Temporal knowledge graphs play a crucial role in Natural Language Processing(NLP). In contrast to well-established static knowledge graphs, temporal knowledge graphs provide temporal information, which can be incorporated into knowledge representations. Establishing a time-aware knowledge graph relevant to practical applications poses both research challenges and engineering value. This paper primarily discusses the key algorithm for constructing a temporal knowledge graph of the Qinghai-Tibet Plateau grasslands–the Quadru-ple Information Extraction Algorithm. This algorithm consists of two main components: a named entity recognition model and a relationship classification extraction model. Firstly, an entity recognition model based on word and word relationship classification is used to extract entities. The model achieves a high accuracy, recall, and F1 score of up to 97% on public datasets and custom datasets. Secondly, with a time-aware graph convolutional neural network relationship classification model, it is possible to further obtain relationships between head and tail entities under temporal constraints, thus achieving the final extraction of quadruple information. Experiments demonstrate that the quadruple algorithm successfully extracts quadruples from scientific literature related to the growth cycle of grassland plants on the Qinghai-Tibet Plateau, facilitating the representation of knowledge in this domain. This provides technical support for subsequent agricultural and pastoral science research and dissemination efforts.