Abstract: Overgrazing is one of the main reasons of grassland degradation. How to monitor the temporal and spatial distribution of feed intake rapidly and effectively is very important for formulating scientific and rational grazing plan and protecting the grassland. Due to Xinjiang??s remote geographic location and vast area, traditional ways of collecting data about feed intake such as recording the data manually is not efficient. With the rapid development of global navigation satellite system, the grazing trajectory data of heads can be collected conveniently. Grazing trajectory records the information of feeding process such as time, position and velocity. In this paper, a temporal and spatial distribution model of feed intake based on trajectory data of grazing is proposed. The establishment of the proposed model consists of 2 steps: constructing the distribution of feed intake with the minimum temporal and spatial granularity, which is used to calculate the feeding intake distribution of the herds in one day, and generating the distribution of feed intake with the multiple temporal and spatial granularities, which is used to describe the feeding intake distribution of the herds in a bigger temporal and spatial granularity. To obtain the distribution of feed intake with the minimum temporal and spatial granularity, the grazing region is computed by building buffer zone using grazing trajectory data. Then the theoretical feeding intake of herds is allocated upon the grazing region evenly to acquire the temporal and spatial distribution model of feed intake. To generate the distribution of feed intake with the multiple temporal and spatial granularities, the target grassland is divided into grids and the distribution of every feed intake with the minimum temporal and spatial granularity is mapped into the grids, and then the multiple temporal and spatial granularities are computed by overlapping the feeding intake grids. The model is subsequently utilized to compute the feed intake distribution. We collected grazing trajectory data of sheep in the studied grassland from July to October in 2015. Taking these grazing trajectory data as input, the model gives the corresponding feed intake distribution. The model result shows that: 1) Feed intake is relatively higher in those places where trajectory points are intensive; 2) Feed intake distribution varies with the different period, which is highly consistent with rotational grazing cycle; 3) There is a significant negative correlation between feed intake and slope of terrain. To verify the accuracy of the proposed model, we randomly chose 59 sample areas of 1 m2 in grazing region and computed the forage surplus of each sample area. We analyzed the correlation between feed intake and forage surplus of each sample area. The result shows the modeled feed intake is significantly negatively correlated with forage surplus, with a correlation coefficient of -0.704, and the accuracy of the model is 86.2%. Thus, the model is useful for rapid acquisition of feed intake??s distribution information in vast area, and it also provides an efficient method to monitor feed intake distribution in grazing grassland.