Developments Of Spatio-temporal Interpolation Methods For Meteorological Elements

Spatial interpolation technique for meteorological elements is often used to obtain the meteorological data of every position in the scope which is covered by the meteorological stations. It is need to execute spatial interpolation in the case that the meteorological stations are of scarcity and uneven distribution. However, it is difficult to obtain the satisfying results when the existing spatial interpolation methods are directly utilized to interpolate meteorological elements. For the purpose of obtaining higher interpolation accuracy, this paper firstly reviews the representative spatial interpolation methods, and makes a detailed comparison among them, and secondly, the spatial interpolation method for meteorological elements is presented based on spatial clustering and the spatio-temporal interpolation method for interpolating the irregular data set and repairing the missing data of the meteorological stations is developed. Lastly, a prototype tool for interpolation is designed and implemented to show the rationales and application of the proposed methods in this paper. Main works can be summarized as follows:1. The background and purpose of this paper are explained, the research progress on spatial interpolation and spatio-temporal interpolation for meteorological elements is reviewed, the problem of the existing spatio-temporal interpolation method is pointed out, and moreover, the research content and structure arrangement are introduced.2. The interpolation algorithms of Thiessen, Inverse Distance Weighting (IDW), Gradient Plus Inverse Distance Weighting (GIDW), Spline, Trend, Area-based, Ordinary Kriging (OK) are realized and used to interpolate for annual average temperature and precipitation of China. Cross-validation result shows that:for the annual average temperature, GIDW method is better than the other six kinds of interpolation methods, and for the annual average precipitation, IDW method is the most effective.3. Spatial interpolation method for meteorological elements based on spatial clustering is proposed. K-means spatial clustering and three-step spatial clustering are separately used to cluster for annual average temperature and precipitation of China. The results of interpolating after clustering and directly interpolating are compared. Interpolation after clustering which has better accuracy is found. Moreover, the interpolation result of three-step spatial clustering is better than that of K-Means spatial clustering. Thus, spatial clustering can be used as an effective preprocessing step before spatial interpolation.4. The shortage of spatio-temporal interpolation proposed by Li is pointed out and has been improved. An improved reduction method and a mixed spatio-temporal interpolation method are proposed. The improved reduction method and the original method are applied to interpolate irregular data set and the interpolation results of two methods are compared. The mixed spatio-temporal interpolation method is used to repair the missing data of the meteorological stations. The result shows that mixed spatio-temporal interpolation is better than a single time interpolation or spatial interpolation for repair the missing data.5. Based on the spatio-temporal interpolation method for meteorological elements proposed by this paper, the module of drawing rain isoline in the Jinan urban flood warning system is developed. The feasibility and correctness of the spatio-temporal interpolation method are validated by practice.Finally, after concluding the main achievement in this paper, some issues for further work are presented.