【作者】 卢小慧；

【导师】 靳孟贵； Karsten H.Jensen；

【作者基本信息】 中国地质大学 ， 水文学及水资源， 2009， 博士

【摘要】 受全球气候变化及人类活动的影响，水资源的短缺及有效管理，成为日益关注的热点问题。采用地表水-地下水耦合的分布式水文模型，分析水循环过程、拟定流域水资源管理方案，对于揭示水循环规律、有效管理水资源具有重要理论与实际意义。这些年来，众多科研机构开发了一系列复杂程度各异的模型软件。在这些软件开发过程中，对非饱和水流运动过程的刻画是水循环中最关键的环节之一，因为其计算精度直接影向着模拟结果的可信度。非饱和水流运动过程一般用传统的Richards方程来刻画。基于Richards方程建立流域水文模型的一个局限性在于需要大量的参数。Richards方程主要适用于刻画小尺度均质的非饱和流，但它经常被用于刻画大尺度非均质流动系统，这个问题引起了人们对Richards模型在大尺度流域模拟中的可行性和有效性的讨论。运用Richards方程来刻画流域尺度上的非饱和流增加了对计算机运算能力的要求，因为获取方程的数值解需要对模型进行空间和时间上的离散。运用Richards方程计算较复杂，在流域尺度上是否有更加简单快捷的模型能够获得和Richards方程相同或者更好的模拟效果呢?全球气候变化和日益加剧的人类活动(土地利用方式变化)对水文循环产生了深远的影响。对同一地区，不同土地利用方式、不同尺度(点和流域尺度)对水文循环有何影响，对于不同地区，不同的气候条件和土地利用方式对水文循环有何影响。针对这些问题，本文通过丹麦水力学研究所(DHI)开发的分布式水文模型(MIKESHIE)软件建立地表水-地下水耦合的Skiern流域分布式水文模型，并对此模型进行参数率定。在此基础上对以下几个方面进行研究：(1)模型结构方面，非饱和流概念模型对流域水文向应的影响(2)模型应用方面，以丹麦Skiern流域流域为例，不同土地利用方式、不同尺度(点和流域尺度)对流域水文响应的影响；以丹麦Skiern流域和华北平原为例，对比分析不同的气候和土地利用方式对水文循环的影响。(1)非饱和概念模型对流域水文响应的影响本文在建立Skjern流域地表水地下水耦合的流域水文模型的基础上，比较了这些非饱和流概念模型模拟的水文响应异同点。基于丹麦土壤属性数据库，采用了Pedo-transfer-functions (PTFs)方法获取土壤水力参数。根据土壤颗分数据，0-30cm的土壤具有2种类型(粗砂，亚粘土)，30cm-70cm的土壤有3种类型(粗砂，粘土，细砂)，70-300cm的土壤有4种类型(粗砂，砂质粘土，亚粘土，粘土)。利用ARCGIS软件进行空间叠加获得24种不同的岩性组合，来刻画Skjern流域非饱和带的非均质性。针对上面提出的问题，本文建立5种不同的非饱和流水文模型：两层水均衡模型TLM (Simple two layer water balance model for the root zone andinstantaneous routing of the excess water to water table)简单参数化的Richard方程模型RI2( Richards’ equation -simple parameterization)基于PTFs和土壤数据库获取参数的Richard方程模型RI24 ( Richards’ equation-parameterization determined on the basis of pedo-transfer functions and detailed soilmaps),考虑优先流的Richard方程模型RIBP (Richards’ equation with bypass to accommodatefor spatial variability)考虑重力流的Richard方程模型RIGR (Simplified version of Richards’ equation onlyconsidering gravity flow)分别对这5种模型进行参数率定之后，得出如下结论：(1)5种非饱和流模型都具有相同的季节变化特征，在春季和冬季，降水量丰富，地下水补给和河流径流量大，而在夏季和秋季，降水量减少，相应的地下水补给和河流径流量减少。(2)水文响应时间RIBP和RIGR快于TLM，RI2和RI24，RIBP和RIGR模拟地下水位和流量数据与观测值的响应时间一致，RI2，RI24和TLM模拟的地下水位和流量数据与观测值比较有一定的滞后性。这是由于Skiern流域大部分是砂土，大孔隙较多，具有优先流存在，而RIBP能刻画这部分优先流，故其水文响应时间和观测值一致，而其他模型忽略这部分优先流存在，故其与观测值相比有一定的滞后效应。(3)TLM模拟的地下水补给量和径流量比其他模型大。对比模拟效果，选取Nash and Sutcliffe系数(R2)，均方差函数和变异系数为评价指标，5种模型中TLM的模拟结果和观测值最为接近，TLM较其它4种模型而言更简单，对模型作参数率定时所需运行时间更短，故TLM相比于RI更适用于Skjern流域。(2)不同土地利用方式对流域水文响应的影响土地利用方式对流域尺度水文响应的影响是一个未解决的问题，并且受到大多数发展中国家的关注(Im et al．2005)。采用极端土地利用法在Skjern流域构建土地利用变化情景，运用考虑植被影响的MIKESHE模型研究土地利用变化情景下的水文响应。场景一：将流域内植被全部调整成森林，流域内林地面积增大996．75 km~2，变化后森林面积占总面积的84%。对比变化前后结果差异发现，在1990年(丰水年)，径流最大峰值为28．67 m~3/s，比变化之前减少了5．48m~3/s，峰值出现时间推后1天；在1996年(干旱年)，径流最大峰值为13．63m~3/s，比变化之前减少了5．44 m~3/s，峰值出现时间推后2天，说明在其他条件(降雨、地形)一定的情况下，森林面积增加能够减少产流总量、径流深度、径流系数，削弱洪峰流量，增加地下水补给，提高地下水位，减少水土流失。同理，根据极端土地利用方法，将Skjern流域植被类型100%调整为草地，欧石楠，谷和果树，对Skjern流域的蒸散发量，径流量，地下水补给量进行模拟研究。结果表明草地的蒸腾量在草地，欧石楠，谷和果树这4种植被中最大，约467mm/y，并引起洪峰流量增加、峰现时间提前。(3)不同尺度(点和流域尺度)对水文响应的影响在Skjern流域选择5个不同土地利用方式(森林，草地，谷，欧石楠和果树)的典型区，用荷兰自由大学开发的集中参数模型(EARTH)建立5个土地利用方式不同的模型，并对这5个模型进行参数率定。运用点面结合的方法，对比EARTH模型获得的点尺度上的模拟结果和MIKESHE流域尺度模拟结果。结果表明，EARTH模型对森林，草地，谷，欧石楠和果树这5种植被进行模拟，草地实际腾发量最高。通过对两个模型的地下水位模拟值和观测值对比，发现两者模拟效果都很好。说明简单的EARTH模型和复杂的MIKESHE模型都适用于Skjern流域。但是，EARTH模型计算的补给量和地下水位偏高，主要是没有考虑地下水开采。(4)不同的气候和土地利用方式对水文循环的影响选取两个典型区丹麦Skjern流域和华北平原，研究不同的气候和土地利用方式对水文循环的影响。运用一维非饱和流模型HYDRUSID模拟华北平原5个典型区的水文响应。运用一维非饱和流模型DAISY，考虑植被作物生长全过程(从作物播种到收割)模拟。在Skjern流域选择4种典型植被研究植被变化对水文响应的影响。a．不同气候条件对水文响应的影响。丹麦属于温带海洋性气候，年降水700-800 mm，40%的降雨集中于9月到11月，4月到8月间降雨量偏少。华北平原属于季风气候影响下的半干旱半湿润气候，年降水在500-600mm，80%的降水集中在6月到9月。故两者的地下水补给季节性变化特征不同，丹麦补给集中在9-12月和1-3月，而华北平原地下水补给多集中在6-9月。b．地下水埋深对地下水补给的影响。这两个研究区有着共性，对于浅层地下水，入渗补给很快到达地下水面，峰值的滞后时间不是很明显，大约都滞后1-2天；对于地下水位埋深在2-5m之间的，年地下水补给量峰值代表一系列降雨量产生的累积效应，滞后现象较为明显，峰值滞后4-5天。对于地下水位埋深大10m，峰值滞后时间很长。c．不同的土地利用方式对水文响应的影响。丹麦Skjern流域农作物主要是夏玉米，冬小麦和果树，华北平原作物主要是夏玉米和冬小麦。本文用DAISY模型研究了作物生长全过程对水文响应的影响，包括从作物播种到收割的全过程，结果表明考虑作物生长全过程，这两个研究区具有相似性，夏玉米生长期地下水补给量大于冬小麦生长期。d．不同的灌溉方式对水文响应的影响。由于丹麦雨量充沛，据统计数据表明全年灌溉量约60mm，灌溉方式对地下水补给量占降水和灌溉总百分比很小，而华北平原属于半干旱半湿润气候，全年的灌溉量是全年降水量的50%左右，华北平原灌溉对地下水产生的补给约占降水和灌溉总量的27-49%。创新点(1)研究分布式流域水文模型中不同非饱和流概念模型(TLM，RI2，RI24，RIBP，RIGR)对流域水文响应的影响。(2)基于高分辨率的DEM与土壤数据库，构建地表水和地下水耦合的流域尺度分布式水文模型。采用PTFs方法土壤水力传导系数，通过GIS进行空间叠加，表征Skjern流域的土壤非均质性。(3)对比丹麦和华北平原两个研究区，研究不同气候、不同土地利用方式条件下水文响应的异同点。更多还原

【Abstract】 Water is one of the most precious natural resource,upon which mankind depends forsurvival.The Water Framework Directive adopted by the European Union prescribes that waterresources management strategies must be developed at catchment scale - the natural geographicaland hydrological unit - instead of according to administrative or political bounds and theecological state of both surface- and groundwater must be considered.The catchment or basinscale is also used as the management scale in the concept of Integrated Water ResourcesManagement (IWRM) that is widely adopted as the management principle in developingcountries.The catchment or basin scale is also used as the management scale in the concept ofIntegrated Water Resources Management (IWRM) that is widely adopted as the managementprinciple in developing countries.In water circle,there is one-way or two-way transformationbetween rainfall,surface water,unsaturated flow and ground water,while surface water andgroundwater separation method in active water resources evaluation separate the complicativetransformation between surface water and ground water.In a word.Integrated Water ResourcesManagement (IWRM) will be a trend of Chinese water resource management in the future.A number of modeling tools have been developed over the past decades with quite diversityin complexity and in how the individual hydrological processes are represented.Water dynamicsin the unsaturated zone is one of the most important processes as it controls model precision.One of the limitations in using a catchment model based on a Richards’ equation formulation isthe huge amount of parameters required to run the model.Using Richards’ equation as themodeling platform for unsaturated flow simulations at catchment scale increase thecomputational demand,as small spatial and temporal discretizations are required in thenumerical approximation of the equation.Given these complications a relevant questiontherefore arises whether simpler and less demanding modeling approaches will provide same results or perhaps even better at catchment scale?Global climate change and human activities (land use change) have a great effect on thehydrological circle.To one study area,different land use and scale have an effect on hydrologicalcircle;to different study areas,and different climate and land use conditions have an effect onhydrological circle.Aiming at the problem,MIKESHE software developed by DHI (Danish hydrology institute),which represent a state-of-art of distributed hydrological model,is used to establish surface waterand ground water coupling model of Skjern catchment,and standard parameterization andcalibration of the model has been done.The objective of this study is to test and analyze theeffect of using different models for unsaturated flow on the dynamic response of an integrateddistributed hydrological model and analyzes effect of different climate and land use condions onhydrological circle.The effect of different unsaturated flow models on dynamic response of afully-distributed hydrological modelBased on soil database of Denmark,Pedo-transfer-functions (PTFs) are used to derive hydraulicparameters of soil.From surface to 30cm depth,there are 4 types of soil,from 30cm to 70cm,there are 2 types of soil,from 70cm to 300cm,there are 3 types of soil,and 24 combinations ofsoil are obtained through ARCGIS software in order to describe the heterogeneity of unsaturatedzone in Skjern catchment.Aiming at the problems,5 unsaturated models have been established:TLM (Simple two layer water balance model for the root zone and instantaneous routing ofthe excess water to water table)RI2 (Richards’ equation -simple parameterization)R124 (Richards’ equation -parameterization determined on the basis of pedo-transferfunctions and detailed soil maps)RIBP ( Richards’ equation with bypass to accommodate for spatial variability)RIGR (Simplified version of Richards’ equation only considering gravity flow)The results show that four kinds of unsaturated flow model have the same seasonalcharacters,in spring and winter,the rainfall is plentiful,the groundwater recharge and riverdischarge is big,while in summer and autumn,the rainfall reduced,the groundwater recharge andriver discharge decreased.The hydrological response of bypass flow model and gravity flow modelis quicker than two layers model and Richard equation model,which the simulated groundwaterlevel and discharge of bypass flow model and gravity flow model have a good agreement with theobservations,the simulated groundwater level and discharge of Richard equation model and twolayer model have a time-lag compared with observations,one reason is the sand covers main areaof Skjern catchment,macropore exist,and preferential flow occurs.The groundwater recharge andriver discharge of two layers model is bigger than the other three models.Compared simulatedeffect of four unsaturated conceptual models,Nash and Sutcliffe coefficient (R2),Root MeanSquare Error (RMSE),Correlation Coefficient (R) are chosen as model performance criteria,thesimulated results of two layers model are the most similar to the observations in four unsaturated conceptual models,and further more it is the most simple one which needs the shortest computertime,therefore two layer model is more fitable for Skjern catchment.The effect of land use change on water balance components such as groundwaterrecharge,evapotranspiration and groundwater dynamic responseSeveral land cover scenarios in the upstream area of Skjern catchment were developed by theextreme land use/cover method.Based on those land cover scenarios,MIKESHE model was usedto predict the possible hydrological variations resulted from land cover change.Results also showthat afforestation in the Skjern catchment might raise evapotranspiration and reduce runoff,and cutdown and delay flood peaks effectively.Conversion from forests to grassland,grain,heather andhomestead covers tends to reduce evapotranspiration and increase runoff volume.This invariablycould make flood peaks increase and appear in advance.Site scale and Catchment scale combinations in Skjern catchmentActual evapotranspiration,percolation,recharge and groundwater level fluctuations weresimulated for a period of twenty years (1990-1999) using EARTH and MIKESHE models Theresults of EARTH model represent one dimensional site scale,while MIKESHE model representcatchment scale.Both models appear to simulate the slow groundwater level fluctuations of thestudy area with high accuracy from 1990-1999.However,the simulated result of EARTH modelis a bit higher than the observed data probably as a result of it does not consider groundwaterabstractions.Nearly identical actual evapotranspiration and groundwater recharge values wereobtained from the simulation results of both models.The goodness of fit between the observedand simulated groundwater levels is also equally good for both models.It is recalled that bothmodels share the same boundary conditions.Hence it can be concluded that the water balancecalculation is depending more on the top boundary conditions than on the complex soil waterflow in the unsaturated zone.In general the actual evapotranspiration for forest is lower than forgrass and heather.On the contrary,recharge values below forest is almost the same as grass andheath lands.This shows that in forested areas,evaporation of intercepted rainfall contributessignifcantly to the total loss of water due to evapotranspiration causing recharge to be small.Theclose agreement between the simulated results of the two models also indicates that the simpleEARTH model is equally suitable as the more complex MIKESHE model at least for thecondition of the Skjern catchment.Comparing the hydrology response of groundwater in Skjern catchment and theNorth China plaina.Different climatic conditionsDenmark is warm climate,of which mean annual precipitation is 700-800 mm.The greatestrainfall comes between September and November.The Hebei plain is a semi-arid to semi-humidarea with a monsoon-dominated climate.Mean annual precipitation is only 500-800 mm.ofwhich more than 80% occurs in the months of June through September.The groundwaterrecharge of Skjern catchment mainly concentrated from September to November,while of theNorth China Plain mainly from June to September if not considering the effect of irrigation.b.Different groundwater depth When water table is shallow,Infiltration reaches groundwater relatively quickly,so thatmost of the individual rainfalls correspond to isolated infiltration recharge events with very smalltime-lags,the peak time-lag is about one or two days.As the depth to groundwater increases,the correspondence between rainfall and rechargedecreases;the recharge events produced by individual rainfalls merge into one single annualprocess although a few peaks of recharge correspond to large rainfall events or concentratedrainfall clusters.When water table is deep,Infiltrated water takes a long time to reach groundwater table sothe peak time lags here are 18 to 35 days even longer.c.Different land use considering crop rotationIn spring barley,winter wheat,maize and grass,the average annual actual evaportranspiration ofspring barley is the smallest one about 384.8mm,and its average annual groundwater recharge isthe biggest one about 255.1 mm in 4 types of crops.The groundwater recharge of maize is higherthan winter wheat in Skjern catchment and the North China Plain.d.Different irrigation methodIn Skjern catchment,the annual irrigation is about 60 mm.The groundwater recharge fromirrigation is a small part occupied in total rainfall and irrigation.In the North China Plain,theannual irrigation is about 300-450 mm.The simulation results show that groundwater recharge decreased significantly and thatrecharge caused by irrigation in the Hebei Plain accounted for 27-49% of the total precipitationand irrigation.Irrigation-induced recharge at Luquan was largest among the five sites (49% ofthe total precipitation and irrigation).Recharge caused by irrigation in the piedmont plain wasfound to be slightly larger than that in the alluvial and lacustrine plainInnovations1.The hydrological response of four unsaturated conceptual models is compared incatchment scale.The applicability of these models are analyzed which enrich theexperience of distributed hydrological models.2.Based on high resolution DEM and soil database,Distributed hydrological models ofSkjern catchment established which coupling the surface water and groundwater.Pedo-transfer-functions (PTFs) have been used to derive soil hydraulic parameters incatchment scale which is benefit for describing soil heterogeneity in Skjern catchment.3.Combine the results of Denmark and China,the effect of different climatic conditionsand land use on hydrological response was analyzed.更多还原