This paper analyzes the flood that occurred between 11^th and 13^th of September 2013 in the upper catchment of the river Geru. The flood was simulated using the program Mike by DHI with the Unitary Hydrograph Method. As input data, we used the precipitation measured at the Automated Hydrological Sensor Station Cudalbi and radar precipitations. We analyzed the importance of accuracy for input data on the simulation results and the direct influence of setting the proper time steps in achieving the simulated discharge hydrograph. It appears that radar precipitations used as input data lead to a discharge hydrograph with low errors for amplitude and phase of the runoff peak. The model can be used in the future to reproduce the floods produced in the analyzed catchment and to study the influence of physical and geographical characteristics of the hillslopes.

MIKE software created by Danish Institute of Hydraulics can be used to perform mathematical modelling of rainfall-runoff process on the hillslopes, resulting in a runoff hydrograph in the closing section of a catchment. The software includes a unitary hydrograph method - UHM in the hydrological module Rainfall - Runoff. Excess rainfall is routed to the river and transited through unit hydrograph method. The model divides the flood generating precipitation in excess rainfall (net rainfall) and losses (infiltration). This paper analyzes data from the flash flood that occurred between the 11^th and 13^th of September 2013 in the upper catchment of the river Geru. The catchment chosen for study, is controlled by the hydrometric station located in the village Cudalbi. Simulations of this flash flood were performed with MIKE by DHI –UHM software, alternatively using as input data the precipitations recorded by AHSS (Automated Hydrological Sensor Station) Cudalbi and radar precipitations generated by ROFFG (Romanian Flash Flood Guidance) software system in ArcGIS module for determining the areas affected by flash floods. The Unitary Hydrograph Method - UHM from the hydrological module Rainfall – Runoff calculates excess rainfall and determines infiltration losses by four methods. For each set of input data, the four methods for calculating infiltration losses were subsequently used. The comparison between the results highlights that the amplitude and phase errors for the maximum discharge are smaller when the model uses for simulation radar precipitations as input data, and calculates infiltration losses with the Proportional Loss method. This method reproduces with a better accuracy the peaks of the discharge hydrograph. The model can be used in the future to forecast a discharge hydrograph based on estimated radar precipitations in the catchment

The dam of the non-permanent reservoir Ezer, located on Jijia river is an earth dam with a maximum height of 6.18 m, which provides a global retention to the canopy of 10.330 million cubic meters. The dam founded on weak, muddy soils suffered in the years 1989 and 1992 downstream slope failures of the fillings. It was found that hydrostatic levels were high in the piezometric wells and that consolidation of the foundation soil was reduced. This paper presents a brief history of the dam and aspects regarding the behaviour monitoring of Ezer non-permanent reservoir during the years 2000-2012.