Some properties of drainage outflow rate on eugley

Abstract

Some properties of drain effluent were considered in this paper. The measurements of ground water table depth and drains outflow rates were carried out during the three consecutive off-vegetation seasons, at non-equidistance time intervals. Preliminary analysis showed that there were not moving statistics for the series of data measured during different of-vegetation seasons. Therefore, all the data can be assumed as one compact time series. The type of distribution was determined. To check out normal distribution of time series, data were plotted on both normal probability and log probability paper. The general trend of data distribution was not linear, due to the curvature appearance in the middle of the trend. Therefore, polynomial character of the curve was observed. The principal statistical analysis was aimed to recognize type of distribution of measured data. Obtained results were used for statistical description. In enables choice of a single value to represent a parameter (in this case drainage outflow rate) to be used in further calculations. If random variable have normal distribution, arithmetic mean may be used as the characteristic of simple value, but for data that have Lap laces distribution the median value is the best choice. If data have complex distribution, (like in the case of drainage outflow rate) it's necessary to carry another statistical analysis such as average absolutely value or some higher moments.

U radu su prikazane neke osobine drenažnog isticanja pri odvodnjavanju močvarno-glejnog zemljšta horizontalnom cevnom drenažom. U toku tri uzastopne vanvegetacione sezone praćeno je drenažno isticanje u neekvidistantnim vremenskim intervalima. Prethodna analiza pokazala je da nema pomerenja statistika za podatke dobijene merenjem tokom različitih vanvegetacionih sezona i da se svi podaci mogu tretirati kao članovi kompaktne vremenske serije. Analiziran je tip raspodele dobijenih merenja. Proverom normalnosti izmerenih serija koja se koja je izvršena prikazom podataka na normalnom papiru, uočavaju se kolena (prevoji u trendu) QQ-krive koji jasno naznačavaju nelinearnu (polinomijalnu) prirodu krive, te je pretpostavka o normalnoj, Gausovskoj raspodeli bez osnova. U cilju testiranja tipa raspodele za merenu seriju korišćen dopunski test, kojim je potvrđeno da skup merenja ne podlezę ni Laplasovom zakonu raspodele. Osnovni cilj ovakvih statističkih analiza se sastoji u tome da se na osnovu zaključka o tipu raspodele merenih sekvenci, izvrši njihova odgovarajuća statistička karakterizacija, na osnovu koje će biti usvojeni reprezentativni podaci koji se mogu koristiti u analizama ili proračunima. Naime, za uzorke koji podležu normalnom zakonu raspodele, dovoljno je izračunati srednju vrednost i varijansu dok je za uzorke sa Laplasovom raspodelom, umesto srednje vrednosti informativnije odrediti medijanu. Za uzorke koji imaju složen zakon raspodele, kao što je to ovde slučaj sa drenažnim isticanjem, potrebno osim ovakvih statistika sračunati i srednju apsolutnu vrednost kao i neke momente višeg reda.