基于超参数优化CatBoost算法的河流悬浮物浓度遥感反演

悬浮物浓度（TSM）是水生态环境评价的重要参数之一,及时掌握河流悬浮物浓度动态变化信息对于内陆水质监测、水环境治理是十分必要的。本研究基于野外实测光谱和悬浮物浓度数据,筛选与悬浮物浓度高度相关的波段组合反射率作为自变量,基于CatBoost、随机森林和多元线性回归算法构建悬浮物浓度遥感反演模型,采用带交叉验证的网格搜索法分别对CatBoost和随机森林2种机器学习模型进行超参数调优,确定模型最优参数配置,并对比不同模型反演精度,确定最优模型。基于最优模型,利用2019—2020年多时相Sentinel-2 MSI遥感影像,反演闽江下游悬浮物浓度,并分析其时空变化特征。结果表明：(1) b4/b3、(b6-b3)/(b6+b3)、(b4+b8)/b3、(1/b3-1/b4)×b5是MSI反演闽江下游TSM浓度的最佳波段组合反射率;(2)对比其他2种模型,基于超参数优化的CatBoost算法建立的悬浮物反演模型精度最高,其决定系数R2为0.95,均方根误差RMSE和平均绝对百分比误差MAPE分别为15.32 mg/L和19.68%;(3) 2019—2020年闽江下游悬浮物浓度分布“西低东...     

曹妃甸近岸海域垂向悬沙含量遥感反演

利用遥感技术监测水体悬浮泥沙含量是海洋水色遥感的一个重要研究方向。本文以曹妃甸近岸海域4个剖面35个站点开展的现场光谱测量、同步采集的不同深度悬浮泥沙含量样品数据为基础,采用Landsat-5 TM遥感数据,建立水体表、中、底层悬沙含量遥感反演模型,并研究悬沙垂向上的空间分布规律。结果表明,曹妃甸近岸海域泥沙含量在垂向上有明显的相关性,以Landsat-5 TM影像反射率比值（R_TM3/R_TM2）为因子,构建优化的悬浮泥沙含量遥感反演模型,各层模型平均相对误差都在30%以内,表层和中层的平均绝对误差在6 mg/L以下,均方根误差小于10 mg/L,底层的精度略低于表层和中层。研究结果为进一步研究海洋环境泥沙运移规律、优化水动力作用下泥沙运移模型提供支持。     

Suspended sediment dynamics in the meltwater of Chhota Shigri glacier,Chandra basin,Lahaul-Spiti valley,India

Present study shows suspended sediment dynamics in the meltwater of Chhota Shigri glacier, Himachal Pradesh, India for different melt seasons during the period 2011-2014. Maximum suspended sediment concentration in the meltwater was found during the month of July 2011, 2012 and 2014 constituting to 55.2%, 48.3% and 46.9%, respectively. Whereas in 2013, maximum suspended sediment concentration was observed in August accounting for 46.1% of the total. On the other hand, maximum suspended sediment load was monitored in the month of July 2011, 2012 and 2014 constituting 59.5%, 63% and 55.7% of the total, respectively. Whereas in 2013, maximum suspended sediment load was observed in the month of August accounting for 49.8% of the total suspended sediment load. Annual distribution of suspended sediment concentration (SSC) and suspended sediment load (SSL) in the Chhota Shigri glacier shows higher value of SSC and SSL during the study period 2012 and 2013, which may be due to the presence of high glacial runoff and negative mass balance of the studied area during these time periods. Marked diurnal variation has been observed in the SSC of meltwater. Strong correlation was observed between SSC and SSL with discharge. On the other hand, SSC and SSL also showed strong exponential correlation with air temperature of the studied area. Sediment yield from the catchment of Chhota Shigri glacier is high during the peak melt season (July and August) and low during the late melt season (September and October). The average value of erosion rate for Chhota Shigri glacier basin during the study period 2011-2014 was calculated to be 1.1 mm/yr, which is lower than the average erosion rate of other Himalayan glaciers such as Rakiot, Chorabari and Gangotri glaciers, which may be caused by its geological setting containing high erosion resistant rocks such as granite, granite gneiss and porphyritic granite.     

Estimation of suspended sediment concentrations using Pulse-coherent Acoustic Doppler Profiler （PCADP）

The objective of the study is to investigate the suitability of using Pulse-coherent Acoustic Doppler Profiler (PCADP) to estimate suspended sediment concentration (SSC). The acoustic backscatter intensity was corrected for spreading and absorption loss, then calibrated with OBS and finally converted to SSC. The results show that there is a good correlation between SSC and backscatter intensity with R value of 0.74. The mean relative error is 22.4%. Then the time span of little particle size variation was also analyzed to exclude the influence of size variation. The correlation coefficient increased to 0.81 and the error decreased to 18.9%. Our results suggest that the PCADP can meet the requirement of other professional instruments to estimate SSC with the errors between 20% and 50%, and can satisfy the need of dynamics study of suspended particles. Supported by National High Technology Research and Development Program of China (863 Program, No. 2008AA09Z113)     

The suspended sediment concentration distribution in the Bohai Sea, Yellow Sea and East China Sea

The distribution of the suspended sediment concentration (SSC) in the Bohai Sea, Yellow Sea and East China Sea (BYECS) is studied based on the observed turbidity data and model simulation results. The observed turbidity results show that (i) the highest SSC is found in the coastal areas while in the outer shelf sea areas turbid water is much more difficult to observe, (ii) the surface layer SSC is much lower than the bottom layer SSC and (iii) the winter SSC is higher than the summer SSC. The Regional Ocean Modeling System (ROMS) is used to simulate the SSC distribution in the BYECS. A comparison between the modeled SSC and the observed SSC in the BYECS shows that the modeled SSC can reproduce the principal features of the SSC distribution in the BYECS. The dynamic mechanisms of the sediment erosion and transport processes are studied based on the modeled results. The horizontal distribution of the SSC in the BYECS is mainly determined by the current-wave induced bottom stress and the fine-grain sediment distribution. The current-induced bottom stress is much higher than the wave-induced bottom stress, which means the tidal currents play a more significant role in the sediment resuspension than the wind waves. The vertical mixing strength is studied based on the mixed layer depth and the turbulent kinetic energy distribution in the BYECS. The strong winter time vertical mixing, which is mainly caused by the strong wind stress and surface cooling, leads to high surface layer SSC in winter. High surface layer SSC in summer is restricted in the coastal areas.     

Factorial analysis of annual erosion-accretion cycles of tidal flats in the front area of the southern Changjiang river delta

ＦＡＣＴＯＲＩＡＬＡＮＡＬＹＳＩＳＯＦＡＮＮＵＡＬＥＲＯＳＩＯＮＡＣＣＲＥＴＩＯＮＣＹＣＬＥＳＯＦＴＩＤＡＬＦＬＡＴＳＩＮＴＨＥＦＲＯＮＴＡＲＥＡＯＦＴＨＥＳＯＵＴＨＥＲＮＣＨＡＮＧＪＩＡＮＧＲＩＶＥＲＤＥＬＴＡＹａｎｇＳｈｉｌｕｎ（杨世伦）Ｃｈｅ...     

Suspended sediment concentration mapping based on the MODIS satellite imagery in the East China inland,estuarine, and coastal waters

The purpose of this research is to improve the retrieval accuracy for the suspended sediment concentration(SSC) from in situ and satellite remote sensing measurements in turbid East China estuarine and coastal waters. For this aim, three important tasks are formulated and solved: 1) an estimation of remote-sensing reflectance spectra R_(rs)(λ) after atmospheric correction; 2) an estimation of R_(rs)(λ) from the radiometric signals above the air-water surface; and 3) an estimation of SSC from R_(rs)(λ). Six different models for radiometric R_(rs)(λ) determination and 28 models for SSC versus R_(rs)(λ) are analyzed based on the field observations made in the Changjiang River estuary and its adjacent coastal area. The SSC images based on the above-mentioned analysis are generated for the area.     

Remote sensing retrieval of surface suspended sediment concentration in the Yellow River Estuary

Accurate assessment of surface suspended sediment concentration(SSSC) in estuary is essential to address several important issues: erosion, water pollution, human health risks, etc. In this study, an empirical cubic retrieval model was developed for the retrieval of SSSC from Yellow River Estuary. Based on sediments and seawater collected from the Yellow River and southeastern Laizhou Bay, SSSC conditions were reproduced in the laboratory at increasing concentrations within a range common to field observations. Continuous spectrum measurements of the various SSSCs ranging from 1 to 5700 mg/l were carried out using an Ava Field-3 spectrometer. The results indicated the good correlation between water SSSC and spectral reflectance(Rrs) was obtained in the spectral range of 726–900 nm. At SSSC greater than 2700 mg/L, the 740–900 nm spectral range was less susceptible to the effects of spectral reflectance saturation and more suitable for retrieval of high sediment concentrations. The best correlations were obtained for the reflectance ratio of 820 nm to 490 nm. Informed by the correlation between Rrs and SSSC, a retrieval model was developed(R2 = 0.992). The novel cubic model, which used the ratio of a near-infrared(NIR) band(740–900 nm) to a visible band(400–600 nm) as factors, provided robust quantification of high SSSC water samples. Two high SSSC centers, with an order of 103 mg/l, were found in the inversion results around the abandoned Diaokou River mouth, the present Yellow River mouth to the abandoned Qingshuigou River mouth. There was little sediment exchange between the two high SSSC centers due to the directions of the residual currents and vertical mixing.     

A method for simulating sediment incipient motion varying with time and space in an ocean model(FVCOM):development and validation

We modified the sediment incipient motion in a numerical model and evaluated the impact of this modification using a study case of the coastal area around Weihai, China. The modified and unmodified versions of the model were validated by comparing simulated and observed data of currents, waves, and suspended sediment concentrations(SSC) measured from July 25~(th) to July 26~(th), 2006. A fitted Shields diagram was introduced into the sediment model so that the critical erosional shear stress could vary with time. Thus, the simulated SSC patterns were improved to more closely reflect the observed values, so that the relative error of the variation range decreased by up to 34.5% and the relative error of simulated temporally averaged SSC decreased by up to 36%. In the modified model, the critical shear stress values of the simulated silt with a diameter of 0.035 mm and mud with a diameter of 0.004 mm varied from 0.05 to 0.13 N/m~2, and from 0.05 to 0.14 N/m~2, respectively, instead of remaining constant in the unmodified model. Besides, a method of applying spatially varying fractions of the mixed grain size sediment improved the simulated SSC distribution to fit better to the remote sensing map and reproduced the zonal area with high SSC between Heini Bay and the erosion groove in the modified model. The Relative Mean Absolute Error was reduced by between 6% and 79%, depending on the regional attributes when we used the modified method to simulate incipient sediment motion. But the modification achieved the higher accuracy in this study at a cost of computation speed decreasing by 1.52%.     

Mechanism of Phase Lag Between Current Speed and Suspended Sediment: Combined Effect of Erosion,Deposition, and Advection

To retrieve and explain the phase lag between current speed and suspended sediment concentration(SSC), erosion, deposition, and advection were isolated as primary processes of sediment movement in a three-dimensional model. The response time was proved to be one of the reasons for the phase lag, as time is needed for suspension to diffuse from bottom to surface. A fitted Shields diagram was introduced into the model to reflect the relationship between SSC and shear stress, between shear stress and critical shear stress, as well as between SSC and critical shear stress for erosion. It takes some time for shear stress to increase to the critical value after high or low tide, and this was proved to be an important contributor to the phase lag. Overall, the variation of vertically integrated SSC is influenced by erosion mass flux, deposition mass flux, and advection flux. The phase pattern of erosion mass flux is consistent with the pattern of current if there was no wave action. However, phase difference is produced by the influence of deposition mass flux and advection. In this study, SSC peak/trough mostly occurred near the moment erosion mass flux approximately equaled deposition mass flux and would be impacted by advection. The time required for instantaneous variation of suspension to get to 0 after current peak/trough represents the phase lag between current speed and SSC.     

A preliminary study on suspended sediment concentration measurements using an ADCP mounted on a moving vessel

ＩＮＴＲＯＤＵＣＴＩＯＮＳｕｓｐｅｎｄｅｄｓｅｄｉｍｅｎｔｔｒａｎｓｐｏｒｔｐｌａｙｓａｎｉｍｐｏｒｔａｎｔｒｏｌｅｉｎｔｈｅｃｏａｓｔａｌｅｎｖｉｒｏｎｍｅｎｔｅｖｏｌｕｔｉｏｎ.Ｔｈｕｓ,ｔｈｅｏｂｓｅｒｖａｔｉｏｎｏｆｓｕｓｐｅｎｄｅｄｓｅｄｉｍｅｎｔｓｉｓｏｆｉｎｔｅｒｅｓｔｔｏｃｏａｓｔａｌｓｃｉｅｎｔｉｓｔｓ,ｅｎｇｉｎｅｅｒｓ,ｃｏｎｓｅｒｖａｔｉｏｎｉｓｔｓａｎｄｐｌａｎｎｅｒｓ,ｉｎｔｈｅｉｒａｃｔｉｖｉｔｉｅｓｏｆｈａｒｂｏｕｒｃｏｎｓｔｒｕｃｔｉｏｎ,ｌａｎｄｒｅｃｌａｍａｔｉｏ…     

Factorial analysis of annual erosion-accretion cycles of tidal flats in the front area of the southern Changjiang river delta

Based on surveyed data from seven coastal sections and the collected data of wind, sea level, tide, nearshore suspended sediment concentration and river flux from adjacent stations, this paper deals with regressive correlation between monthly average flat elevation and monthly average figures of the influential factors. All sections except one which is located within the river mouth showed negative correlation between flat elevation and sea level and between flat elevation and tidal range, with correlation coefficients being −0.53 – −0.91 (−0.77 on the average) in the former condition and −0.56 – −0.97 (−0.80 on the average) under the latter. Each of the sections with available suspended sediment concentration (SSC) data shows a positive correlation between flat elevation and SSC, with correlative coefficients being 0.35 – 0.97 (0.66 on the average). Only two sections (one in the Changjiang River Estuary and the other in the Hangzhou Bay) which are similar to beaches in sediment grain-size and slope gradient showed a negative correlation between flat elevation and onshore wind frequency and between flat elevation and average wind velocity, with correlative coefficients being respectively −0.57 and −0.69 (−0.63 on the average) in the former situation and −0.61 and −0.75 (−0.68 on the average) in the latter. Other sections did not show uniform relationship between flat elevation and wind conditions. Due to local marine factors the nearshore SSC in the studied area is negatively correlated with the Changjiang River sediment flux (r=−0.78), which results in false negative correlation between flat elevation and river sediment flux. The paper also gives sediment dynamic and morphodynamic explanation for the above correlations. Sea level rise results in the spread of breaker zone from subtidal area to intertidal area and then increases the intertidal water energy. The larger the tidal range, the stronger the tide currents and the easier for the flat to be eroded. The higher the SSC of flood water, the easier for the sediment to deposit down. Because of correlations among the influential factors, it is difficult to give the prime one which dominates the erosion and accretion processes in tidal flats.     

基于HJ-1A CCD1数据的台湾相思树叶面积指数反演

基于HJ-1A CCD1环境卫星数据,以福建沿海地区普遍分布的台湾相思树为研究对象,利用回归分析法(NDVI、OSAVI、EVI、HJVI)和PROSAIL辐射传输模型,构建台湾相思树LAI反演模型。同时,利用同步野外地面实测数据,将模型估算LAI值与实测LAI值进行对比。结果表明：(1)相比归一化植被指数NDVI、优化土壤调节指数OSAVI和增强型植被指数EVI 3种常用植被指数,引入修正大气、土壤背景影响的蓝、绿波段的环境植被指数HJVI来反演相思树LAI具有更高的精度(R²=0.7344,RMSE=0.1421);(2)本研究所选4种植被指数构建的最优反演模型均为非线性模型,其中,环境植被指数HJVI反演LAI最优模型为幂函数模型,表明相思树LAI与植被指数之间呈非线性变化;(3)PROSAIL辐射传输模型法比回归分析法反演相思树LAI的精度有较大提高(R²=0.7903,RMSE=0.1303),可见PROSAIL模型法构建反演模型能更好地反演相思树LAI。     

Application of MODIS data in monitoring suspended sediment of Taihu Lake,China

Application of MODIS in ocean color is mainly based on bands 8-16 with the spatial resolution of 1 000 m.This spatial resolution,however,can not meet the application demand of inland waters where the areas are relatively small.With the assumption of the black water at shortwave infrared(SWIR) wavelengths(>1 000 nm),we first propose an atmospheric correction method for bands 1 and 2 with their spatial resolution of 250 m,and we then establish a quantitative retrieval model for suspended sediment concentratio...     

Determining suspended sediment concentration and settling velocity from PC-ADP measurements in the Beibu Gulf, China

Modeling sediment transport depends on several parameters, such as suspended sediment concentration (SSC), shear stress, and settling velocity. To assess the ability of Pulse-Coherent Acoustic Doppler Profiling (PC-ADP) to non-intrusively quantify spatial and temporal SSC and settling velocity at seabed, a field experiment was conducted in the Beibu Gulf (Tonkin Gulf), in the South China Sea. The spatial profiles and temporal variations in SSC at 1 m above bottom were derived from PC-ADP acoustic backscatter intensity determinations after being calibrated with the optical backscatter sensor (OBS) measurements at the same elevation. The PC-ADP and OBS results agreed well. The temporal settling velocity obtained from Rouse profiles agreed well with the Soulsby formula based on size information by LISST (laser in situ scattering and transmissometry). Tides and tidal currents are diurnal in the gulf. SSC increased with increasing ebb and flood flow, and it rapidly decreased with the increase of distance from the seabed. The maximum SSC at 0.16 m and 1.3 m above bottom reached 816 mg/L and 490 mg/L during spring tides, respectively. The sediments consisted of mineral particles 23-162 μm in diameter and 0.05-2.04 cm/s in settling velocity. Generally, both the SSC and settling velocity followed variations in the bottom friction. Results suggest that PC-ADP is able to provide reasonable SSC and settling velocity measurements of both profiles and time series for a long study period.     

混合光谱分解算法的太湖水体组分浓度反演分析

二类水体组份的遥感定量反演一直是水色遥感的难点和热点问题,原因在于其水体组分(纯水、叶绿素、悬浮物及CDOM)之间复杂的相互作用.本文引入光谱分解算法,通过Hydrolight软件模拟叶绿素、悬浮物和CDOM的标准反射率光谱,解决光谱分解算法中“纯端元”难以获取的问题.在此基础上建立了二类水体组分光谱分解反演模型.模型...     

基于自组织网络的黄河口浑浊模式研究

传统浑浊度时空变化模式研究依靠野外观测实验,需要投入大量人力物力,模型适用范围亦非常有限。本文利用自组织网络（SOM）,直接从长时序遥感影像中提取典型浑浊模式,分析浑浊度年内、年际变化特征。以黄河河口附近海域为研究区,提取出近15年的6类典型浑浊模式。典型特征显示,研究区内主要有2个浑浊区,位于渤海湾西部和南部,以及河口外和莱州湾西北部;6类模式中四类以年为周期过渡更替,冬春季浑浊度较高,夏秋季浑浊度较低;多年浑浊模式逐渐由中浑浊向清澈模式变化,整体浑浊度有降低趋势。浑浊水体分布主要受河口潮流、环流等海洋动力和风浪影响,结合研究区气象观测数据分析,海面风浪变化是造成浑浊模式更替的主要原因,黄河入海泥沙影响范围仅局限于河口口门周边。利用统计参数分析和2007年各月悬浮泥沙浓度反演结果比较,评价SOM分类效果,结果表明SOM提取的模式间具有显著差异,一定程度上能够代替经验模型反映区域浑浊特征。SOM神经网络能够从长时序遥感影像中直接提取浑浊水体典型分布模式,分析海岸带地区水体浑浊度变化的时空特征,对了解复杂水体泥沙输运及优化水资源利用具有重要应用价值。     

Filling of the Three Gorges Reservoir to the 135-m Level： Instant Effects on the Yangtze Discharge and Suspended Sediment Concentration Entering the Estuary

Via the valuable opportunity of the Three Gorges Reservoir (TGR) 135-m filling in June 2003, the Yangtze discharge and suspended sediment concentration (SSC) entering the estuary during the period from 15 May to 15 July 2003 were analyzed to examine the instant effects of the filling on them. The Yangtze discharge and SSC entering the estuary in the periods before, during and after the filling clearly indicated three phases: 1) the pre-storage phase characterized by natural conditions, in which the SSC incr...     

Retrieval of total suspended matters using field spectral data in Shitoukoumen Reservoir, Jilin Province, Northeast China

From August to October in 2006, three times of field spectral measurements with a Field Spec FR spectroradiometer (Analytical Spectral Devices, Inc., USA) were carried out in Shitoukoumen Reservoir, Jilin Province, Northeast China. Owing to the serious soil and water loss in the upstream, reflectance curves of the reservoir were characterized by high concentrations of total suspended matter (TSM). Extending the spectral analysis to 1200nm in the near-infrared band, this research revealed an obvious reflectance peak around 1070nm which was caused by the strong backscattering of high TSM. The method of partial least squares (PLS) regression was applied to retrieving the TSM. Reflectance in two spectral bands, i.e., 675-948nm and 1029-1105nm, were used as variables to develop PLS models. Traditional linear regression, first derivative model and logarithmic model were also used for the comparison of different models. Results showed that the PLS model based on Rrs(675)-Rrs(948) gave out best results with high precision and stability. Although the PLS model based on Rrs(1029)-Rrs(1105) did not have an outstanding performance due to lots of noise, the reflectance peak in the near-infrared band was an important TSM feature and its efficient exploitation would have a eunsiderable significance in TSM remote sensing.     

A new method for the estimation of fine-sediment resuspension ratios in estuaries-taking the turbidity maximum zone of the Changjiang(Yangtze) estuary as an example

Based on the principle of conservative matter removal in estuary,a new method is proposed for estimating the ratio of sediment resuspension in estuaries with fine suspended sediments in the turbidity maximum zone(TMZ) of the Changjiang(Yangtze) estuary during 2005.Results show that there was a range of 18.7%±27.9% to 73.9%±22.5% per annum of total suspended particulate matter(SPM),with an average of 49.2%.Nearly half of the particulate matter in the TMZ originates from sediment resuspension.This indicates that sediment resuspension is one of the major mechanisms involved in formation of the TMZ.Compared with traditional method for calculating these ratios in the estuary,this new method evaluates the dynamic variation of SPM content carried by river runoff from the river mouth to the ocean.The new method produced more reliable results than the traditional one and could produce a better estimation of resuspension flux for particulate matter in estuaries.