Effect Of Groundwater On Vegetation Index In Hailiutu River Catchment

Hailiutu River Catchment is a typical semi-arid region in northwest China. Due to the lowprecipitation and high evaporation, the local vegetation is controlled by spatial and temporaldistribution of water resources, especially the groundwater resources. The relationshipbetween shallow groundwater and vegetation becomes an important problem of EcologicalHydrogeology. The solution can be a scientific basis to ease the conflict between the utilizationof groundwater and environment protection. With the application of remote sensing technology,this article used Normalized Difference Vegetation Index (NDVI) as a symbol of vegetation,studied the quantitative relationship between NDVI and depth to groundwater table (DWT),defined the dependent pattern and level of vegetation index to groundwater in Hailiutu RiverCatchment.For collecting the DWT data comprehensively, this article proposed a method to designand optimize groundwater level monitoring networks. This method is also appropriate forother ungauged area. Firstly, the referential water table distribution was generated through arough numerical simulation with dense meshes. Secondly, the groundwater level was sampledfrom the modeling results at the nodes on an initially designed network, as a monitoring test.Then, a new contour of groundwater level was obtained with the monitoring test using Krigingmethod and the interpolation error was calculated through comparison with the referentialwater table distribution. With consideration of both the precision and investment, an optimalscheme among the alternative observation networks was chosen. Combined with the previousdata, this optimal method collected the accurate samples of DWT in Hailiutu River Catchment.Combined TM NDVI with SPOT VEGETATION data, interannual and seasonalvariations of average NDVI, with the influencing factors were analyzed. The results showedthat interannual variation of NDVI was not strong and may have positive correlation withprecipitation of the last year. The seasonal variation of NDVI was negatively correlated withdrought index and positively correlated with temperature.After the statistical analysis of NDVI and DWT spatial distribution, it indicated that bothof them showed the skewed pattern of frequency distribution, and the coefficients of skewnesswere positive. The statistical distribution of NDVI for different intervals of DWT was Gammadistribution.The frequency distribution curves of NDVI with respect to different DWT were obtained.The statistical distributions of NDVI values at different DWT intervals indicated that higher vegetation coverage and more plant diversity exist at places of shallow groundwater. Both themean and the standard deviation of NDVI values decreased with the increase of groundwaterdepth when DWT was less than10m. Beyond that depth, a low level of vegetation coverageand diversity was maintained. Comparisons of different sub-areas within the region withdifferent dominant species showed that the NDVI of shrubs was sensitive to DWT. In contrast,NDVI of herbs was not significantly influenced by DWT. The relationship between NDVI andgroundwater depth in farmlands could not be reliably determined because of disturbance byhuman activities. S. matsudana and Populus simonii Carr are the dominant species in grovesof trees, and they can access deep groundwater (7-8m) with their deeply extending roots.For confirming the relationship between NDVI and DWT,3DEM-NDVI-DWT profiles,which locating upstream, midstream and downstream of Hailiutu River Catchment andstretching across the whole district from west to east were chosen. The results verified theconclusion above-the response of vegetation distribution on increase of DWT was positivecorrelation, except the area with manual irrigation.However, the dependency of vegetation on groundwater was significantly different in thearid Ejina area and Yinchuan Plain in China. Both in Ejina area and Yinchuan Plain, a suitableDWT interval could be found for the vegetation growth, but not in Hailiutu River Catchment.These differences may attribute to multiple factors such as climate, soil cover andhydrogeology.The distribution of vegetation can influence the land surface evapotranspiration(ET), andalso be effective to groundwater. As a result, a coupled model of DWT, NDVI and ET wasbuilt. In this model, the relationship between NDVI and DWT, NDVI and ET were presentedas linear formula respectively. The software of Processing Modflow was used to simulate thegroundwater. The Recharge and Evapotranspiration packages were managed specially basedon the linear formula. The model inverted the distribution of average NDVI frequency.This article made a quantitative description of the relationship between NDVI and DWTin Hailiutu River Catchment, built a coupled model of DWT-NDVI-ET. We concluded that thisresearch may significantly improve the contents and methodology of Ecological Hydrogeology,and will provide theoretical significance to the development of Ecohydrology.