Wyniki wyszukiwania - BazTech

Ograniczanie wyników

1 2002

Znaleziono wyników: 1

Liczba wyników na stronie

Wyniki wyszukiwania

Sortuj według:

Ogranicz wyniki do:

Model wahań zwierciadła wód podziemnych w Puszczy Niepołomickiej

Chełmicki W., Ciszewski S., Żelazny M.

Czasopismo Techniczne. Środowisko

2002

R. 99, z. 4-Ś

19--26

The Niepołomice Forest is situated in the vicinity of the Nowa Huta steel complex, one of the largest industrial enterprises of this type in Poland built in the early 1950's (Fig. 1). Extemsive industrial emissions over the last 50 years are influencing the condition of the forest environment. The magnitude of this influence and the resulting destruction have been investigated by many forest-ecology teams. The ground-water table, alongside industrial emissions, has the potential to influence the state of the forest. Information on ground-water levels before and after the opening of the steel complex should make an assessment possible as to what extent the ground-water table fluctuations have been influencing the state of the forest environment, and to what extent its destruction has resulted from industrial emissions. The paper aims to reconstruct the ground-water fluctuations in the Niepołomice Forest for the period 1901-1960, having avaiable the ground-water table data for 1961-2000 and meteorological data for the whole of the 20th century. The Niepołomice Forest is situated in the Vistula River valley. The dominant bedrock is Pleistocene sand, clay and gravel of glacifluvial and fluvial origin covering deeper Miocene silt and clay. Part of the sands is wind-transformed and forms chains of dunes. The ground-water table depth is usually between 1-4 m except glacifluvial hills and dunes, where it locally exceeds 8-10 m. The dominant plant communities are coniferous fresh and wet mixed forests - ca 50%, deciduous forests - 24% and fresh and wet mixed forests - 20% [2]. Available data and method of reconstruction of ground-water fluctuations. There was only one water-ground monitoring post (Poszyna) in the Niepołomice Forest. There the ground-water levels were measured on a weekly basis since 1953, although reliable data without major gaps only covers 1957-2000. Other posts distributed in the surroundings of the Niepołomice Forest were operating irregularly and just for short periods. Long-term meteorological data are available for Kraków. Assuming the relationship between the meteorological and ground-water variables and having available meteorological data for 1901-2000 (Kraków) and the ground-water table data for 1961-2000 (Poszyna), an attempt was made to build a neural-network model of ground-water fluctuations for Poszyna, to reconstruct ground-water table fluctuations before 1961. The monthly means of the ground-water levels and air temperature, and monthly totals of precipitation were used for model calibration. A predictor concept was applied to solve the issue of ground-water level modelling based on artificial neural networks. The predictor is a conversion of a past series of data into a simulated present value. The method of learning artificial neural networks is a classical algorithm of back-propagation with monumentum, mixed with pruning by optimal brain surgery [4]. There are two modelling tasks. The first gives an answer to the question: how useful for ground-water level modelling at Poszyna are the meteorological data from Kraków. The series of both meteorological data and the ground-water level are the model inputs. The results are satisfactory (Tab. 1). The pruning method gives us a small networks, which are easily interpreted. The goal of the second task is the construction of a model that can be used to find a relationship between the ground-water level in Poszyna and the Kraków meteorological data. Although the results are not as in the previous case (Tab. 1) they are still useful in obtaining values close to the real ones. This model was used to perform a full conversion of the Kraków meteorological data into the Poszyna water level series. The obtained series conform to the forest behaviour at the time of measurements in the learning set. Both, reconstructed and observed ground-water levels at Poszyna mirror the periods of dry, medium and wet years. Having the ground-water levels for Poszyna for the whole 20th century, the next step of the procedure was to reconstruct ground-water table fluctuations in other parts of the Niepołomice Forest. In order to find the relationship between ground-water fluctuations at Poszyna and other parts of the area, nine piezometric wells were established in different parts of the forest , and then used for a one-year-long (May 1999-April 2000) monitoring, which was simultaneous with the one at Poszyna. The one-year-long series of measurements has shownn a well expressed linear relationship between water levels in the piezometric wells and Poszyna (Fig. 3 and Tab. 2). This relationship was used for calculation of the ground-water levels at the monitoring sites for the whole of the 20th century. As the linear relationship between the Poszyna and piezometric wells was based on a one-year-long series of observations, the question arises whether this relationship is valuable for the whole of the 20th century. One of the factors to consider is the changing age of the tree stands and resulting changes in transpiration efficiency influencing ground-water levels over the period studied. The trial approach to the problem was made using the formula of Suliński [5] showing the dependence between the mean depth to a ground-water table H [m], a stands age A [years], and the coefficient of filtration of water bearing deposits K [10-5 m/s]: H[A]=40,05 K0,198 A0,278 e-0,020A. Fig. 4 shows the ground-water levels for the 140-year-old pine stand (Molinio Pinetum) at site 15; one without the correction for tree age, and another - with the correction based on the formula (1). The doubts connected with use of the formula arise due to the unknown effect of the tree ageing at the Poszyna post, which is situated in a woodland glade, surrounded by forest of unknown effect on the ground-water level at the monitoring site. The whole reconstruction procedure is shown in Fig. 5.