| Submarine Fresh Groundwater Discharge(SFGD),an important component of the coastal zone water cycle,is an important source of land-based material in the ocean.In this study,we address the problem of the lack of studies on the fluctuation of SFGD in the Bohai Sea basin over a large area and the inadequate use of spatial data products,construct a water balance equation for the coastal basin.The datasets containing three spatial precipitation products and 15 precipitation-related feature parameters are constructed.Six different complex precipitation data fusion machine learners including stacked generalized integrated learning used and compared in precipitation data fusion.Then we a multi-source precipitation data fusion product with 0.05° resolution for the Bohai Sea basin was generated,the applicability of the stacked generalized integrated learning fusion precipitation product in China was verified,and the accuracy of SFGD estimation was improved.Finally,combining all the water balance component data products including actual evapotranspiration and groundwater storage variables,the SFGD fluxes in the secondary tributary basins of the Bohai Rim from 2004 to 2018 were estimated and analyzed.The results show that:(i)the improved large-scale and small-scale terrain undulation parameters used to characterize terrain position are able to identify the degree of terrain undulation at different spatial scales and distinguish different terrain positions,besides,the related custom parameters can be determined by priori knowledge of the study area without exhaustive tests,all indicate the strong usability of proposed method;The constructed stack generalized integrated learning models EM3 and EM4 can combine the advantages of multiple learners,reduce the inter-site and inter-seasonal accuracy differences,improve the spatial and temporal heterogeneity of data accuracy,and have strong application value in the context of many model and data choices;The precipitation data fusion results based on the five learners of SVR,RF,MLP,EM3 and EM4 all improve the mean values of various indicators,among which the precipitation fusion based on the SVR of the 2nd linear kernel function data with the observed values of CC,NSE and KGE on the validation set have been able to reach 0.93,0.86 and 0.89,and the improvement of model complexity on this basis has limited effect on the optimization of accuracy;The monthly-scale precipitation data generated by the study for the Bohai Basin from 2001-2018 show that the trend of decreasing annual precipitation in the Bohai Basin has eased,with the boundary between the main stream basin of the Liao River and the Huntai River basin and the Tuhaimajia River basin as the boundary,and the basin to the west has a trend of increasing precipitation despite low precipitation,with the largest increase in the northern system basin of the Hai River basin,and conversely,the area to the east of the boundary has higher precipitation but no significant trend of change;(ii)The recharge-discharge difference calculated from precipitation,actual evapotranspiration,river discharge/recharge,interbasin transfer and groundwater storage variables is defined as the water balance difference excluding the exchange of water at the land-sea subsurface interface(ΔW),when ΔW is positive,there is a surplus of water at the land end,which is the annual SFGD flux.From 2004 to 2018,the annual average SFGD for the region was 108.96×10~8 m~3/a,mainly happened in the Luan River and Jidong coastal basin,the below-Huayuankou basin,and the coastal river basins of Shandong Peninsula,with annual average SFGD fluxes per unit area of 0.07 m/a,0.05 m/a,0.04 m/a,respectively.SFGD is influenced by other water balance components and fluctuates greatly over time,with no obvious trend.In addition,the annual average AW in the whole region is-149.71×10~8 m~3/a,which occurs as a recharge component,and groundwater recharge or seawater intrusion may occur at the land and seawater boundaries of the basin;the difference in ΔW between wet and dry years can reach 938.84×10~8 m~3/a.Among them,the ΔW in the Liaohe River basin,the Huntai River basin,river basins northeast along the Yellow and Bohai sea,the Haihe northern system and the below-Huayuankou basin oscillates between positive and negative values,and is a sensitive area for ΔW changes with water resources;while the Luan River basin and the Jidong coastal and coastal river basins of the Shandong Peninsula river basins are predominantly SFGD,i.e.ΔW is predominantly positive;ΔW in the Tuhaimajia River and Haihe southern River basins is negative all year round.Compared to river recharge/discharge,ΔW is more strongly coupled to precipitation-evapotranspiration differences and is more influenced by water resources,especially in the Luanhe and Jidong coastal,Haihe northern and southern river basins,where river recharge/discharge is rare and ΔW is particularly sensitive to changes in the balance of other recharge-discharge components.ΔW is more sensitive to changes in the balance of other recharge-discharge components in the study area from 2004 to 2018.In the period 2004-2018,the precipitationevapotranspiration differential across the region showed a decreasing trend in the basins in the central and western part of the study area,especially in the below-Huayuankou basin,which,together with the excessive transfer of water from the Yellow River in some years,resulted in a negative value of ΔW.Inter-basin transfers across the region were negatively correlated with groundwater storage variables and ΔW,reflecting to some extent the positive role of inter-basin transfers in alleviating groundwater resources problems. |
PDF Full Text Request