Cristian Vasilică SECU¹, Ionuţ MINEA¹ & Iulian STOLERIU^2
1Alexandru Ioan Cuza University of Iasi, Faculty of Geography and Geology, Department of Geography, Bd. Carol I 20A, 700505, Iasi, Romania, e-mail: ionutminea1979@yahoo.com
²Alexandru Ioan Cuza University of Iasi, Faculty of Mathematics, Bd. Carol I No.11, 700506, Iasi, Romania,

GEOSTATISTICAL MODELING OF WATER INFILTRATION IN URBAN SOILS

Full text

The post-communist period (after 1989) caused important changes within the urban space of Iaşi municipality. The partial dismantling of the industrial zone, the urban sprawl against the periurban and rural areas, the new infrastructure works (water and sewage network) led to significant changes in soil morphology and physical properties, including in the infiltration rate (IR). The objectives of this study are: 1) to analyze the spatial variability of IR in urban soils of Iaşi municipality; 2) to identify correlations between IR and some auxiliary physical properties; 3) to choose the optimal geostatistical model, which will accurately estimate the IR in locations with no measurements, enabling the best version of cartographic representation. Technosols are characterized by high spatial variability of infiltration rate, when compared to other soils. Their physical properties differ from one location to another, even in the same landforms, fact that is also confirmed by a high coefficient of variation of IR (82.6%). There is no direct correlation between the IR and the auxiliary properties, and thus the co-kriging method will not useful here. Due to the high values of skewness, we have transformed the original data and used the lognormal ordinary kriging. A comparison between the predicted IR values for the unmodified grid with the measured values do not reveal significant differences, having obtained N/S=0.1798 and R²=0.5184. The elimination of the maximum values of IR or their replacement with some interpolated values for similar conditions of soil and relief are not desirable tasks. Their consideration will lead to high values of the mean square error (MSE) and low goodness of prediction (G).