Abstract
Outflow loads were measured using three kinds of watersheds: a forest area (F), an agricultural area without a pig (A) and an agricultural area with many pig farms (P) by an automatic water sampler at intervals of 6 hr for 6 months. Each number of data is 720. From these data, some data (n=6, 12, 18, -----) were picked up using 4 methods: same interval, random, corresponding to discharge and notincluding any storms. Then, the mean of outflow loads were calculated using 4 methods: average by equation (4) (La), revise by flow rate by Eq.(5) (Lw), LQ method by L=aQ+b (straight type) Eq.(3) (Lq1) and LQ method by L=aQb (curved type) (Lq2).
Next, the values of mean X, standard deviation δ and mean squared error e2 were calculated, and squared error ratio E and range ratio R were obtained by Eq.(1) and (2).
Results: 1) LQ method is better than the method of average and revise by flow rate.
2) In the case of NO3-N load by the LQ method, the squared error ratio E is smaller than 10% for the agricultural area when the number of picked up data (n) is larger than 18. In the forest area, it is smaller than 5% when n equals 6 as shown in Fig. 3.
3) A small difference exists between the results calculated by the LQ equation of the straight type and that of the curved type. In the case of a small number of data (n=6), the value of E calculated by the curved type in an agricultural area (A) became large as shown in Fig. 3.
4) In the case of SS, the value of E were so large that it would be impossible to use LQ equation for calculating the outflow load as shown in Fig. 5.
