MODELLING OF REGIONAL AQUIFER RECHARGE USING REMOTE SENSING METHODS
2006, 21 x 29.7 cm
Price: 30,00 EUR
In this book the possibility of using remote sensing methods in the process of set up and validation of a dynamic physically based distributed hydrological model (MIKE SHE) in the catchment area of Rižana spring is described. The constructed model simulates the whole hydrological circle. Groundwater flow in the aquifer is modelled with the conceptual model. More accurate are modelled processes that influence the aquifer recharge (evapotranspiration, flux in unsaturated zone). To these processes is also limited the use of remote sensing. The data sources for remote sensing methods were satellite images (Landsat TM and NOAA-AVHRR) which were used for the landuse determination, then as a source of information on the development (state) of vegetation and for validation of spatial distribution of simulated evapotranspiration.
The landuse input data in the model were determined with the classification of the Landsat TM image into six vegetation cover classes. The advantage of this procedure is its fastness in acquiring new data and its ability to determine user-defined classes that incorporate vegetation types according to their influence in the hydrological cycle.
The classification was also the basis for the use of the linear mixture model, which was used for determining the development of vegetation classes. Twenty-one NOAA-AVHRR satellite images were used in the linear mixture model. The resulting temporal NDVI profiles of vegetation cover classes are, in general, in agreement with the observed vegetation characteristics. The exception is in the case of the vegetation cover class that covers the smallest part of the study area (2 %). This shows the ability of the linear mixture model to extract subpixel information for the vegetation classes that are well-represented in terms of cover portion within the study area.
By using NDVI temporal profiles for determining the leaf area index temporal profiles in the hydrological model, accuracy of model estimated on the comparison of simulated and observed discharge of Rižana spring was improved. The improvement based on overall RMSE (Root Mean Square Error), Nash-Sutcliffe coefficient and correlation coefficient is small.
The orographic interpolation (Šercl & Lett, 2002) that was used for the precipitation distribution in the hydrological model doesn’t improve the accuracy of the simulated Rižana spring discharge. Improvement is noticed in spatial distribution of simulated evapotranspiration that is estimated on the basis of comparison with the distribution of remotely sensed dryness index TVDI (Sandholt et al., 2002). It was established that the method TVDI might be useful for the comparison of spatial distribution of evapotranspiration (within each satellite image) and limited useful for the comparison between different periods.