Significant Factor Identification, Short Term Forecast, And Regional Cooperative Control For Cyanobacterial Blooms

Many lakes and reservoirs all over the world become eutrophic because of thedevelopment of society and economics, and cyanobacterial blooms occur in thesewater bodies. This dissertation discusses the environmental factors significantlyassociated with cyanobacterial bloom outbreaks and intensity, tries to predictcyanobacterial blooms ahead of a short period, and explores the cooperativeapproach for cyanobacterial bloom control among several regions in a lake basin.This dissertation uses Lake Tai and the Lake Tai basin for empirical studies. Theinnovations in this dissertation include the following content.1. A binary dependent variable was constructed based on inverted pictures forcyanobacterial blooms, and a probit model was used to identify environmentalfactors significantly associated with cyanobacterial bloom outbreaks.A probit model was constructed in this dissertation for identifying significantenvironmental factors for cyanobacterial bloom. The model for the first time directlyuses the index of occurrence or not (which is the judgement result on invertedpictures for cyanobacterial blooms) of cyanobacterial bloom as a dependent variable,and simultaneously uses the monitoring variables of water quality, hydrology andweather (altogether19variables) as independent variables. This model was appliedto the Hill Dagong water area of Lake Tai in China, and4significant environmentalfactors associated with bloom outbreak were systematically identified. Nitrateconcentration, pH, and water depth are positively correlated with the probability ofcyanobacterial bloom outbreak; wind speed is negatively correlated with theprobability. Among the4significant factors, nitrate concentration is mostsignificantly correlated with the probability. The above conclusions as a wholecoincide with the general opinions in this research field.2. The intensity grades of cyanobacterial blooms were built based oncyanobacterial bloom area and aggregation intensity, and a multivariate linearregression model was used to identify significant environmental factors based on proper spatial and temporal precisions.To identify significant environmental factors of cyanobacterial bloom intensity andto overcome the defaults such as improper choices for the dependent variable, coarsetime and space precision in the previous research, a multiple linear regression modelwas developed in this study. In this model, the intensity rank of cyanobacterialbloom is used as the dependent variable, and the monitoring variables of waterquality, hydrology, and weather are used as the independent variables. Based on thedata of bloom area and aggregation intensity, I used a7-level scale to generate thevalues for bloom intensity ranks, which makes the dependent variable have amoderate macroscopic characteristic and avoids the default of using indices such aschlorophyll a concentration to represent bloom intensity. This model was applied tostudying cyanobacterial blooms in Lake Tai. The analysis result of this model revealsthat the intensity rank of cyanobacterial bloom in the Hill Dagong water area hadsignificant positive correlation with air temperature and nitrate concentration, andhad significant negative correlation with wind speed, all at the significance level of1%. These conclusions coincide with the general conclusions in this research field,which manifests the validity of this identification model.3. The index of presence or absence of cyanobacterial blooms was applied as apredicted variable, and a short-term probit forecast model was used to predictcyanobacterial bloom outbreaks.Inverted pictures of cyanobacterial blooms were used for judging whethercyanobacterial blooms outbroke in a given water area at given time, then a probitshort-term forecasting model for bloom outbreak was developed. The model directlyuses the binary variable of occurrence or not of cyanobacterial blooms as a predictedvariable, and uses the significant factors identified in the above two studies aspredictive variables. The Hill Dagong water area was used as a case for the empiricalstudy of this forecast model. The results indicate that the values of evaluationindicators of this model are good; the mean relative error of this model is13.3%which is close to or lower than that of two previous models; compared with the previous models, this model has obvious advantages in both spatial precision andtime precision. The accuracy of the forecast model for the next day is the highest, thepredictive accuracy lowers when predictive cycle lengthens. In addition, thepredictive accuracy and the validity indicators of this forecast model are higher thanor close to those in the case of bringing all the available monitoring variables intothe forecast model, revealing the cost-efficiency of our model.4. Based on exploiting nutrient salt reduction cost differential among differentdepartments and regions, a bi-level optimization model was developed topromote the cooperative reduction for nutrient salt, therefore realizing thelong-term control for cyanobacterial blooms.The author combined ecological reduction, industrial reduction, and municipalsewage plant reduction to solve water eutrophication problem. Based on thistechnical frame, an optimal allocation model for reduction targets for nutritive salt ina lake basin is proposed in this dissertation. This model seeks the optimal reductionquantity of each region in the basin, and compensates among the regions by transferfee for the transfer of initial reduction targets. This model can give full play to thecost advantages of nutrient reduction across the regions and departments in the basin,can promote cooperative reduction for nutritive salt among the regions anddepartments, thus optimizes allocation of resources and decreases the reduction costsfor nutritive salt in the basin. Transfer fee method can make full use of theadvantages of horizontal transfer payment. The empirical study indicates thatcomparing with the current management mode, if this model is implemented forammonia nitrogen reduction in the Lake Tai basin in2005, an amount of CNY (orRMB)10540.08×104can be saved for the reduction cost, and the saving proportionis as high as16.6%.The above studies are helpful for understanding the mechanism for cyanobacterialbloom outbreaks from a new viewpoint, and provide a long-term control approachfor cyanobacterial blooms.