【作者】 高鹏；

【导师】 赵庚星；

【作者基本信息】 山东农业大学 ， 土地资源利用， 2022， 博士

【摘要】 黄河三角洲是我国濒海区土壤盐渍化典型地区,土壤盐渍化影响农业生产与土地资源的有效利用,是制约该地区可持续发展的主要因素。探明黄河三角洲土壤盐渍化的变异特征与驱动机制,进而实现盐渍化的预测预防与预警治理,是土壤资源合理利用的前提,对保障黄河三角洲粮食安全和农业可持续发展具有重要意义。本研究以黄河三角洲典型地区垦利区为研究区,分区域、地类、剖面三个尺度进行研究。区域尺度对应整个研究区,地类尺度分耕地、荒地代表类型,剖面尺度以土体构型为主要对象。区域和地类尺度通过野外采样,运用空间插值分析土壤盐渍化空间变异特征,提取气候、高程等因素空间分布数据,筛选确定主要影响因子,采用回归分析和地理探测器方法进行驱动力分析,探明其驱动机制,利用随机森林实现驱动模型构建。剖面尺度利用室内土柱试验方法,摸清垂直方向盐分分布规律,探明土体构型、地下水矿化度、地下水位、覆盖度对剖面水盐运移的驱动机制,利用多元线性回归实现剖面尺度盐渍化驱动模型构建。基于各尺度下驱动模型和历史数据预测秋季土壤盐渍化空间分布,以区域土地生态利用和作物产量作为预警等级划分依据,实现多尺度盐渍化预警,为盐渍化预防和治理提供理论依据和技术支持。主要研究结果如下:（1）多尺度盐分空间变异规律区域尺度上,盐分空间分布为东高西低、北高南低的格局,自西南内陆至东北临海区土壤盐分含量逐渐升高,土壤盐渍化逐渐加重,黄河径流附近有明显差异,局部降低。地类尺度上,耕地土壤盐渍化水平较低,多数面积土壤含盐量在1.5–5.0 g·kg^-1范围,整体分布为自西部至黄河口镇为轻、中度盐渍化区,东部平行海岸线分布条带状重度盐渍化区;荒地自西向东、自北向南土壤盐分等级不断提升,其盐分空间分布与区域尺度一致。剖面尺度上,矿化地下水对水位以上土体盐分影响厚度为130 cm,随地下水位抬升近地面最大可达140 cm;供水初期盐分自表层至模拟地下水位不断升高,土壤含盐量与剖面深度呈线性正相关,随观测时间增加表层积盐显著,盐分含量表层高,中部低,靠近供水位置再次升高。夹粘土体构型表层含盐量低于均质壤土构型,同一深度夹粘层上方盐分含量相对均质壤土构型降低。（2）多尺度盐渍化驱动机制区域尺度上,各因子对盐渍化驱动力强度为:海岸距离（0.7623）>地下水埋深（0.7266）>累计降水量（0.7241）>地下水矿化度（0.6974）;多因子综合驱动力大于单因子驱动力,其中海岸距离减小、地下水埋深变浅、矿化度增加和累计降水量减少是区域盐渍化加重的重要驱动力;利用随机森林建立的盐渍化驱动模型精度最高,R²达到0.9541。地类尺度上,不同地类盐渍化驱动力发生显著变化,各区域因子对耕地盐渍化驱动力强度为:黄河径流距离（0.6323）>海岸距离（0.6277）>累计降水量（0.5933）>地下水埋深（0.5776）,特征因子中灌排设施有效抑制耕地盐渍化水平;各区域因子对荒地盐渍化驱动力强度为:累计降水量（0.8286）>地下水埋深（0.8216）>海岸距离（0.7895）>地下水矿化度（0.7625）,特征因子中,微地形对荒地盐渍化影响最大;针对不同地类构建的驱动模型精度较高,耕地盐渍化驱动模型R²为0.8792,荒地盐渍化驱动模型R²为0.5266。剖面尺度上,各因子对剖面盐分分布驱动力强度为:地下水矿化度（0.6809）>地下水水位（0.3924）>夹粘层厚度（0.3268）>夹粘层深度（0.2925）。地下水矿化度、夹粘层深度与剖面电导率呈线性正相关,地下水埋深、夹粘层厚度与剖面电导率呈线性负相关。表层干草覆盖度与蒸发量呈线性负相关,通过蒸发量影响土体积盐。利用多元线性回归建立剖面尺度盐渍化驱动模型精度较高,R²为0.9517。（3）盐渍化预测预警利用多尺度驱动模型实现了研究区不同尺度下秋季盐渍化的预测,区域盐渍化预测结果表明,秋季区域盐渍化水平增加,其中非盐渍化、轻度盐渍化面积共减少17.3%,中度和重度盐渍化土壤分别增加10.3%和3.45%;耕地盐渍化预测结果表明,东部盐渍化明显加重,面积呈南北条带状增加;剖面预测结果显示土壤含盐量随地下水位波动上升,在9月达到峰值。基于预测结果,针对区域土地生态和作物产量实现预警,区域预警结果表明:东北部土壤盐渍化预警级别较高,各预警等级分布范围自东向西平移且扩大;作物产量影响预警结果显示,棉花在秋季返盐时红色预警面积较小,小麦红色预警面积中等,玉米红色预警面积最广。本文通过对黄河三角洲濒海区土壤盐渍化的系统性、一体化研究,较好摸清了土壤盐渍化的多尺度变异特征及驱动机制,完成了剖面尺度土壤水盐运移过程的细致刻画,较好揭示了土体内不同驱动力影响下土壤水盐动态的控制机理,为濒海盐渍区水盐调控以及盐渍土利用管理提供了科学依据。更多还原

【Abstract】 The Yellow River Delta is a typical area of soil salinization in China’s coastal areas.Soil salinization is the main factor restricting the sustainable development of the region.Exploring its salinization variation characteristics and driving mechanism,and then realizing the prediction,prevention and early warning management of salinization is the premise of rational use of salinized soil,which is of great significance to ensure food security and achieve sustainable agricultural development in the Yellow River Delta.This study took Kenli District,a typical area of the Yellow River Delta,as the research area,and explored it in three scales:region,land type and profile.The regional scale corresponded to the whole study area.The land type scale was divided into cultivated land and wasteland.The profile scale was based on the soil configuration.In terms of regional and land type scale,this study conducted field sampling,used spatial interpolation to analyze the spatial variation law of regional and land type scale salinization,extracted the spatial distribution data of factors such as climate and elevation,screened and determined the main influencing factors,used regression analysis and geographic detector methods to analyze the driving force,explored its driving mechanism,and adopted random forest to build the driving model.In terms of the profile scale,the indoor soil column test method was used to find out the salt distribution law on the vertical scale,to explore the driving mechanism of soil configuration,groundwater salinity,groundwater level and coverage on the profile water and salt transport,and to build the profile scale salinization driving model by using multiple linear regression.Besides,the spatial distribution of soil salinization in autumn was predicted based on the driving model and historical data at various scales,and the regional land ecological use and crop yield are taken as the basis for early warning classification,so as to realize multi-scale salinization early warning and provide theoretical basis and technical support for the prevention and treatment of salinization disasters.The main results are as follows:（1）Multi-scale spatial variation of salinityThe spatial distribution of salt in the Regional Scale was the distribution pattern of high in the East and low in the west,high in the north and low in the south.From the southwest inland to the Northeast coastal area,the soil salt content gradually increased,the soil salinization gradually increased,and there was an obvious difference near the runoff of the Yellow River,which decreased locally.Land type scale:The salinization level of cultivated land was low,and most areas were 1.5-5.0 g·kg^-1.The overall distribution was light and moderate salinization from the west to the Huanghekou Town,and the eastern coastal area was distributed with strip-shaped severe salinization along the Northeast southwest parallel coastline.The soil salinity level of wasteland increased from west to East and from north to south,and its spatial distribution of salinity was consistent with the regional scale.Profile scale:The maximum influence thickness of mineralized groundwater on soil salinity above the water level was 130cm,and it could reach 140 cm near the ground.At the initial stage of water supply,the salt content in the soil raised from the surface to the simulated groundwater level,and the soil salt content had a linear positive correlation with the profile depth.With the increase of observation time,the salt accumulation in the surface was significant,the salt content in the surface was high,the middle was low,and it increased again near the water supply location.The salt content in the surface layer of the clay layer soil configuration was lower than that in the homogeneous loam configuration,and the salt content above the clay layer was lower than that below.（2）Driving mechanism of multi-scale salinizationThe driving forces of regional scale factors on salinization are as follows:coastal distance（0.7623）>Groundwater Depth（0.7266）>Cumulative Precipitation（0.7241）>Groundwater Salinity（0.6974）.The multi factor comprehensive driving force was greater than the single factor driving force,in which the core driving force of regional salinization was the decrease of coastal distance,the shallower underground depth,the increase of groundwater salinity and the decrease of cumulative precipitation.Meanwhile,the salinization driving model based on random forest had the highest accuracy,and R² reached 0.9541.The driving forces of different land types changed significantly.The driving forces of various factors of cultivated land on salinization are as follows:the Yellow River（0.6323）>Coastal Distance（0.6277）>Cumulative Precipitation（0.5933）>Groundwater Depth（0.5776）.Irrigation and drainage facilities in the characteristic factors effectively inhibited the salinization level of cultivated land.The driving force intensity of various factors of wasteland on salinization was:Cumulative Precipitation（0.8286）>Groundwater Depth（0.8216）>Coastal Distance（0.7895）>Groundwater Salinity（0.7625）.Among the characteristic factors,micro terrain had the greatest impact on wasteland salinization.The driving model constructed for different land types had high accuracy,R² of cultivated land salinization driving model was0.8792,and R² of wasteland salinization driving model was 0.5266.At the profile scale,the driving forces of various factors on the profile salinity distribution are as follows:Groundwater Salinity（0.6809）>Groundwater Level（0.3924）>Clay Layer Thickness（0.3268）>Clay Layer Depth（0.2925）.Beyond that,the salinity of groundwater and the depth of intercalated layer were positively correlated with the conductivity of the profile,while the depth of groundwater and the thickness of intercalated layer were negatively correlated with the salinity of the profile.There was a linear negative correlation between surface hay coverage and evaporation,which affected soil volume salt by affecting evaporation.It had a high accuracy to establish profile scale salinization driving model using multiple linear regression,and R² was 0.9517.（3）Salinization prediction and early warningThe multi-scale driving model was used to predict the autumn salinization in the study area at different scales.According to the results,the regional salinization level increased in autumn,in which the non-salinization and mild salinization areas decreased by 17.3%,and the moderate Salinization Soil and severe Salinization Soil increased by 10.3%and 3.45%respectively.As shown by the prediction result of cultivated land,the salinization in the East was obviously aggravated,and the area increased in a north-south strip.According to the profile prediction results,the soil salt content increased with the fluctuation of water level and reached the peak in August.Based on the prediction results,the early warning for regional land ecological use and crop yield was realized.As shown by the results,the early warning level of soil salinization in the northeast of the region was high,and the distribution range of each early warning level shifted and expanded from east to west.The early warning results of crop yield impact reveal that the red early warning area of cotton in autumn was small,the red early warning area of winter wheat was medium,and the red early warning area of corn was the widest.Through the systematic and integrated study of soil salinization in the coastal area of the Yellow River Delta,this paper better understands the multi-scale variation characteristics and driving mechanism of soil salinization,completes the detailed description of soil water and salt transport process at the profile scale,better reveals the control mechanism of soil water and salt dynamics under the influence of different driving forces in the soil,and provides a scientific basis for water and salt regulation and saline soil utilization management in the coastal saline area.更多还原