基于GIS的多污染因子复合地面浓度场模拟与评价

针对工业大气多因子复合作用的污染物连续地面浓度分布和评价问题,提出了一种基于GIS的地面浓度场模拟和综合评价的新方法,该方法将大气污染物扩散高斯模式和大气质量综合污染指数评价模型在GIS中有机集成,实现了污染物分布空间信息的显示和分析。通过将此方法应用于某一化工厂污染源不同气象条件下污染扩散分析计算,结果表明此种方法简单实用,特别是对多污染因子复合作用的综合评价模型具有任意空间分辨精度,可以满足不同分级需求。     

基于深度学习的高分辨率遥感影像土地覆盖自动分类方法

土地覆盖变化是全球变化研究的核心,而精准分类是开展土地覆盖变化研究的基础。高分辨率遥感卫星技术的快速发展对地表分类的速度和精度提出了双重挑战,近年来人工智能等新技术的发展为图像自动分割提供了实现途径,而以卷积神经网络为代表的深度学习方法在遥感图像分类领域也具有独特的优势。为对比深度学习模型设计对高分辨率图像分类结果的影响,本文以郑州市2019年高分1号影像作为输入,对比研究了基于UNet模型改进的4种不同深度学习网络模型在高分辨率影像土地覆盖自动分类应用中的差异,探讨了残差网络、模型损失函数、跳层连接和注意力机制模块等编码和解码设定对于分类精度的影响机制。研究发现：同时加入多尺度损失函数、跳层连接和注意力机制模块的MS-EfficientUNet模型对郑州市土地覆盖分类结果最优,基于像元评价的整体分类精度可达0.7981。通过在解码器中引入多尺度损失函数可有效提高林地、水体和其他类别的分类精度;而对编码器进行改进,加入跳层连接和注意力机制可进一步提高草地、水体和其他类别地物的分类精度。研究结果表明,深度学习技术在高分辨率遥感影像自动分类中具有潜在应用价值,但分类结果精度的进一步提高和多级别大范围的精细分类方法仍是下一步研究的重点。     

大气负荷对新疆地区地壳形变和地面重力变化的影响

基于ECMWF全球大气模型和中国地面气候资料日值数据集中新疆及周边的气象站气压数据,根据弹性地壳形变理论,采用移去-恢复法,利用负荷格林函数法和球谐函数法,计算新疆及周边区域2011～2015年大气负荷变化对地壳垂直、水平形变的影响,同时采用大气导纳方法计算大气变化对地面重力的影响。结果表明,大气负荷对新疆地区垂直形变的影响达到cm级,对地面重力的影响可达10μGal;大气负荷影响具有明显的季节性。     

基于移去-恢复法的区域大气负荷影响精化——以滇西地区为例

传统的大气负荷改正根据全球大气压模型计算获取,但全球大气压模型中、长波信号占优,缺乏突显区域特征信息的短波信号,而在研究高精度的区域大气负荷影响时,空间信息中的短波信号不可忽视。为此,根据地球重力场理论,引入移去-恢复思想,结合ECMWF提供的全球大气压变化数据和区域高分辨率大气压变化数据（CLDAS-V2.0）,基于负荷球谐系数和区域负荷格林函数方法得到大气压变化对滇西地区的负荷影响。结果表明,得到的大气负荷形变场在空间和时间分变率上均有一定提高。利用移去-恢复法计算区域大气压负荷影响,可为固体地球形变和CORS站时序分析提供参考。     

大气负荷变化对CORS位置影响分析与其移除研究

利用全球大气模型和CORS位置时间序列数据,基于负荷形变理论,采用球谐系数法计算大气负荷变化对CORS位置地壳形变的影响。结果显示,对研究区CORS位置垂直方向影响为-15.9~13.2 mm,水平形变影响为0~2.7 mm。根据移除大气负荷对CORS影响前后序列的RMS值变化可知,移除大气负荷影响对CORS时间序列水平方向改善较小,垂直方向改善明显;根据最小二乘方法计算研究区大气负荷变化对地壳垂直方向的形变导纳值（回归系数）为-0.7~-0.43 mm/hPa。本文研究方法和成果对分离修正CORS中大气负荷影响具有借鉴意义。     

一种自适应时间步长的场地地表污染扩散元胞自动机仿真模型

针对传统扩散模型难以动态模拟地表污染物时空不均匀扩散过程的问题,本文提出一种基于元胞自动机模型的污染物地表扩散仿真模型,在综合考虑地表高差及粗糙度对污染物扩散过程影响的基础上,确定了不规则污染场地的元胞边界条件、划分了元胞空间、提出了一种降雨和非降雨条件下污染物扩散流速计算方法,基于分子扩散建立了地表污染扩散模型演变规则。为更好模拟地表污染物扩散情况,本文提出了一种污染物随坡度和质量衰减的元胞自适应时间步长调整算法,该算法能够动态调整元胞自动机的时间步长,防止固定时间步长在污染物快速扩散时错过细节,而缓慢扩散时消耗计算资源。实验设计了降雨和非降雨两种情形对污染物随时间扩散的过程进行仿真与分析。实验结果表明,不同下垫面对污染物扩散速度有很大的影响,污染物在糙率为0.012的水泥地表上的扩散速度约为其在糙率为0.035的一般性土壤地表上的2.7倍;降雨强度和时长能够加快污染物的扩散,且扩散速度随着降雨曲线变化而改变,并在雨强峰值附近达到最大;污染物扩散服从坡度分布特征,且随着时间变化,高污染区域范围和污染物浓度差异渐渐变小,并在一段时间后,浓度变化渐渐趋于平稳;自适应时间步长演变算法能够较好地体现一次演变过程中污染物扩散在不同邻域元胞方向上的细微时间差异,提高污染物时空分布的计算精度。     

基于Android终端和UC Map平台的移动GIS开发与应用

面向服务的移动GIS开发与应用逐渐成为移动GIS领域研究热点和发展趋势。本文首先介绍了Android终端和UCMap开发平台等相关知识;其次详细研究了基于移动GIS应用的Android平台开发环境搭建和基于UCMap平台的Android终端应用二次开发模块;最后提出了一种基于移动GIS应用的UCMap平台开发与应用模式,并在Android平台上进行了相关技术验证,对移动GIS应用研究具有较高的技术参考价值。     

不同海潮模型对中国沿海区域海潮负荷位移改正影响分析

选取FES2004、EOT11a、TPXO7.2和Chinasea2010等4个海潮模型,首先通过矢量差评价不同海潮模型的潮波参数差异,然后利用中国大陆构造环境监测网络23个沿海测站的GNSS连续观测数据,通过计算不同海潮模型改正前后GNSS时间序列的wRMS,评价基于不同海潮模型的海潮负荷位移改正的有效性。结果表明,不同海潮模型的潮波参数存在mm级差异,且沿海区域差异远比内陆区域显著。此外,海潮负荷改正对中国沿海区域测站坐标时间序列wRMS影响较大,大部分测站坐标时间序列进行海潮负荷改正后,wRMS减小10%~70%;改正后wRMS改善幅度与测站所属区域有关,东海沿岸测站wRMS的改善较渤海、黄海和东海沿岸测站更显著,可减小50%以上;wRMS改善程度的模型间差异为1%~2%,其中基于FES2004模型的改正对序列wRMS的影响最大,可达67.5%。     

新西兰海潮负荷位移建模精度分析

为评价新西兰海潮负荷位移建模精度,利用新西兰189个GPS站11 a的实测数据,基于静态精密单点定位测定8个半日潮波及周日潮波的海潮负荷位移参数,并将其与7种全球海潮模型及4种地球模型计算的海潮负荷位移改正值进行比较。结果表明：1）TPXO7.2模型负荷位移改正值与GPS解算的海潮负荷位移参数最符合,M2、N2、O1和Q1潮波均方根误差在水平方向小于0.5 mm,垂直方向小于0.7 mm;2）不同地球模型对确定海潮负荷位移的影响主要体现在M2和N2潮波;3）GPS估值和海潮模型值之间的残差矢量呈现出大小及方向上的区域一致性,部分站点异常的残差值可能反映出当前SNREI地球模型的缺陷。     

确定性系数与随机森林模型在云南芒市滑坡易发性评价中的应用

编制科学的滑坡易发性分区图,可以有效降低灾害带来的损失。以云南省芒市为研究区,利用确定性系数模型（certainty factor,简称CF）方法计算各个因子的敏感值,作为随机森林（random forests,简称RF）的分类数据,选取合适的训练数据和最优化的模型参数进行模型预测,从而对研究区进行滑坡易发性评价分区。采用频率比方法将连续性因子离散化,从而通过确定性系数计算因子不同区间的滑坡易发性,同时利用CF先验模型,对研究区负样本进行选取。通过计算袋外误差得到最优化的RF参数,随后利用RF模型对研究区模型进行训练及预测。绘制ROC曲线和三维遥感影像对预测模型结果分别进行定量和定性评价,结果表明,所得到的模型精度为91%,优于随机抽样得到的结果。最后,采用平均基尼不纯度减少和平均准确度下降两种计算方法计算、评价了研究区各个因子的重要性。基于以上对研究区进行的滑坡易发性评价结果,可以为该区灾害风险评估和管理提供依据。      

城市暴雨内涝模拟模型优化与精度验证

本文采用“Vegetation-Impervious Surface-Soil”模型和线性光谱混合分解方法,获取像元中不透水面、植被、土壤覆盖信息,用于计算SCS模型产流参数综合CN（Curve Number）值;基于土地利用类型,采用经验值与数值实验逐步求精相结合的方法,确定水动力汇流模型参数曼宁系数,并用实测积水数据验证两次参数修正的模拟效果。以上海中心城区为例进行验证,研究结果表明：①将采用V-I-S模型得到的不透水面、植被、土壤信息设定CN值,能够降低积水分布的极值化现象,提高SCS产流模型产流量和产流分布精度;②采用经验法和数值模拟逐步求精法,按土地利用类型设定曼宁系数,使各时段最大积水深度高于原模型,说明曼宁系数是汇流模型的敏感参数。     

Application of SCS model in estimation of runoff from small watershed in Loess Plateau of China

Soil Conservation Service （SCS） model, developed by U. S. Soil Conservation Service in 1972, has been widely applied in the estimation of runoff from an small watershed. In this paper, based on the remote sensing geo-information data of land use and soil classification all obtained from Landsat images in 1996 and 1997 and conventional data of hydrology and meteorology, the SCS model was investigated for simulating the surface runoff for single rainstorm in Wangdonggou watershed, a typical small watershed in the Loess Plateau, located in Changwu County of Shaanxi Province of China. Wangdonggou watershed was compartmentalized into 28 sub-units according to natural draining division, and the table of curve number （CN） values fitting for Wangdonggou watershed was also presented. During the flood period from 1996 to 1997, the hydrograph of calculated runoff process using the SCS model and the hydrograph of observed runoff process coincided very well in height as well as shape, and the model was of high precision above 75%. It is indicated that the SCS model is legitimate and can be successfully used to simulate the runoff generation and the runoff process of typical small watershed based on the remote sensing geo-information in the Loess Plateau.     

Comparison of SCS-CN Determination Methodologies in a Heterogeneous Catchment

The aim of this study was to assess the runoff amount from a catchment characterized by diverse land uses by using the Soil Conservation Service Curve Number(SCS-CN) method based on Curve Number(CN) defined for dominant homogeneous elementary sub-regions.The calculations employed the SCS-CN method,involving the division of the catchment in two homogeneous parts and determining the runoff amount.The obtained results were compared with the results provided by three other CN determination methods,i.e.the Hawkins function,the kinetics equation,and a complementary error function peak.The catchment is located in a mountain dominated by forest land cover.Empirical CN-Precipitation(CN_(emp)-P) data pairs were analyzed using the mentioned methods,and the highest quality score was achieved from model 1.The results suggest that dividing a catchment into two homogeneous areas and determining their separate CN parameters,used later on to calculate the runoff by means of the presented approach,could be an alternative to the standard methods.The described method is relatively easy,and as it does not require an adoption of numerous parameters,and it can be employed for designing hydraulic facilities.     

A Modified Groundwater Module in SWAT for Improved Streamflow Simulation in a Large,Arid Endorheic River Watershed in Northwest China

Interactions between surface water and groundwater are dynamic and complex in large endorheic river watersheds in Northwest China due to the influence of both irrigation practices and the local terrain. These interactions interchange numerous times throughout the middle reaches, making streamflow simulation a challenge in endorheic river watersheds. In this study, we modified the linear-reservoir groundwater module in SWAT (Soil and Water Assessment Tools, a widely used hydrological model) with a new nonlinear relationship to better represent groundwater processes; we then applied the original SWAT and modified SWAT to the Heihe River Watershed, the second largest endorheic river watershed in Northwest China, to simulate streamflow. After calibrating both the original SWAT model and the modified SWAT model, we analyzed model performance during two periods: an irrigation period and a non-irrigation period. Our results show that the modified SWAT model with the nonlinear groundwater module performed significantly better during both the irrigation and non-irrigation periods. Moreover, after comparing different runoff components simulated by the two models, the results show that, after the implementation of the new nonlinear groundwater module in SWAT, proportions of runoff components changed-and the groundwater flow had significantly increased, dominating the discharge season. Therefore, SWAT coupled with the non-linear groundwater module represents the complex hydrological process in the study area more realistically. Moreover, the results for various runoff components simulated by the modified SWAT models can be used to describe the hydrological characteristics of lowland areas. This indicates that the modified SWAT model is applicable to simulate complex hydrological process of arid endorheic rivers.     

气候变化情景下淮河上游流域氮排放预测研究

淮河流域是水体遭受营养盐污染较严重的地区,本研究选择淮河上游的淮滨流域(淮滨站以上,流域面积1.6万km²)为研究对象,首先构建了淮滨流域SWAT水文水质模型,然后利用2011—2017年淮滨站实测的月径流和月氨氮浓度对SWAT模型进行了校正与验证,最后基于全球气候模型(GCM)气象数据,预测了未来30年(2020—2029年、2030—2039年、2040—2049年)不同气候变化情境(RCP2.6、RCP4.5、RCP6.0、RCP8.5)下的径流、氨氮浓度和非点源总氮负荷。结果发现,径流在校正期和验证期的Nash-Suttcliffe系数均为0.79,氨氮在校正期和验证期的Nash-Suttcliffe系数均高于0.5,表明模型的适用性良好。研究发现本研究区施肥量与土地利用类型是非点源氮负荷空间分异的主导因素。2020—2049年,不同气候变化情景下,本研究区的降水量和气温均为增长趋势。假如保持基准期(2011—2016年)污染排放强度,仅考虑气候变化影响,流域内非点源污染总氮负荷将比基准期最多增加31.8%,流域出水口淮滨站的年均氨氮浓度将最多减小42.6...     

Spatial analysis of the impact of land-use on surface runoff and soil erosion-a case study of western Jilin

The impact of land-use on surface runoff and soil erosion is still poorly understood at basin scale. Thus in the Western Jilin Ecosystem (WJE), surface runoff and soil erosion were measured against identified land-use types in the basin. Due to the spatial nature of the analysis, GIS ArcMap version 9.1 and the WetSpass model were used in the simulation process. In the study, the WetSpass model was extended with the Dynamic Sediment Balance Equation (Ziegler et al., 1997), to simulate and quantify soil erosion. A hypothetical natural grassland scenario was developed for the study area and compared with the present land-use management conditions. The results indicate significant differences in runoff and soil erosion across the different land-use types both within and between the two scenarios. Calculated averages of surface runoff and soil erosion for the present land-use management were 48.03 mm/a and 83.43 kg/(m2·a) respectively. Those for the hypothetical natural grassland scenario were 24.70 mm/a and 78.36 kg/m2·a) . Thus an overall decrease in runoff and soil erosion was observed as the conditions changed from the present land-use management to the hypothetical natural grassland state. Under the present land-use management, urban settlements exhibited the highest surface runoff but one of the least soil erosions, while bare-lands showed the highest soil erosion. It was more generally observed that runoff and erosion varies with vegetation type/density. It was concluded based on the research findings that the present land-use management might not be the best scenario for the ecosystem as it showed increased basin runoff and soil erosion in comparison with the natural grassland vegetation. Since no best scenario was simulated for or advanced in the study, further research to develop a more balanced land management system is thus required. The findings of the study can assist in the identification of vulnerable/fragile ecosystems in the basin and to guide sustainable future planning and development of the basin.     

Probabilistic Properties of a Curve Number:A Case Study for Small Polish and Slovak Carpathian Basins

The proper determination of the curve number(CN) in the SCS-CN method reduces errors in predicting runoff volume. In this paper the variability of CN was studied for 5 Slovak and 5 Polish Carpathian catchments. Empirical curve numbers were applied to the distribution fitting. Next, theoretical characteristics of CN were estimated. For 100-CN the Generalized Extreme Value(GEV) distribution was identified as the best fit in most of the catchments. An assessment of the differences between the characteristics estimated from theoretical distributions and the tabulated values of CN was performed. The comparison between the antecedent runoff conditions(ARC) of Hawkins and Hjelmfelt was also completed. The analysis was done for various magnitudes of rainfall. Confidence intervals(CI) were helpful in this evaluation. The studies revealed discordances between the tabulated and estimated CNs. The tabulated CNs were usually lower than estimated values; therefore, an application of the median value and the probabilistic ARC of Hjelmfelt for wet runoff conditions is advisable. For dry conditions the ARC of Hjelmfelt usually better estimated CN than ARC of Hawkins did, but in several catchments neither the ARC of Hawkins nor Hjelmfelt sufficiently depicted the variability in CN.     

Hydrological Response to Environment Change in Himalayan Watersheds: Assessment from Integrated Modeling Approach

Land use changes such as deforestation,increase in cropping or grazing areas and built-up land, likely modify the water balance and land surface behavior in the Himalayan watersheds.An integrated approach of hydrological and hydraulic modeling was adopted for comparative analysis of hydrological pattern in three Himalayan watersheds i.e.Khanpur,Rawal and Simly situated in the Northern territory of Pakistan.The rainfall-runoff model SWAT- Soil and water assessment tool and Hydro CAD were calibrated for the selected watersheds.The correlation analysis of the precipitation data of two climate stations i.e.Murree and Islamabad, with the discharge data of three rivers was utilized to select best suitable input precipitation data for Hydro CAD rainfall-runoff modeling.The peak flood hydrograph were generated using Hydro CAD runoff to optimize the basin parameters like CN, runoff volume, peak flows of the three watersheds.The hydrological response of the Rawal watershed was studied as a case study to different scenarios of land use change using SWAT model.The scenario of high deforestation indicated a decline of about 6.3% in the groundwater recharge tostream while increase of 7.1% in the surface runoff has been observed under the scenario of growth in urbanization in the recent decades.The integrated modeling approach proved helpful in investigating the hydrological behavior under changing environment at watershed level in the Himalayan region.     

Assessing effects of “source-sink” landscape on non-point source pollution based on cell units of a small agricultural catchment

Ascertaining the relationship between "source-sink" landscape and non-point source(NPS) pollution is crucial for reducing NPS pollution, however, it is not easy to realize this target on cell unit scale. To reveal the relationships between "sourcesink" landscape and NPS pollution based on cell units of a small catchment in the Three Gorges Reservoir Region(TGRR), the runoff and nutrient yields were simulated first by rainfall events on a cell unit scale based on the Annualized AGricultural Non-Point Source Pollution Model(AnnAGNPS). Landscape structure and pattern were quantified with "sourcesink" landscape indicators based on cell units including landscape area indices and locationweighted landscape indices. The results showed that:the study case of small Wangjiagou catchment highlighted a good prediction capability of runoff and nutrient export by the AnnAGNPS model. Throughout the catchment, the spatial distribution trends of four location-weighted landscape indices were similar to the trends of simulated total nitrogen(TN) and total phosphorus(TP), which highlighted the importance of spatial arrangement of "source" and "sink" landscape types in a catchment when estimating pollutant loads. Results by Pearson correlation analysis indicated that the location-weighted landscape index provided a more comprehensive account of multiple factors, and can better reflect NPS-related nutrient loss than other landscape indices applied in single-factor analysis. This study provides new findings for applying the "source-sink" landscape indices based on cell units in small catchments to explain the effect of "source-sink" landscape on nutrient export based on cell unit, and helps improve the understanding of the correlation between "source-sink" landscape and NPS pollution.     

Development of a location-weighted landscape contrast index based on the minimum hydrological response unit

The changing patterns of watersheds in a landscape, driven by human activities, play an important role in non-point source pollution processes. This paper aims to improve the location-weighted landscape contrast index using remote sensing and GIS technology to account for the effects of scale and ecological processes. The hydrological response unit(HRU) with a single land use and soil type was used as the smallest unit. The relationship between the landscape index and typical ecological processes was established by describing the influence of the landscape pattern on non-point source pollution. To verify the research method, this paper used the Yanshi River basin as a study area. The results showed that the relative intensity of non-point source pollution in different regions of the watershed and the location-weighted landscape contrast index based on the minimum HRU can qualitatively reflect the risk of regional nutrient loss.