A Spatio-temporal Interpolation Method Based On The Stationarity Of Time Series

In order to get the continuous surface of monitoring data, spatial interpolation is one of major spatial analysis method to deal with the discrete spatial monitoring data. In recent years, with the development and wide application of GIS, RS, GPS and Sensor Network, large number of discrete spatio-temporal data is accumulated. Regarding the discrete spatio-temporal data, many researchers have designed a large number of analysis method from either spatial or time respective. However, few researchers focused on the spatio-temporal interpolation method which was based on the integration of space and time. Firstly, the monitoring data has sophisticated spatial relationship and time relationship, however traditional method didn’t take the spatio-temporal relationship into consideration and just try to solve the solution from either spatial dimension or time dimension. Secondly, during the long-time operation of monitors, all kinds of mechanical and anthropogenic problem easily caused the loss of monitoring data. The traditional methods just utilize time series method or experience to fix the lost data regardless of the spatial relationship among the Sensor Network. From what have been mentioned above, we can conclude that, the traditional spatial interpolation method can obtain the continuous surface and the time series method can solve the data loss problem in some extent, however the methods mentioned above separated the time dimension and space dimension into two parts and hence caused the information missing and affected the future researches potentially. This paper studied the spatio-temporal interpolation method based on shape function method, and pointed out the great drawbacks of extensive method. By studying the stationary process of time series of Econometrics and put forward one spatio-temporal interpolation method based on stationary process. Designed the experiment using the PM2.5data in Jiangsu Province and compared the method based on stationary process with spatial interpolation and time series analysis. In the last part, this paper develop a prototype system. The major research points are listed below.1. Summarized fundamental theory of the shape function, analyzing construction algorithms and characteristics of the2-D and3-D shape function.Studied and compared the two kinds of spatio-temporal interpolations, pointed out the great drawbacks of extensive method. 2. Studied the time series analysis method mentioned in Econometrics, and focused on the stationary test of time series analysis. Studied the stationary process of non-stationary time series data. Using the real data, examined the validity of stationary process and concluded that the PM2.5data is not a stationary time series.3. Put forward a spatio-temporal interpolation method based on stationary process and designed the algorithm process. Using the PM2.5data in Jiangsu province, tested the correctness of the method. Compared the spatio-temporal interpolation based on stationarity with the spatial interpolation method and time series analysis.4. Designed and develop a new prototype system using the spatio-temporal integrated method studied in this paper. Tested the availability of this system using some experimental data.