If the distribution is Gaussian, 99.7% of the data would lie in the range of m±3σ. However, the cluster of the data points of temperature (Fig. 3, left)is ; located much higher than m 3σ, and it extends continuously beyond the line of m+3σ.
The skewed nature is most conspicuous for temperature near 300 m depth. The frequency of the temperature values at 300 m is shown in Fig4. No data exist below 0C, and significant frequency can be seen even for the value of m+9σ. Thus, m±3σ criterion cannot be applied in this region. If the distribution is skewed in the same manner for variable depths, we may determine the range by considering the values of skewness, kurtosis and so on. However, the distribution pattern is changeable with depth and with season. The distribution patterns of temperature for variable depth are shown in Fig. 5. As seen in this figure the distribution pattern for shallower depths than 150 m depth is quite different to those for deeper depth than 150 m depth.
Further investigation is needed to set the reasonable range parameter.
3. The region to the south of the Kii Peninsula
Another example of the skewed distributions is shown in Fig. 6 for a region to the southeast of the Kii Peninsula:the distribution pattern has three eminent peaks. The highest peak corresponds to the state of the Kuroshio Large meander: when the Kuroshio flows in meandering path, very warm Kuroshio Water is carried around the Large Cold Water, and intrudes into the coastal region under consideration. Thes second peak near 24C corresponds to the case of the Straight Path of the Kuroshio:the region is just located in the shadow zone of the Kii Peninsula when the Kuroshio has Straight Path, and the surface water temperature is relatively low. The lowest peak corresponds to the time of upweIling caused by relatively strong and persistent Southwest wind.
