定量评估气象条件对滇池蓝藻水华发生的影响及预测OA

Predicting the probability of cyanobacterial blooms in Dianchi Lake is crucial for effective water quality management.In this study,a meteorological model for predicting cyanobacterial blooms in Dianchi Lake during the recovery period(March to June)and the high bloom occurrence period(July to December)was developed using the random forest algorithm and daily MODIS data from 2001 to 2021.Information on cyanobacterial blooms was extracted using the Normalized Difference Vegetation Index(NDVI).The relationship between cyanobacterial bloom occurrence and meteorological factors was quan-titatively assessed using characteristic variable importance and partial dependence plots.We aimed to pro-vide scientific evidence for the prevention and control of the cyanobacterial blooms in Dianchi Lake.Re-sults show:(1)During 2001-2021,the average annual cumulative frequency and scale of cyanobacterial blooms in Dianchi Lake were 26.9 times and 7.30％over the 21 years of study period.The occurrence of cyanobacterial blooms showed clear seasonality.Over the study period,the occurrence frequency of cya-nobacterial blooms decreased,while the bloom scale increased.(2)The key meteorological factors affect-ing the occurrence of cyanobacterial blooms were water temperature and wind speed during the recovery period,with water temperature the more important.The key meteorological factors in the high occurrence period were water temperature,wind speed,sunshine,and precipitation,with wind speed most important,followed by water temperature,sunshine and precipitation.(3)Overall,water temperature and precipita-tion intensified cyanobacterial blooms,while wind speed and sunshine had inhibitory effects.The effects of water temperature,sunshine,and precipitation had a cumulative effect on the occurrence of cyanobacte-rial blooms.(4)Each factor affects cyanobacterial blooms within a certain range.Values above or below the effective range may decrease the occurrence of cyanobacterial blooms.When the water temperature was＞18℃ and wind speed was＜2.5 m/s,the probability of cyanobacterial bloom was relatively high.(5)The accuracy,recall rates,comprehensive evaluation score,and Area Under the Curve(AUC)of the model were,respectively,80.1％,62.3％,63.4％,and 87.6％during the recovery period,and 83.1％,85.2％,88.8％,and 86.0％during the high occurrence period.In conclusion,the model has good applica-bility for predicting the occurrence of cyanobacterial blooms in Dianchi Lake.