共查询到20条相似文献，搜索用时 343 毫秒
1.
Neural network (NN) modeling techniques were used to predict flowability behavior of distillers dried grains with solubles (DDGS) prepared with varying levels of condensed distillers solubles (10, 15, and 20%, wb), drying temperatures (100, 200, and 300°C), cooling temperatures (–12, 25, and 35°C), and storage times (0 and 1 month). Response variables were selected based on our previous research results and included aerated bulk density, Hausner ratio, angle of repose, total flowability index, and Jenike flow index. Various NN models were developed using multiple input variables in order to predict single‐response and multiple‐response variables simultaneously. The NN models were compared based on R2, mean square error, and coefficient of variation obtained. In order to achieve results with higher R2 and lower error, the number of neurons in each hidden layer, the step size, the momentum learning rate, and the number of hidden layers were varied. Results indicate that for all the response variables, R2 > 0.83 was obtained from NN modeling. Compared with our previous studies, NN modeling provided better results than either partial least squares modeling or regression modeling, indicating greater robustness in the NN models. Surface plots based on the predicted values from the NN models yielded process and storage conditions for favorable versus cohesive flow behavior for DDGS. Modeling of DDGS flowability using NN has not been done before, so this work will be a step toward the application of intelligent modeling procedures to this industrial challenge. 相似文献
2.
3.
Riry Wirasnita Tony Hadibarata Abdull Rahim Mohd Yusoff Zulkifli Yusop 《Water, air, and soil pollution》2014,225(10):1-12
Land use specific deposition velocities of atmospheric trace gases and aerosols—particularly of reactive nitrogen compounds—are a fundamental input variable for a variety of deposition models. Although the concept is known to have shortcomings—especially with regard to bi-directional exchange—the often limited availability of concentration data and meteorological input variables make it a valuable simplification for regional modeling of deposition fluxes. In order to meet the demand for an up-to-date overview of recent publications on measurements and modeling studies, we compiled a database of ammonia (NH3) deposition velocities published from 2004 to 2013. Observations from a total of 42 individual studies were averaged using an objective weighing scheme and classified into seven land use categories. Weighted average and median deposition velocities are 2.2 and 2.1 cm s?1 for coniferous forests, 1.5 and 1.2 cm s?1 for mixed forests, 1.1 and 0.9 cm s?1 for deciduous forests, 0.9 and 0.7 cm s?1 for semi-natural sites, 0.7 and 0.8 cm s?1 for urban sites, 0.7 and 0.6 cm s?1 for water surfaces, and 1.0 and 0.4 cm s?1 for agricultural sites, respectively. Thus, values presented in this compilation were considerably lower than those found in former studies (e.g., VDI 2006). Reasons for the mismatch were likely due to different land use classification, different averaging methods, choices of measurement locations, and improvements in measurement and in modeling techniques. Both data and code used for processing are made available as supplementary material to this article. 相似文献
4.
降雨侵蚀力简易算法是较大尺度应用USLE/RUSLE进行土壤侵蚀评价研究的必要内容。基于降雨量和降雨时间建立月降雨侵蚀力计算模型,并以陕北黄土丘陵沟壑区为例,进行模型的拟合。结果表明:随着自变量中降雨量和降雨时间表示方式的改变,模型的拟合优度表现出明显的差异;对于不同因变量而言,以ΣEI30(或lg(ΣEI30))和以ΣEI10(或lg(ΣEI10))为因变量的模型拟合优度在整体上比较接近甚至相同,而以ΣE60I10(或lg(ΣE60I10))为因变量的模型拟合优度在整体上略低;就尺度效应而言,在时间尺度上,整个汛期的模型拟合优度低于1个月份或多个月份模型的拟合优度,在空间尺度上,区域模型中的拟合优度低于至少1个流域的模型拟合优度;在实际应用中,可以选择以ΣEI30为因变量的月降雨侵蚀力公式对该区域进行土壤侵蚀评价。 相似文献
5.
Using support vector machines to predict cation exchange capacity of different soil horizons in Qingdao City,China 
下载免费PDF全文
下载免费PDF全文 Agricultural, environmental and ecological modeling requires soil cation exchange capacity (CEC) that is difficult to measure. Pedotransfer functions (PTFs) are thus routinely applied to predict CEC from easily measured physicochemical properties (e.g., texture, soil organic matter, pH). This study developed the support vector machines (SVM)‐based PTFs to predict soil CEC based on 208 soil samples collected from A and B horizons in Qingdao City, Shandong Province, China. The database was randomly split into calibration and validation datasets in proportions of 3:1 using the bootstrap method. The optimal SVM parameters were searched by applying the genetic algorithm (GA). The performance of SVM models was compared to those of multiple stepwise regression (MSR) and artificial neural network (ANN) models. Results show that the accuracy of CEC predicted by SVM improves considerably over those predicted by MSR and ANN. The performance of SVM for B horizon (R2 = 0.85) is slightly better than that for A horizon (R2 = 0.81). The SVM is a powerful approach in the simulation of nonlinear relationship between CEC and physicochemical properties of widely distributed samples from different soil horizons. Sensitivity analysis was also conducted to explore the influence of each input parameter on the CEC predictions by SVM. The clay content is the most sensitive parameter, followed by soil organic matter and pH, while sand content has the weakest influence. This suggests that clay is the most important predictor for predicting CEC of both soil horizons. 相似文献
6.
Because soil hydraulic properties are indispensable for determining soil water retention and soil solute movement, their input for simulation models is essential. Many of these parameters cannot be estimated directly at the scale of interest, but can only be derived through inverse modeling. During this process, the parameters are generally adjusted using least-squares approach with Levenberg–Marquardt (LM) algorithms in which numerically simulated models are fitted to measured data. In this study we used a new inverse method to estimate the unsaturated soil hydro-dispersive properties from in-situ experiments. The method employs complex-variable-differentiation method (CVDM) to accurately predict of the hydraulic properties of the van Genuchten–Mualem models (θr, θs, α, ks, n). To the knowledge of the authors, it is first study use CVDM in soil physics. The optimization procedure was performed by using a continuous data set of daily in situ soil water content and bromide concentration measurements. Estimated parameters during the inversion showed high correlation (R2 = 0.88, RMSE = 0.013 and the model efficiency CE = 0.77) by using the CVDM-methods with the actual field measurements, compared with the traditional LM-algorithm (R2 = 0.81, RMSE = 0.021 and CE = 0.626). The results show that the new inverse analysis in the present work has the high accuracy, validity, uniqueness, and higher inversion efficiency. Meanwhile, the convergence and stability of the modified LM-algorithm are improved. Overall, it was concluded that the CVDM is promising method to estimate hydro-dispersive parameters in soil physics. 相似文献
7.
[目的]探索一种在不损伤枣树条件下估算枣树生物量的方法,为研究灌溉施肥与枣树生长关系提供技术支撑。[方法]在大量实际测定调查数据的基础上,用线性和非线性生物量模型分别对枣树枝条、枣吊、叶片及果实的生物量进行建模,以相关系数R2,拟合指数W,标准误SEE,变异系数CV和预估精度p作为模型的精度评价指标,对模型进行了检验,并对两类模型进行了对比分析。[结果]线性模型适用于枝条生物量的预测,R2为0.980~0.984,p介于0.909~0.926;叶片生物量和枣吊生物量的最优模型为非线性模型,R2介于0.880~0.943,p介于0.897~0.976;果实生物量最优模型为二元非线性模型,R2为0.999,p为0.998;枝条、叶片和枣吊的鲜重与干重的比例系数分别为2.085,2.854和2.675;单个枣吊长度和枣吊上的叶片数目呈现出良好的一元线性关系,R2达到0.88。[结论]本研究中使用的枣树生物量模型的模拟精度较高。 相似文献
8.
水稻冠层图像分割方法对比研究 总被引：1，自引：0，他引：1
水稻冠层信息在自动化管理上具有重要指导意义,但田间多变光照强度环境显著降低了水稻冠层图像分割和信息提取的精度。为降低光照强度的干扰,本文基于RGB、CIEL~*a~*b~*、HSV色彩空间和多色彩空间(包括RGB、CIEL~*a~*b~*和HSV色彩空间)构建水稻冠层图像的色彩特征组合,然后通过支持向量机(SVM)的线性核函数对水稻冠层图像进行分类识别,其分割方法分别定义为rgb-SVM、lab-SVM、hsv-SVM和Multi-SVM。同时,利用此方法对不同光照强度下的水稻冠层图像进行分割,并与常用的Ex GOtsu分割方法进行对比,比较不同方法的分割效果和光强稳健性。结果表明,rgb-SVM的分割效果优于Ex GOtsu方法,但对晴天条件下获取的水稻冠层图像的分割误差仍较大,光强稳健性低;lab-SVM和hsv-SVM分割方法的分割精确度较低,存在一定的欠分割现象;基于多色彩空间和支持向量机的Multi-SVM分割方法的分割效果最佳,该方法对不同光强下获取的水稻冠层图像的分割误差均控制在4.00%以内,具有较好的光强稳健性。因此,基于多色彩空间和支持向量机的Multi-SVM分割方法能够相对准确地将水稻像元从水稻冠层图像中分割出来,且对田间多变光强条件具有一定的稳健性,可为田间水稻生长发育监测和自动化管理提供一定的技术支持。 相似文献
9.
Mechanistic rhizosphere models are mathematical models that calculate the uptake of solutes (e.g. nutrients such as P or K) in a plant root and/or the concentration profiles of ions or microorganisms across the rhizosphere. They consist of partial differential equations (pde's) and their associated initial and boundary conditions, and often require numerical solutions due to non‐linearity. The suitability of the pde‐solver FlexPDE for rhizosphere modeling was evaluated by reproducing the outputs of three original rhizosphere models and comparing the results with the original model outputs. The models considered were NST 3.0 (Syring and Claassen, 1996), Kirk (1999), and Newman and Watson (1977). From visual comparison, the curves as produced by the original model and FlexPDE match very well. All calculated R2s (coefficients of determination) were equal to 1 with a rounding precision of three decimal places. Rescaling the time, using an upwind scheme in the presence of convection terms and controlling the relative error with the FlexPDE function errlimit were helpful in finding the most accurate numerical solution. These results showed that FlexPDE is a valuable tool in rhizosphere modeling. 相似文献
10.
为更好地研究利用光谱反映的土壤重金属信息，实现具有多重金属复合污染问题的铅锌矿区土壤重金属含量高光谱快速估测，该研究以河北省某铅锌矿区为例，首先对研究区土壤的Cu、Cr、Ni、Zn、Cd、Pb污染状况进行了评价分析，其次基于实验室高光谱数据，组合变换光谱、特征变量和反演算法形成不同反演策略，通过各反演策略下的重金属反演精度比较，定量分析不同光谱预处理、特征选择和建模算法的优劣与适应性，构建最优反演模型。研究结果表明：1）研究区土壤Cr、Ni清洁程度较好，其余Cu、Zn、Cd、Pb均有不同程度污染；参比当地土壤背景值，区域内梅罗综合污染指数均值29.7，为重度污染，潜在生态风险因子均值1330.3，处于高生态风险状态；2）光谱预处理可以增强土壤重金属信息表达。其中，光谱微分效果较好，但易受噪声影响，而多元散射校正、标准正态变量、倒数对数变换可以进行光谱去噪，提升处理效果；3）特征选择方法中，相关系数法选择特征波段数目多，不同重金属反演R2 差异较大；Boruta法选择特征波段数目少，不同重金属反演R2 差异较小；4）BPNN、XGBoost可以较好描述重金属含量与光谱的非线性关系，相较于其他算法具有更好表现，分别实现了Cr、Ni、Zn和Pb、Cd的最优反演，SVMR实现了Cu的最优反演。研究表明，不同的光谱预处理、特征选择与建模算法对于土壤重金属含量的反演均具有较大影响，选择合适的处理、建模算法可以有效提升反演精度。该研究为进一步实现高效、准确、大范围遥感监测铅锌矿区土壤重金属污染状况提供参考依据。 相似文献
11.
M.P. Gonzalez-Dugo C.M.U. Neale L. Mateos W.P. Kustas J.H. Prueger M.C. Anderson F. Li 《Agricultural and Forest Meteorology》2009,149(11):1843
The integration of remotely sensed data into models of evapotranspiration (ET) facilitates the estimation of water consumption across agricultural regions. To estimate regional ET, two basic types of remote sensing approaches have been successfully applied. The first approach computes a surface energy balance using the radiometric surface temperature for estimating the sensible heat flux (H), and obtaining ET as a residual of the energy balance. This paper compares the performance of three different surface energy balance algorithms: an empirical one-source energy balance model; a one-source model calibrated using inverse modeling of ET extremes (namely ET = 0 and ET at potential) which are assumed to exist within the satellite scene; and a two-source (soil + vegetation) energy balance model. The second approach uses vegetation indices derived from canopy reflectance data to estimate basal crop coefficients that can be used to convert reference ET to actual crop ET. This approach requires local meteorological and soil data to maintain a water balance in the root zone of the crop. Output from these models was compared to sensible and latent heat fluxes measured during the soil moisture–atmosphere coupling experiment (SMACEX) conducted over rain-fed corn and soybean crops in central Iowa. The root mean square differences (RMSD) of the estimation of instantaneous latent and heat fluxes were less than 50 W m−2 for the three energy balance models. The two-source energy balance model gave the lowest RMSD (30 W m−2) and highest r2 values in comparison with measured fluxes. In addition, three schemes were applied for upscaling instantaneous flux estimates from the energy balance models (at the time of satellite overpass) to daily integrated ET, including conservation of evaporative fraction and fraction of reference ET. For all energy balance models, an adjusted evaporative fraction approach produced the lowest RMSDs in daily ET of 0.4–0.6 mm d−1. The reflectance-based crop coefficient model yielded RMSD values of 0.4 mm d−1, but tended to significantly overestimate ET from corn during a prolonged drydown period. Crop stress can be directly detected using radiometric surface temperature, but ET modeling approaches-based solely on vegetation indices will not be sensitive to stress until there is actual reduction in biomass or changes in canopy geometry. 相似文献
12.
Paolo Porto Des E. Walling Vito Ferro Costanza di Stefano 《Land Degradation \u0026amp; Development》2003,14(4):389-408
Increasing concern for problems of soil degradation and the off‐site impacts of accelerated erosion has generated a need for improved methods of estimating rates and patterns of soil erosion by water. The use of environmental radionuclides, particularly 137Cs, to estimate erosion rates has attracted increased attention and the approach has been shown to possess several important advantages. However, the use of 137Cs measurements to estimate erosion rates introduces one important uncertainty, namely, the need to employ a conversion model or relationship to convert the measured reduction in the 137Cs inventory to an estimate of the erosion rate. There have been few attempts to validate these theoretical conversion models and the resulting erosion rate estimates. However, there is an important need for such validation, if the 137Cs approach is to be more widely applied and reliance is to be placed on the results obtained. This paper reports the results of a study aimed at validating the use of two theoretical conversion models, namely the exponential depth distribution model and the diffusion and migration model, that have been used in several recent studies to convert measurements of 137Cs inventories on uncultivated soils to estimates of soil erosion rates. The study is based on data assembled for two small catchments (1.38 and 1.65 ha) in Calabria, southern Italy, for which measurements of sediment output are available for the catchment outlet. The two catchments differ in terms of the steepness of their terrain, and this difference is reflected by their sediment yields. Because there is no evidence of significant deposition within the two catchments, sediment delivery ratios close to 1.0 can be assumed. It is therefore possible to make a direct comparison between the estimates of the mean annual erosion rates within the two catchments derived from 137Cs measurements and the measured sediment outputs. The results of the comparison show that the erosion rate estimates provided by both models are reasonably consistent with the measured sediment yields at the catchment outlets. However, more detailed assessment of the results shows that the validity of the erosion rate estimates is influenced by the magnitude of the erosion rates within the catchment. The exponential depth distribution model appears to perform better for the catchment with higher erosion rates and to overestimate erosion rates in the other catchment. Similarly, the basic migration and diffusion model performs better for the catchment with lower erosion rates and overestimates erosion rates in the other catchment. However, the improved migration and diffusion model appears to perform satisfactorily for both catchments. There is a need for further studies to extend such independent validation of the 137Cs technique to other environments, including cultivated soils, and to other conversion models and procedures. Copyright © 2003 John Wiley & Sons, Ltd. 相似文献
13.
Balances of alteration and migration of clay fractions and clay minerals in Gray Brown Podzolic Soils from Loess Balances of alteration and migration of clay fractions, clay minerals, K, Mg and Fe bound in clay silicates were calculated for Gray Brown Podzolic Soils (Parabraunerden) derived from Young Pleistocene Loess below arable and forest land in the district south of Würzburg. The extent of clay formation lies between 63 and 98 kg/m2, of which 60–80 % belong to the fraction <0.1 μm. The clay migration varies between 35 and 51 kg/m2, the fractions < 0.2 μm (especially the fraction <0.1 μm) predominating, the coarse clay being involved only to a smaller extent. During Holocene soil formation 121 kg illite, 16 kg vermiculite and 11 kg kaolinite per m2 are formed in the clay fraction (< 2 pm). The loss of smectite amounts to 68 kg/m2. The biggest alterations of the clay mineral quantities occur in the fraction <0.1 pm; they indicate a smectite-illite transformation. Illite, at 23 kg/m2, accounts for half of the clay migration, followed by smectite and vermiculite each at 9 kg/m2 and kaolinite at 5 kg/m2. When comparing the migrated with the present amounts no preference of certain clay minerals during clay migration can be determined. In the clay fractions the gained Fe (3.07?4.32 kg/m2) and K (2.75?3.84 kg/m2) predominate over Mg (0.57?1.15 kg/m2). The three elements accumulate to the greatest extent in the fine clay fraction. The element migration parallels the gain. The balances of the elements are discussed in connection with pedogenic illite formation as well as biotite and vermiculite disintegration. 相似文献
14.
Alper Unal 《Water, air, and soil pollution》2009,200(1-4):47-57
The objective of this study was to improve the ability to model the air quality impacts of biomass burning on the surrounding environment. The focus is on prescribed burning emissions from a military reservation, Fort Benning in Georgia, and their impact on local and regional air quality. The approach taken in this study is to utilize two new techniques recently developed: (1) adaptive grid modeling and (2) direct sensitivity analysis. An advanced air quality model was equipped with these techniques, and regional-scale air quality simulations were conducted. Grid adaptation reduces the grid sizes in areas that have rapid changes in concentration gradients; consequently, the results are much more accurate than those of traditional static grid models. Direct sensitivity analysis calculates the rate of change of concentrations with respect to emissions. The adaptive grid simulation estimated large variations in O3 concentrations within 4?×?4-km2 cells for which the static grid estimates a single average concentration. The differences between adaptive average and static grid values of O3 sensitivities were more pronounced. The sensitivity of O3 to fire is difficult to estimate using the brute-force method with coarse scale (4?×?4 km2) static grid models. 相似文献
15.
准确获取及预测光合色素含量可为精细化种植管理提供数据依据,为探究花生冠层叶片色素吸收特征,该研究以开花下针期的花生冠层叶片为研究对象,以ASD Field Spec4野外便携式高光谱仪采集的光谱数据为数据源,进行花生叶片叶绿素含量和类胡萝卜素含量反演。通过对比7种单一筛选特征波长变量算法及结合4种模型(PLSR、SVR、GBDT和XGBoost)的结果,优选出3种算法进行两两耦合。结果表明:1)在单一算法试验中IRIV、UVE和GA算法结果较优;2)在耦合算法试验中UVE-IRIV、GA-IRIV和GA-UVE方法都能有效降维,且模型稳定性提升。在叶绿素含量反演模型中,GA-IRIV-XGBoost模型精度最高,R2=0.622,RMSE=0.235 mg/g;在类胡萝卜素含量反演模型中,UVE-IRIV-XGBoost模型精度最高,R2=0.575,RMSE=0.056 mg/g;3)比较两种色素反演模型的预测精度,表明叶绿素的预测精度优于类胡萝卜素。该结果可为快速、准确预测花生叶片光合色素含量提供一种方法。 相似文献
16.
煤矿区土壤有机碳含量的高光谱预测模型 总被引：2，自引：0，他引：2
可见—近红外光谱已被证明是一种快速、及时、有效的土壤有机碳含量预测工具。利用Field Spec4对济宁鲍店矿区的104个土壤样品进行光谱测量,采用Savitzky-Golay卷积平滑(SG)、多元散射校正(MSC)及数学变换等多种方式组合对光谱预处理,并运用偏最小二乘回归分析建立土壤有机碳含量预测模型,进而探讨煤矿区土壤有机碳含量的高精度预测方法。结果表明:(1)不同的光谱预处理方法对建模结果影响差异较大,建模结果以SG加MSC预处理再结合光谱反射率的一阶微分变换最优,建模R~2=0.86,RMSE=2.0g/kg,验证R~2=0.78,RMSE=1.81g/kg,RPD=2.69。(2)倒数和倒数的对数与土壤有机碳含量的相关性曲线接近重合,与反射率曲线成反比,但是建模效果远低于反射率;光谱反射率的一阶微分能明显提高500~600nm波段相关性。(3)光谱反射率随土壤有机碳的含量减少而增大,当有机碳含量较低时,其波谱的近红外波段反射率响应能力也随之降低,反射率直接建模难度加大。 相似文献
17.
18.
It is common practice to fit mathematical models to radionuclide activity–depth profiles in soils in order to quantify rates of vertical migration through the soil profile. We have fitted six such models to 21 different activity–depth profiles of radiocaesium (137Cs) derived from Chernobyl and determined relations between the models and the values of their parameters. The advection and dispersion parameters obtained using four solutions to the advection–dispersion equation (each based on different initial and boundary conditions or different simplifications) are in good agreement. We further develop a relation between parameter values obtained using the advection–dispersion models and those determined by a simpler exponential function of the form Aexp(–Bt) where t is the time and A and B are parameters to be estimated. One of the advection–dispersion models proved to be significantly better than the others in terms of goodness-of-fit, versatility and ease of use. A simple model, using calculations based on measured characteristics of the activity–depth profile, was shown to accord well with parameters derived from more complex models based on statistical curve fitting. We have also evaluated the ‘residence time’ or ‘compartmental’ model approach to characterizing radionuclide activity–depth profiles. We relate such models to a numerical solution of a simple advection equation, and we show that apparent dispersion in compartmental models is an artefact of numerical dispersion, which can be quantified by the Courant condition. For activity profiles that have a significant advection component, using solutions to the advection–dispersion equation, we have observed a strong positive correlation between advection and dispersion in the profile. 相似文献
19.
We evaluated potential soil nitrogen mineralization of 46 sandy fields of the Pampas for determining the contribution of deep layers to mineralization and modeling its trend in depth as a possible tool for improving current existing mineralization models based on surface data. Mineralization, total and mineral nitrogen decreased with depth. A potential model fitted well to these variables (R2 = 0.95–0.99), but mineralization showed a more stratified profile. Consequently, the fraction of total nitrogen mineralized decreased with depth despite soils had constant texture across the profile. Potential mineralization to 1 m depth could be estimated using data from the 0–0.2-m soil layer and the average curvature of the potential model (R2 = 0.60) or linear regression methods (R2 = 0.71). Another estimation of potential mineralization could be performed by developing a pedotransfer function which used as predictors total nitrogen and depth (R2 = 0.62), without the need of laboratory incubations. Our results showed that for sandy soils, deep nitrogen mineralization account for 40% of soil mineralization and can be assessed using surface data or the total nitrogen content of the soils. Because surface soil mineralization and whole profile mineralization were highly correlated, it is improbable that field mineralization modeling may be improved using deep data in these soils. 相似文献
20.
Kinetically controlled release of uranium from soils 总被引：1，自引：0，他引：1
A. BRAITHWAITE F. R. LIVENS S. RICHARDSON M. T. HOWE K. W. T. GOULDING 《European Journal of Soil Science》1997,48(4):661-673
Although trade element uptake on and release from solid phases are fundamental controls on the migration of the elements in the environment, the controls are incompletely understood. The extraction of uranium from two soils, both of which have been labelled naturally with uranium, was therefore studied using a cation resin exchange technique. One soil was a peat from the Needle's Eye natural analogue site, Scotland, and the other was a calcareous brown earth from Derbyshire, England. The effects of different exchanging cations, solution pH and the presence of complexing anions (Cl?, CO32?, SO42?) in solution on uranium extraction were assessed. The extraction could be described by a simple, first-order kinetic model with up to three rate constants being identifiable in individual experiments. In both soils no single reaction pathway appeared to dominate, and extraction was slow, with rate constants of 10?3?10?4 h?1 in acid conditions and around 10?6 h?1 in neutral conditions. Half-times for uranium release in the experiments were in the range 30–60 days in acid and around 10 years in neutral conditions; in the field they are therefore expected to be several years at both sites. Incorporation of kinetic factors into a simple one-dimensional migration model illustrates that their overall effect is to retard migration. Ideally, therefore, reaction rates should be taken into account in predictive modelling of element transport. 相似文献