Рад у часопису

објављена верзија

српски

Saša MILANOVIĆ, Branislav PETROVIĆ, Vesna RISTIĆ VAKANJAC, Ljiljana VASIĆ, Veljko MARINOVIĆ, Petar VOJNOVIĆ

SIMULACIJA ISTICANJA ODABRANIH KARSTNIH VRELA PIROTSKE KOTLINE

Vodoprivreda

Српско друштво за одводњавање и наводњавање, Београд, Академија инжењерских наука Србије

2022

Analiza recesije, autokorelaciona i kroskorelaciona analiza vremenskih serija padavina i isticanja podzemnih voda karstnih vrela Krupac i Gradište (Stara planina, JI Srbija) izvršena je radi formiranja stohastičkog autoregresivno-krosregresivnog modela (ARCR) i kratkoročne prognoze dnevnih vrednosti izdašnosti. Usled postojanja heterogenosti i anizotropnosti karstne izdani veoma je komplikovano i zahtevno odrediti prostornu distribuciju podzemnih voda u okviru karstnog hidrogeološkog sistema, te je primenljivost determinističkih modela minimizirana. Stoga je upotreba stohastičkog modela u ovom slučaju svrsishodnija. Autokorelogram izdašnosti vrela Krupac pokazao je da memorija sistema iznosi oko 30 dana, a memorija sistema vrela Gradište 14 dana. Kroskorelogram izdašnosti vrela Krupac i padavina pokazuje da je reakcija velikih karstnih kanala sa kašnjenjem od pet dana, dok je statistički uticaj padavina na isticanje vrela Krupac do 25 nakon čega uticaj padavina na izdašnost postaje beznačajan. Kroskorelogram izdašnosti vrela Gradište i padavina ukazuje da je reakcija velikih karstnih kanala sa kašnjenjem od pet odnosno sedam dana. Analizom primenjenog modela ustanovljeno je da su periodi visokih i srednjih voda dobro simulirani, dok su najveće oscilacije u periodu recesije (niske vode), stoga je na podatke o padavinama primenjen filter linearnog pokretnog proseka. Transformisani ARCR-MA model dao je mnogo bolje rezultate u periodu niskih voda, dok su rezultati simulacije gotovo isti u periodima visokih i srednjih voda. Model je pokazao dobro poklapanje izmerenih i prognoziranih vrednosti isticanja na karstnim vrelima, uz koeficijente korelacije RK = 0.9739 i RG = 0.980, što omogućava relativno pouzdanu kratkoročnu prognozu dnevnih isticanja karstnih vrela.

Recession analysis, autocorrelation and cross-correlation analysis of time series of precipitation and groundwater outflow of karst springs Krupac and Gradište (Stara planina, SE Serbia) was carried out in order to form a stochastic autoregressive-crossregressive model (ARCR) and short-term forecast of daily yield values. Due to the existence of heterogeneity and anisotropy of the karst outcrop, it is very complicated and demanding to determine the spatial distribution of groundwater within the karst hydrogeological system, the applicability of deterministic models is minimized, therefore the use of a stochastic model is more expedient in this case. The autocorrelogram of the yield of the Krupac spring showed that the memory of the system is about 30 days, and the memory of the system of the Gradište spring is 14 days. The cross-correlogram of the yield of the Krupac spring and precipitation shows that the response of large karst channels is delayed by five days, while the statistical impact of precipitation on the discharge of the Krupac spring is up to 25 days, after which the effect of precipitation on the yield becomes insignificant, while the cross-correlogram of the yield of the Gradište spring and precipitation indicates that the reaction of large karst channels with a delay of five or seven days. It was found that the periods of high and medium water are well simulated, while the greatest oscillations are in the period of recession (low water), therefore a linear moving average filter was applied to the rainfall data by analyzing the applied model. The transformed ARCR-MA model gave much better results in the low water period, while the simulation results are almost the same in the high and medium water periods. The simulation models showed a good match between the measured and predicted values of groundwater discharge at the karst springs, thus can provide a relatively reliable short-term forecast of the daily discharge of karst springs.

Vol. 54 (2022)

No. 319-320

185

197

0350-0519

Srbija

karstno vrelo, recesija, autokorelacija, kroskorelacija, ARCR MA model, JI Srbija

karst spring, recession, autocorrelation, crosscorrelation, ARCR MA model, SE Serbia

551.49.08

https://www.vodoprivreda.net/#

M50

М52

Отворени приступ

Creative Commons – Attribution-NonComercial-Share Alike 4.0 International

.pdf