The Research Of Data Classification Aggregation Of RTM Based On NetCDF

Land surface runoff is the most important functions in ecological environment.It participates in the whole process of biophysics such as atmosphere,ocean,vegetation and soil.Changes in land surface runoff will have a direct impact on climate,ocean,vegetation,and soil changes,and even affect the global ecosystem environment.Therefore,grasping changes in surface runoff can predict changes in the ecological environment and play an important role in responding to and remedying ecological disasters.However,the current surface runoff model still has some problems in the simulation of surface runoff.CESM is a global climate-ocean-land-glacier coupled climate model,in which the RTM module is an effective way to study surface runoff.Due to the different types of data acquisition equipment,the data storage format cannot be unified,and the surface runoff data has features such as high dimensionality and high computational difficulty,and it is difficult to be coupled to the RTM module.For this reason,this paper proposes an RTM data classification aggregation method based on NetCDF.The focus of this method is: 1)How to classify and aggregate high-dimensional surface runoff data;2)How to perform NetCDF conversion of multi-source high-dimensional surface runoff data to meet the requirements of RTM for input data.The study of RTM data classification is of great significance for analyzing the temporal spatial distribution and evolution rule of surface hydrology.In order to further use CESM to study the interaction mechanism of surface runoff and human activities,understand the laws of climate and environment evolution in the past,and lay the foundation for predicting the change of the future global ecological environment.The main contents of the paper include:1.a high-dimensional RTM data classification aggregation method based on neural network is proposed.This method is based on neural network and Autoencoder algorithm to classify and aggregate high-dimensional surface runoff data.2.an RTM multi-source data classification aggregation model based on NetCDF is proposed,and the pipeline data processing model is designed from the perspective of pipeline communication.The model uses an abstract factory to mask device differences and pass the device's incoming data information by designing a message pump in the pipeline.This kind of information is transmitted to the data signal processor.Then,in the data signal processor,neural network is used to perform data classification on the input data of the device to extract surface runoff data.Finally,based on the Net CDF library file,an interface function for reading and writing Net CDF format files in the.Net environment is proposed to create a data file format that meets the RTM input data requirements.In this paper,the FM evaluation index is used to analyze the results of the improved Autoencoder algorithm model based on the K-means algorithm.The results show that the clustering effect of the high dimensional data is obviously better than the direct clustering;the proposed multi source data aggregation model processor can integrate the input data in real time,and the system is stable.State;the interface of reading and writing NetCDF format data in the.Net environment can well access the scientific data data sets such as surface runoff.