Data Mining on Algae Concentrations (Chlorophyll) Time Series in Source Water Based on Wavelet

The time series composed by daily chlorophyll measurement of Tianjin raw water was transformed by use of wavelet techniques. Then the original time series would be decomposed as 5 low- and lower-frequency period components and 3 high-frequency ´noise´ components. Based on the analysis, it was shown that these messages from the 5 period compounds were mostly caused by seasons, predominant algae kinds switch, algae growth dynamic equilibrium under given conditions and so on, while high-frequency detail message might come from rain, wind direction and the depth of sample raw water, of course, also from the measurement error. At last, these conclusions which were integrated with air temperature, radiant intensity and rain data were farther discussed, furthermore the meaning of these components and conclusions were validated.