Evaluating dynamic hydrological connectivity of a floodplain wetland in North Bihar, India using geostatistical methods.

Spatio-temporal connectivity patterns of a wetland as a function of the land use/land cover (LULC) of its catchment have been analysed in a GIS environment. An innovative method has been implemented for mapping 'dynamic hydrological connectivity' for a water-stressed wetland of Kosi-Ganga interfluve area in the middle Ganga Plains, India for pre- and post-monsoon seasons over a time-span of 29 years (1989 to 2017). It was accomplished by using the time-series NDVI (Normalized Difference Vegetation Index) data and the connectivity response unit (CRU) approach by applying geostatistical methods namely the Getis-Ord Gi* and Mann-Kendall trend test statistics. The study area is principally a rain-fed wetland located in flat terrain (average slope of ~2°) under intensive agriculture and receives water as overland flows. The agriculture dominated LULC in this region is controlling the wetland-catchment connectivity scenarios and the overall connectivity potential of the catchment is higher in the pre-monsoon compared to the post-monsoon season. High and low connectivity potentials of different areas of the catchment with respect to the wetland have been classified into three types: persistent, intensifying, and diminishing. The areas with 'persistent' high or low connectivity potentials have been attributed to the topographic factors which are static in nature, such as the proximity to the wetland and the presence of other geomorphic features. The 'intensifying' and 'diminishing' clusters have been linked to changing LULC patterns. The proposed method holds significant implications for the restoration of wetland-catchment connectivity and can be applied in any flatland terrain where hydrological connectivity is strongly influenced by the surface impedance induced by LULC.