Numerical Analysis Of Statistical Power In Precipitation Enhancement Of Non-randomized Experiment And Statistical Method Improvement

This paper probe into the basic issues of non-randomized artificial rainfall statistical evaluation and and some factors influencing assessment effectiveness are revealed from the results of numerical analysis and "false results" statistical analysis[0], in succession, some evaluation of precipitation enhancement has been done based on this selection method. Finally, three kinds of improved programs are put forward.There are still many difficulties and problems in the selection of the target area, control area and statistical variables. Considering practical and scientific factors[0], scientifically accepted simple methods[0] are recommended for estimating in non-randomized operations. The results of numerical analysis show that the statistical power has something to do with the historical samples, operating samples, the effect of precipitation enhancement options and different statistical evaluation methods. Double Ratio Method and Historical Regression Method are selected for evaluation of the precipitation enhancement effectiveness in Beijing from 2002 to 2007 here and all the results of these three methods illuminate that the relative effectiveness is about 10% at the significance of 0.05. Numerical analysis shows that the testing effectiveness is low and assessment accuracy is inadequate by using existing statistical evaluation method for effectiveness estimation of the precipitation enhancement which has few operating samples. It is prescribed to estimate the effectiveness of single operational cloud seeding experiment or operational cloud seeding experiments of a whole single year and the Historical Regression Method are recommended for effect evaluation of these programs. The Historical Regression Method is used in this study for estimation of precipitation enhancement effectiveness in Beijing 2006 and the results show that it has a lot of fluctuation of the single experiment. A series of blank experiment from 1980 to 2000 is used to analyze the stability of the multiple event test method from historical regression method. The result suggests that as the number of operational days increase, the statistical result tends to be more stable and the number of operational days can not meet stability requirements for one year experiment.In order to improve the Historical Regression Method, three kinds of improved programs are put forward and then some examples are given to verify and analyze their shortcomings and applicability. False results are removed from the estimated Precipitation enhancement effectiveness in Method 1 that makes the results more reasonable in the concept. A second assessment from the comparison of precipitation enhancement cumulative effect get from Historical Regression Method and Double Ratio Method is done in Method 2, but the results of an instance show that the method itself requires more stringent conditions, so this issue till requires further study. Method 3 introduces the Regression Ratio Method which is never used at home.