improve the level of refined weather service for modern agriculture, we compared and tested the applicability of temperature prediction products in Qingdao by the methods of the nearest neighbor interpolation, bilinear interpolation and inverse distance weight interpolation, based on the meteorological elements prediction products of 5 km grid in the national region issued by the national meteorological center. The results show that: in inland areas, the bilinear interpolation has the highest accuracy and the minimum mean absolute error and root mean square error; the nearest neighbor interpolation and the inverse distance weighted method are both effective in coastal areas; for mountainous area, the inverse distance weighted interpolation method is more suitable considering the accuracy, mean absolute error and root mean square error; the accuracy of the maximum temperature prediction in inland areas is higher than that in coastal and mountainous areas, and the accuracy is the highest from August to October, while relatively low from March to June; but the accuracy of the minimum temperature prediction is the highest in coastal area, and it is higher in summer than that in winter, with the highest accuracy in July and the lowest accuracy in January. The results can provide a basis for producing refined weather service products with fine resolution and high accuracy.

Key words: Qingdao, terrain, temperature, weather service, nearest neighbor interpolation, bilinear interpolation, inverse distance weighted interpolation