A Precipitation Forecast Displacement Correction Algorithm Based on Object Diagnosis

Abstract

The invention discloses a precipitation forecast displacement correction algorithm based on object diagnosis. The algorithm overcomes the deficiencies of the MODE method and the CRA method, and designs a reasonable object displacement deviation calculation method, and combines the ideas of the object diagnosis and the observation similarity method. Combining, avoiding the drawbacks of point-to-point diagnostic methods, and directly improving the model's precipitation forecast performance from the perspective of object displacement bias correction. Data experiments and statistical cross-checks show that after the displacement correction of the present invention, the forecasting performance of different forecasting timeliness has been improved, the critical success index and detection rate have been increased, and the false alarm rate has been reduced.

Description

technical field [0001] The invention relates to the technical field of numerical forecast statistical post-processing in weather forecast business, in particular to a numerical model precipitation forecast correction technology. Background technique [0002] In the operational practice of weather forecasting, forecasters often find that the heavy precipitation rainbands predicted by numerical models have intensity deviations and north-south shifts compared with the actual observations. At present, most of the existing methods correct the intensity deviation of the model precipitation forecast. Although the forecast performance of the model can be improved to a certain extent, the intensity of heavy precipitation is often accompanied by too much false alarm rate (False Alarm Ratio, FAR). However, the correction methods for the deviation of rainband displacement are rare. The observational similarity method can indirectly correct the displacement deviation of the precipitatio...

AI Technical Summary

Problems solved by technology

However, this method is based on traditional point-to-point error indicators (such as root mean square error, correlation coefficient, etc.) There is a "double penalty" problem in the observation grid

[0003] The Method for Object-based Diagnostic Evaluation (MODE) is a new type of spatial inspection technology, which can extract forecast and observed precipitation objects and calculate displacement deviations, avoiding the disadvantages of point-to-point diagnostic methods, but no scholars have yet Combining the method with the observation similarity method for displacement correction of precipitation forecasting objects

In addition, in the calculation of the displacement deviation of the object, the MODE method only uses the center position of the object to calculate the displacement deviation, and the continuous rain area method (Contiguous Rain Area, CRA) calculates the displacement deviation by moving the object to achieve the maximum overlap. In practice, two methods have been found None of them can adapt well to the precipitation objects with different shapes, and often produce calculation results that deviate greatly from the subjective judgment of the forecaster

Images