辽宁人工增雨试验区功效研究  

Study on Statistical Power of Artificial Precipitation Experimental Area in Liaoning

作　　者：房彬[1,2,3] 班显秀[3] 肖辉[1] 

机构地区：[1]中国科学院大气物理研究所,北京100029 [2]辽宁省人工影响天气办公室,辽宁沈阳110016 [3]中国科学院研究生院,北京100049

出　　处：《高原气象》2009年第3期586-593,共8页Plateau Meteorology

基　　金：国家自然科学基金项目(40875080,40333033);国家科技支撑计划项目(2006BAC12B00-1);辽宁省科技厅“十一五”重点攻关项目“人工增雨实用催化系统集成关键技术研究”(2006210001)共同资助

摘　　要：采用复随机化试验法对辽宁人工增雨试验区进行功效分析,并通过对基于聚类的浮动对比区历史回归统计检验评估方案(CA-FCM)等几种评估方案的比较,确定了最适宜辽宁人工增雨试验区的试验方案和样本容量。结果表明:在辽宁人工增雨试验区以80%检出概率(0.05显著度)检验出10%增雨效果,区域控制试验和区域回归试验需要400个样本,区域交叉试验、多区双比交叉回归试验、CA-FCM需要250个样本。其中CA-FCM评估方案在相同的样本容量和催化效果情况下,功效值最大,评估效果最佳。This paper analyses the statistical power of Liaoning artificial precipitation experimental area by re-randomization method, and finds most suitable evaluation method and sample size through comparison between Cluster-Analysis-based Floating Control historical regression evaluation method （CA-FCM） and the other evaluation methods. The result indicates that the target control randomized evaluation method and historical regression randomized evaluation method need 400 experimental samples in order to detect 10% cloud seeding effect with 80% statistical power at 0.05 significant level, however cross-over randomized evaluation method, multi-area double radio regression evaluation method and CA-FCM only need 250 experimental samples. To the same cloud seeding effects and sample size, CA-FCM has greatest statistical power and best evaluation effect.

关 键 词：辽宁 人工增雨 复随机化试验法 CA-FCM 样本容量 功效 

分 类 号：P481[天文地球—大气科学及气象学]