关键词: 青海湖北岸, 草地, 牧草产量, 归一化植被指数, 预测模型

Abstract:

The objective is to realize the dynamic prediction of grass yield, grasp the change of grass yield in time, and provide scientific basis for animal husbandry production efficiency and structural adjustment in alpine region. Taking the north bank of Qinghai Lake as the research area, the grass yield prediction models for different grassland types (alpine meadow, alpine steppe and temperate steppe) were established by grass yield data, meteorological data and NDVI data from 2003 to 2020. The effect of fencing and grazing on the grass yield prediction models were analyzed. The results showed that alpine meadow had the best prediction effect among the three different grassland types. The regression effect between grass yield data and meteorological elements was better than that between NDVI and meteorological elements. The regression correlation coefficients between grass yield data and meteorological data in Haiyan were above 0.8, which passed the significant test of 0.001. Five models of Haiyan, four models of Qilian, two models of Yanglong and two models of Gangcha were screened with high prediction accuracy, which had good estimation ability and could meet the needs of grass yield prediction application in or outside the fence in these four areas.

Key words: northern shore of the Qinghai Lake, grassland, grass yield, NDVI (normalized difference vegetation index), prediction model