Construction of Precipitation Forecasting Model in Asia-Pacific Region using Atmospheric-Oceanic Indices

首先，采用简单但稳健的方法揭示了南方涛动指数（SOI）对分布在韩国和日本福冈的5个站点的月降水的定量和统计显著影响。将逐月降水数据转化为非超越概率时间序列。SOI分为五类。根据相关结果得出ENSO影响的空间分布。韩国、日本福冈的月降水量普遍受拉尼娜事件的影响。这种影响在南部沿海地区有4个月的滞后时间，在韩国中北部地区有5个月的滞后时间。其次，详细分析了福冈地区SOI、PDOI、NPI、NAOI 4个指数与正常标准化月降水量的相互关系。对于原始时间序列，在任何滞后时间内，这些指数与降水之间的相关性几乎为零。然而，当将这些指标数据根据其大小分为五组时，获得了更强的统计显著相关性。特别是，在某些类别下，降水与SOI、NPI之间具有很高的相关系数，但存在一定的滞后时间。第三，利用动力系统理论的思想，研究了福冈地区SOI与标准化温度和降水数据联合相空间特征的研究方法。从结果可以看出，三个被调查变量之间的联合关系是复杂的，没有明显的线性关系。该方法可作为联合相关变量的确定性动力学的基础。并对福冈市8月降水进行了中期预报。选择海温异常和3个大气-海洋指数作为候选预测指标。结果发现，以NPI和选择海温作为输入的模型表现相当好。少

First, simple but robust approaches were used to reveal the quantitative and statistically significant influence of the Southern Oscillation Index (SOI) on monthly precipitation at five stations distributed over South Korea and Fukuoka, Japan. The monthly precipitation data were transformed into non-exceedence probability time series. SOI is classified into five categories. The spatial distribution of ENSO influence is obtained from the correlation results. The monthly precipitation in South Korea and Fukuoka, Japan is generally influenced by La Nina events. The influence has a four-month lag time in the southern coastal area, and a five-month lag for the middle to northern area of South Korea.Second, the cross-correlations between four indices of SOI, PDOI, NPI, NAOI and normally standardized monthly precipitation in Fukuoka are investigated in detail. For the original time series, the correlations between those indices and precipitation are almost zero for any lag times. However, qui … More te strong statistically significant correlations are obtained when using those indices data categorized into five groups according to their magnitudes. In particular, very high correlation coefficients are obtained between precipitation and SOI, NPI with some lag times under some categories.Third, a methodology to investigate joint phase space characteristics of SOI and the standardized temperature and precipitation data in Fukuoka was studied byusing ideas from dynamical systems theory. From the results, it can thus be said that the joint relationships between the three investigated variables are complex with no obvious linear relationships. The methodology may serve as a basis for deterministic dynamics of jointly interrelated variables.Further, the medium term forecasting of August rainfall in Fukuoka city was conducted. As the candidate predictors, sea surface temperature (SST) anomalies and three atmospheric-oceanic indices are selected. It was found that a model with the NPI and selected SST as inputs performed reasonably well. Less