论文部分内容阅读
近年来,太湖水体出现多次严重的蓝藻水华过程,所以探索气象条件引发蓝藻水华的暴发机制及预报方法有着重要的理论意义及实际价值.由于chl-a含量的动态变化可反映水华与否,可利用粗糙集决策调节回归模型(rough decision-adjusted logistic regression model,RDALR)将粗糙集理论与二元逻辑回归(binary logistic regression,BLR)结合起来,用来建立气象条件和太湖chl-a含量二者之间的联系,为蓝藻暴发期水华预报提供一可行方法,且该方法具有一定可移植性.2012年5月1日-9月5日,对东山24项气象数据和东太湖6个取水口水质监测数据分析表明:RDALR模型解释了东太湖128 d水华监测日71.6％的chl-a动态变化,与BLR模型相比,将39次水华暴发的预报精度提高15％.对RDALR模型分析得出,当模型显著性水平小于0.05时,东太湖水域水华强度与平均风速、平均相对湿度呈显著正相关,与日降水量(P)呈显著负相关,同时水华强度与气温(T)、日照时数(SD)及最大风速风向(WD)等其他气象条件相关不显著.“,”Numerous severe cyanobacterial blooms have appeared in Taihu Lake in recent years.Exploring the influence mechanism,and the forecast method,relevant to the effect of weather conditions on cyanobacterial bloom outbreaks is theoretically significant and has practical value.The dynamic change of chlorophyll-a reflects the presence of blooms and,therefore,the rough decision-adjusted binary logistic regression model (RDALR),combined with the rough decision theory with binary logistic regression (BLR),was used to construct the relationship between weather conditions and the chlorophyll-a content of Taihu Lake.The RDALR model provides a feasible forecast method for the bloom period,and it has certain portability.Analyses were conducted of 24 h meteorological data in Dongshan and the water-quality monitoring data of six water sites in East-Taihu from May 1 to September 5,2012.The results indicated that the RDALR model accurately explained 71.6％ of the dynamic changes of chlorophyll-a in East-Taihu for 128 bloom-monitoring days.Compared with the BLR model,forecasting 39 bloom days,the bloom forecast accuracy was enhanced by 15％.The analysis of the RDALR model indicated that when the significance level was less than 0.05,the bloom intensity in East-Taihu was significantly positively related to AWV and RH,and significant negatively correlated with P.At the same time,T,SD,WD,and other meteorological conditions were not significantly related.