• <tr id="yyy80"></tr>
  • <sup id="yyy80"></sup>
  • <tfoot id="yyy80"><noscript id="yyy80"></noscript></tfoot>
  • 99热精品在线国产_美女午夜性视频免费_国产精品国产高清国产av_av欧美777_自拍偷自拍亚洲精品老妇_亚洲熟女精品中文字幕_www日本黄色视频网_国产精品野战在线观看 ?

    Study on the Comparisons of the Establishment of Two Mathematical Modeling Methods for Soil Organic Matter Content Based on Spectral Reflectance

    2016-06-15 16:39:51ZHANGPeiLIYi
    光譜學與光譜分析 2016年3期
    關鍵詞:黑河適用性反射率

    ZHANG Pei, LI Yi,2*

    1. College of Water Resources and Architecture Engineering, Northwest A&F University, Yangling 712100, China

    2. Water Saving Agriculture Academy in China Arid Zone, Northwest A&F University, Yangling 712100, China

    Study on the Comparisons of the Establishment of Two Mathematical Modeling Methods for Soil Organic Matter Content Based on Spectral Reflectance

    ZHANG Pei1, LI Yi1,2*

    1. College of Water Resources and Architecture Engineering, Northwest A&F University, Yangling 712100, China

    2. Water Saving Agriculture Academy in China Arid Zone, Northwest A&F University, Yangling 712100, China

    Existing prediction models of soil organic matter content (SOC) are restricted by some factors, such as sampling scale, soil type and spectral parameters of samples. Therefore, it is necessary to make a comparative analysis on larger scales to build a quantitative model with better feasibility and greater accuracy. A total of 225 soil samples were collected in an extensive region of the upper reaches of Heihe river basin. SOC and spectral reflectance were being measured. All the samples were divided into 2 subsets-a modeling subset (180 samples) and a validation subset (45 samples). Six indices were obtained through transformation of soil spectral reflectance (R), continuum-removal (CR), reciprocal (REC), logarithm of reciprocal (LR), first-order differential (FDR) and Kubelka-Munck transformation coefficient (K-M). To build the mathematical model of SOC with 12 spectral indices, two methods, i.e., stepwise linear regression and partial least-square regression were used based on the modeling subset, respectively; the validation subset is used for model evaluation. The results indicated that: (1) Regardless of different modeling methods, model between SOC and LR index was always the best among the 6 reflectance-related indices. LR was the best index for predicting SOC; (2) For the model based on the LR index, the accuracy of model using partial least-square regression method was better than that using stepwise linear regression method; (3)225 samples were compared to verify the former available published SOC model. Both the predicted and measured values passed the mean value t-test, and the Pearson correlation coefficient reached 0.826. It shows that local prediction model can be applied to the research of predicting SOC in the larger scale.

    Upper reaches of Heihe; Soil organic matter; Stepwise linear regression; Partial least square regression; Mathematic model; Spectral reflectance

    Introduction

    Soil nutrients play a significant role in the high efficient utilization of water and fertilizer. In which, soil organic matter content (SOC) reflects the fertility characters and the health degree of soil[1]. High accuracy can be achieved for determining SOC in laboratory test using the traditional methods, including complete wet-combustion method and potassium chromate volumetric method[2], etc. However, these methods are complex, in the meantime, the environment are easily to be polluted to some extent, and therefore they cannot be extended on large scales. With the development of quantitative remote sensing technology, many researches have been performed to obtain SOC values easily and rapidly in a large area. The utilization of the models for SOC based on spectral reflectance characteristics can reduce the workload of experimental analysis. And the application of this method to SOC estimation in a large area can provide references for agricultural production planning.

    Several international achievements have been made in predicting SOC based on spectral reflectance. For instance, Palacios-Orueta et al. studied the influences of SOC, iron content and texture on spectral reflectance through multivariate analysis, principal component analysis, discriminant analysis and waveband depth analysis[3]and then provided reference spectral wavebands for the prediction model of SOC. According to the reference spectral wavebands mentioned above, Liu et al. established a spectral prediction model for SOC in the saline land in Beiwucha Town, Manasi County, Xinjiang, China using the stepwise linear regression method (SLR). Their results indicated that, SOC models which established by different indices would have different prediction accuracy[4]. Bilgili et al. constructed a spectral prediction model for SOC in the semi-arid area of northern Turkey using the multivariate adaptive regression splines and partial least-square regression (PLSR) method. Their results revealed that higher prediction accuracy of the model could be achieved by improving the modelling method[5]. Based on the measured SOC values, Zhang et al. established a prediction model by calculating the pixel brightness value of the remote sensing image of the Landsat/TM waveband image for Fuxin, Liaoning, China. They found that the model showed different adaptability influenced by research scale and soil type[6]. Shi et al. also constructed a prediction model for SOC based on the spectral data of 1581 soil samples from 14 provinces in China. Their results suggested that the accurate prediction of SOC was influenced by soil factors, and therefore, accurate trans-regional modelling couldn’t be realized[7]. Using the SLR method, Liu Jiao et al. proposed a spectral prediction model for SOC in typical farmlands and high mountain areas on the upper reaches of Heihe, Gansu, China[8]with high prediction accuracy, but the established model could not be applied large scale due to the limitation of sampling scale. A better modelling method can be proposed to further improve the model accuracy and provide a reliable way to predict SOC on large scales.

    The prediction accuracy of a model, to a great extent, decides whether or not the model is reliable and practical. However, different modelling methods have different degrees of adaptability for different areas and scales, which brings difficulties for SOC modelling. Based on the measured SOC values and spectral reflectance, this research constructed spectral prediction models of SOC on the upper reaches of Heihe using the PLSR and the SLR methods. Through a comparison of the stability and prediction accuracy of the models obtained using the two methods, the best model for predicting SOC based on spectral reflectance was determined. Besides, by comparing the results in this research with previous achievements[7], in the case of expanding the sampling scale, the applicability of the previous model was validated. The result in this research is expected to provide more references for prediction of large-scale SOC based on spectral data.

    1 Materials and methods

    1.1 Soil sampling

    The upper reaches of Heihe are geographically in the range of 37.7°—42.7°N and 97.1°—102.0°E. In this area, the annual average precipitation is 300~700 mm, and the elevation varies from 1,700 to 5,066 m, with significant vertical zones, including the landscapes of glacier, grassland, forest, river shoal and waste land[9]. Owing to the Heihe section of Qilian Mountains runs through the upper reaches of Heihe in northwest-southeast direction, rich variation information concerning soil properties is found along the way. Therefore, the samples were collected along the ridge line of Qilian Mountains from Yugur Autonomous County of Sunan to Shandan County, Gansu, China direction. The distance between adjacent sampling points was about 3 km. Additionally, sampling points were distributed more extensively in different landscape zones, including the Heihe valley of Qilian mountains, Biandukou in Minle County, the military horse-breeding farm in Shandan County, Dayekou in Zhangye City and Kangle grassland. Totally 225 samples were collected at 0~10 cm depth range. These samples were all dried in laboratory, ground and screened by a 2 mm-diameter siever.

    1.2 Analysis of soil properties and classification of sample sets

    The particle size distribution of soil was obtained using a laser diffraction particle size analyzer (Mastersizer 2000)[10]. According to the international classification standards for soil texture, among the 225 soil samples, there were 77 clay loam samples, 78 loam samples, 21 sandy loam samples, and 49 samples of sand, light clay, silt clay, silt clay loam and silt loam. The SOC of these samples was measured using potassium chromate volumetric method[2]. Afterwards, the SOC data of the 225 samples were ranked in an increasing order. Beginning with the third sample, 45 samples occupying 20% of the total were selected with an interval of 4 samples to make up the validation subset, and the rest 180 ones were taken as modelling subset, accounting for 80% of the total. The statistical properties of the total samples, the modelling and the validation subsets are listed in Table 1.

    Table 1 Descriptive statistics of soil organic matter (Unit: g·kg-1)

    Table 1 reveals that the statistical parameters of the modelling and validation sets are close to that of the total samples and can reliably reflect the SOC distribution properties of the total samples. Therefore, the classification of subsets is suitable for further modelling and validation analysis.

    1.3 Acquisition and preprocessing of the spectral data of soil samples

    The spectral experiment was conducted in the Key Lab of Agricultural Soil and Water Engineering in Arid and Semiarid Areas, Ministry of Education, Northwest A & F University, China. The tests were conducted during 10:30—14:30 in sunny days with sufficient sunshine and clear sight. The soil spectrum was measured using a portable hyper-spectrometer ASD FieldSpec (Analytical Spectral Devices). The waveband ranges from 350~1 830 nm, among which, two wavebands of 350~999 nm and 1 000~1 830 nm were sampled at intervals of 1.4 nm and 2 nm, respectively. Before measurement, the apparatus was preheated for half an hour to reduce the error. When the measurement started, considering the factors including variation of light intensity et al., dark current was removed and whiteboard was calibrated every 10 min. The soil samples were placed in the aluminium boxes with 5 cm in diameter and 4 cm in height. The spectrometer was equipped with a sensor probe with 30° of field angle and was vertically fixed at 5 cm above the center of the soil sample. The coverage area of the sensor probe was within the boundaries of the aluminium boxes. Each sample was measured for 10 times and then the arithmetic mean was calculated. Given the non-uniform preheating of the apparatus, the step was observed at some spectral curves at 1000 nm wavelength. Therefore, the short infrared wavebands of the curves were corrected using ViewSpec Pro Version 6.0 (ASD Inc.). Besides, because water absorption waveband, iron oxide, and OH-in kaolinite clay mineral exerted influences on the soil spectral reflectance at 1 400 nm around, the spectral reflectance curve fluctuated violently[11], such wavebands were excluded.

    After transform processing of the observed original spectral reflectance (R), the following parameters were obtained: the reciprocal ofR(REC), the logarithm of REC (LR), the first-order differential ofR(FDR), the transform coefficient of Kubelka-Munck (KM)[12]and continuum removal (CR). Among which, REC, LR and FDR were determined with ViewSpec Pro Version 6.0; CR was obtained using the spectral analysis module of ENVI 4.7 (Exelis VIS Inc.); KM was calculated by the following

    (1)WhereRis reflectance value corresponding to spectral wavelength.

    1.4 Modelling and validation

    A model for spectral reflectance of SOC was established using SLR and PLSR methods respectively. By evaluating the modelling effects of the two methods, the optimal model was obtained.

    1.4.1 To establish models using the SLR and the PLSR methods

    (2)Where the unit of SOC is g·kg-1;c0is a constant;cirefers to coefficient andi=1, 2, 3…,n;Xjdenotes the corresponding spectral parameter value of the sensitive wavelength of the model;jis wavelength;Nis number of the selected wavelengths.

    (3)Wherenis sampling number; SOCOBSis measured SOC value(g·kg-1); SOCPREdenotes predicted SOC value (g·kg-1).

    In model establishment process using SLR and PLSR methods, spectral waveband information for establishing models were both selected through a statistically strict regression process. Both the selected sensitive wavelengths and principal components were significantly correlated to the observed SOC values. The obtained models met the statistical requirement of reliability. And there are some statistical bases for the selection of sensitive wavelengths. Stepwise linear regression analysis process was realized using IBM SPSS 22.0 (IBM Inc.) while PLSR was achieved with the Unscrambler X 10.3 (CAMO Inc.).

    1.4.2 Model evaluation

    (4)

    (5)

    (6)

    WhereSdis the standard deviation of the difference between the measured and predicted values;ndenotes sample size.

    2 Results and analysis

    2.1 The models and their validations using the SLR method

    Based on the R data of the 180 soil samples from the modelling subset, the SLR models of SOC for different indices includingR, CR, REC, LR, FDR and KM were constructed. The model etstablishment results of SOC~Rwas taken as an example. The corresponding equations for the models and validated statistical parameters are listed in Table 2.

    Likewise, by comparing the SOC models using the indices of CR, REC, LR, FDR and KM, the optimal models are determined, as shown in Table 3.

    Table 2 Establishment and validation of SOC-R model using the SLR method

    Table 3 The established SOC models using different spectral indices and the validation parameters using the SLR method

    To compare the model establishment accuracy of the models with different indices using the SLR method, the predicted SOC values were compared with the measured values. The results are illustrated in Figure 1.

    Fig.1 Chart of predictive effects of various models using the SLR method

    Figure 1 shows that, the scatter diagram of the SOC~LR model is closest to the 1∶1 line, with the maximal slope (0.937 7) and greaterR2(0.583 9). There is little error between the measured and estimated values. The SOC~LR model exhibits the best prediction effect, which is consistent with the results in Table 3.

    2.2 The models and their validations using the PLSR method

    Similarly, the PLSR models of SOC for multiple spectral indices, consisting ofR, CR, REC, LR, FDR and KM were established. To simplify the models, the maximum number of extracted principal component is 10. Taking the model for SOC~Ras an example, the corresponding modelling and validation parameters are presented in Table 4.

    Table 4 Establishment and validation of SOC-R

    Table 4 indicates that, with the increase in the number of the principal components, the explanation effects in SOC~Rmodel are better. In the case of 10 principal components, both theRC(0.877) and the accumulative contribution rate (76.89%) are largest and the RMSECV (16.68) is the smallest. Although the increase of principal components improves the stability of the models, the over-fitting might appear to reduce their prediction accuracy[15]. Therefore, to determine the best SOC model, the validation effects of these models need to be compared further. When 7 principal components are selected, minimal values ofrp(0.864) and RMSEP (17.78) are achieved; besides, the model passes theTtest of mean values for the measured and fitted values at the significant level of 0.05. Thus, the 7-principal-component model was selected as the best SOC~Rmodel.

    Likewise, the best models using other 6 spectral indices were selected, shown in Table 5.

    In Table 5, the SOC model using the index LR shows the largestRC(0.881), the largest accumulative contribution rate (77.6%), and the smallest RMSECV(16.63), the explanation effects of fitting the data from modelling subset is the best. The validation results indicate that the SOC~LR model had the largestrp(0.869) and the smallest RMSEP (17.23). The model passes thettest of mean values for the measured and fitted values at the significant level of 0.05. Therefore, the model of SOC~LR is the best when using the PLSR method.

    Table 5 The established SOC models using different spectral indices and the validation parameters using the PLSR method

    SpectralindexModelestablishmentValidationPC?NumberRCCumulative/%RMSECVrpRMSEP|t|R70.85873.6217.820.86417.780.267CR60.84471.2818.590.85618.380.372REC60.87476.3016.890.84418.630.156LR70.88177.6016.630.86917.230.117FDR20.85072.2718.270.79021.520.546K?M60.87376.1716.940.84018.870.13

    To contrast the prediction effects SOC models with various indices using the PLSR method, the predicted SOC were compared with the measured values. The results are shown in Figure 2.

    Figure 2 demonstrates that, the scatter plot of the SOC~FDR model is the closest to the 1∶1 line among all plots, with a slope of 0.927 6, but had the smallestR2(0.435 3) and larger prediction error, thus the model was not good. All of the other 5 scatter diagrams are close to the 1∶1 line, of which, theR2(0.658 6) of the SOC~LR model is the largest, and its prediction effect is the most effective. Hence, the model for SOC~LR is the best when using the PLSR method.

    For both the SLR and PLSR method, LR is the best predictor. By comparing Figures 1 d and 2 d which illustrate the models for SOC~LR using the two methods, the slope (0.937 7) of the model for SOC~LR based on the SLR method is close to 1, which is larger than slope (0.91) of the model for SOC~LR based on the PLSR method; thus, the model based on the SLR method is better in prediction effects. However, theR2(0.658 6) of the PLSR-based model for SOC~LR is larger. Therefore, the validation effects of the two models are similar. The validation parameters in Tables 3 and 5 suggest that the RMSEP (17.23) andrp(0.869) in the model using the PLSR method are obviously superior to the corresponding parameters in the model using the SLR method. Thus, the SOC~LR model based on the PLSR method is more accurate than the SOC~LR model based on the SLR method.

    2.3 Discussions

    Liu Jiao et al. established different models for predicting SOC at the small sampling scale on the upper reaches of Heihe based on measured R and the derived variables, among which

    Fig.2 Chart of predictive effects of various models using the PLSR method

    the best model is the SOC~Rmodel[8], and expressed as

    SOC=-189.3R1 632-740.3R836+615.9R1 326+314.3R474

    (7)

    By using their model, the SOC values of the 225 soil samples in this study were validated. By comparing with the observed values, thervalue of the measured and predicted SOC values is 0.826, the RMSEP is 20.24, and the model passes T test of mean value. It implies that the SOC model which had been established in local area can be applied to larger scaled to predict SOC. However, the extended scale of SOC model obtained at a small scale and its prediction accuracy for upscaling needs to be further explored because there is the influence of spatial variability in soil properties.

    Sun Yongmeng pointed out that, when the studied scale is large, and there is strong spatial variability, RMSEP increased accordingly, so the models established at different scales should not be compared directly[16]. Although the value of RMSEP of the empirical model proposed by Liu Jiao et al. is relatively large, the model shows strong prediction within certain accuracy range. Thus, their model can be employed for predicting SOC on the upper reaches of Heihe.

    With the diversity of the methods, it is necessary to explore more effective modelling methods so as to achieve more accurate prediction and more comprehensive information concerning the distribution of SOC. This research established the best model for predicting SOC using the SLR and PLSR methods at a larger sampling scale. The obtained models in this research are more effective than the existing models[8]both in stability or prediction accuracy. Particularly, the SOC model based on PLSR method shows superior parameters, i.e., larger RMSEP (17.23) andrp(0.869) than those of existing empirical models and is more suitable for predicting SOC on the upper reaches of Heihe.

    3 Conclusion

    According to six spectral-related indices, the SOC models on the upper reaches of Heihe using the two methods of SLR and PLSR were established. By comparing the related parameters during model establishment and validation processes, among the established SOC models using the six spectral-related indices, The two models using the index LR for both the PLSR and SLR methods were very effective for predicting SOC on the upper reaches of Heihe. Furthermore, the SOC~LR model obtained by the PLSR method was more accurate than the model obtained by the SLR method. Therefore, SOC~LR model obtained with the PLSR method is selected as the best model among several established models and it is suitable for predicting SOC on the upper reaches of Heihe.

    [1] Wang H J, Shi X Z, Yu D S, et al. Ecology and Environmental Sciences, 2010, 19: 197.

    [2] Bao S D. Soil and Agricultural Chemistry Analysis. Beijing: China Agriculture Press, 2000. 27.

    [3] Palacios-Orueta A, Ustin S L. Remote Sensing of Environment, 1988, 65(2): 170.

    [4] Liu S B, Li Y, He C S. Soil Science, 2013, 178(3): 138.

    [5] Zhang F S, Qu W, Yin G H, et al. Chinese Journal of Applied Ecology, 2010, 21(4): 883.

    [6] Zhou S, Wang Q L, Peng J, et al. Earth Sciences, 2014, 57(7):1671.

    [7] Bilgili A V, van Es H M, Akbas F, et al. Journal of Arid Environments, 2010, 74: 229.

    [8] Liu J, Li Y, Liu S B. Spectrosc. Spectr. Anal., 2012, 33(12): 3354.

    [9] Qin J, Ding Y J, Wu J K, et al. Journal of Arid Land, 2013, 5(3): 366.

    [10] Peng H T. Horton R, Lei T W, et al. Journal of Soils and Sediments, 2015, 15: 937.

    [11] Ge Y F, Morgan C L S, Grunwald S, et al. Geoderma, 2011, 161: 202.

    [12] Feng Y Z, Sun D W. Talanta, 2013, 105: 244.

    [13] Li Y, Liu S B, Liao Z H, et al. Canadian Journal of Soil Science, 2012, 92(6): 845.

    [14] Mark J, Michael A. The Appraisal Journal, 2001. 89.

    [15] Li L N, Li Q B, Yan H L, et al. Spectrosc. Spectr. Anal., 2010, 30(11): 3041.

    [16] Sun Yongmeng, Ding Jianli, Qu Juan, et al. Transactions of the Chinese Society of Agricultural Engineering, 2012, 28(20): 180.

    *通訊聯(lián)系人

    S123

    A

    基于光譜反射率的兩種土壤有機質數(shù)學建模方法對比

    張 沛1, 李 毅1,2*

    1. 西北農林科技大學水利與建筑工程學院, 陜西 楊凌 712100

    2. 西北農林科技大學中國旱區(qū)節(jié)水農業(yè)研究院, 陜西 楊凌 712100

    已有土壤有機質的光譜預測模型其適用性受建模樣本的采樣尺度、 土壤類型及光譜參數(shù)限制, 需要在大尺度及范圍上進一步檢驗適用性, 并比較分析不同建模方法的建模效果以尋求適用性更好、 精度更高的定量模型。 在黑河上游大尺度范圍采得225個土壤樣品, 進行了土壤有機質(SOC)及光譜反射率測定后將樣本劃分為建模集(180個土樣)與驗證集(45個土樣)。 將土壤光譜反射率(R)變換處理后得到連續(xù)統(tǒng)去除(CR)、 倒數(shù)(REC)、 倒數(shù)之對數(shù)(LR)、 一階微分(FDR)及Kubelka-Munck變換系數(shù)共6種指標, 針對建模集分別采用逐步線性回歸與偏最小二乘回歸方法建立12種光譜指標與SOC的數(shù)學模型, 并采用驗證集進行模型預測效果評價。 結果表明: (1)采用逐步線性回歸或偏最小二乘回歸方法建模, LR指標對SOC變化的解釋效果都是最好, 是SOC的最優(yōu)預測因子。 (2)基于LR指標建立的SOC模型中, 采用偏最小二乘回歸模型比逐步線性回歸模型的預測精度更好, 相較于黑河上游已有的經驗模型, 偏最小二乘回歸法建立的模型的預測效果也更好。 (3)采用本實驗的225個土壤樣品對比驗證了黑河上游僅有的SOC模型。 該模型的SOC預測值與實測值通過了均值T檢驗且Pearson相關系數(shù)達0.826, 表明在局部典型區(qū)域建立的SOC預測模型, 可以應用到更大尺度上的土壤有機質預測研究。

    黑河上游; 逐步線性回歸; 偏最小二乘回歸; 有機質預測模型; 光譜

    2014-12-13,

    2015-04-18)

    2014-12-13; accepted: 2015-04-18

    National Natural Science Foundation of China (No. 51579213), the National High Technology Research and Development Program of China (No. SS2013AA100904), the China 111 Project (No. B12007), and China Scholarship Council for Studying Abroad (201506305014)

    10.3964/j.issn.1000-0593(2016)03-0903-08

    Biography: ZHANG Pei,(1990-),Master Degree in College Water Resources and Architecture Engineering, Northwest A&F University e-mail: zpr0803@163.com *Corresponding author e-mail: liyikitty@126.com

    猜你喜歡
    黑河適用性反射率
    影響Mini LED板油墨層反射率的因素
    近岸水體異源遙感反射率產品的融合方法研究
    海洋通報(2022年4期)2022-10-10 07:40:26
    具有顏色恒常性的光譜反射率重建
    強調簡潔和適用性 MICHI by Rotel X5/X3合并功放
    黑河的“護衛(wèi)隊”
    當代陜西(2019年22期)2019-12-18 05:32:08
    環(huán)保技術在土木工程領域的適用性探討
    到張掖看黑河
    文學港(2019年5期)2019-05-24 14:19:42
    小議階級分析理論的適用性
    化學腐蝕硅表面結構反射率影響因素的研究*
    電子器件(2017年2期)2017-04-25 08:58:37
    張掖黑河濕地國家級自然保護區(qū)
    成人午夜精彩视频在线观看| h视频一区二区三区| 亚洲av综合色区一区| a 毛片基地| 性色avwww在线观看| 毛片一级片免费看久久久久| 黑人猛操日本美女一级片| 免费不卡的大黄色大毛片视频在线观看| 亚洲精品日韩av片在线观看| 国产永久视频网站| 亚洲欧美一区二区三区黑人 | 成人18禁高潮啪啪吃奶动态图 | 一区二区三区四区激情视频| 国产男女内射视频| 在线观看三级黄色| 最黄视频免费看| 中文欧美无线码| 免费高清在线观看视频在线观看| 成人毛片a级毛片在线播放| 欧美区成人在线视频| 亚洲色图av天堂| 性高湖久久久久久久久免费观看| 亚洲色图av天堂| 五月开心婷婷网| 亚洲性久久影院| 熟女av电影| 久久久欧美国产精品| av视频免费观看在线观看| 夜夜爽夜夜爽视频| 男人添女人高潮全过程视频| 在线观看人妻少妇| 亚洲精品一区蜜桃| 伊人久久国产一区二区| 久久久久久久大尺度免费视频| 新久久久久国产一级毛片| 如何舔出高潮| 国产无遮挡羞羞视频在线观看| 久久精品国产亚洲网站| 欧美精品亚洲一区二区| 久久久久精品性色| 国产精品一区www在线观看| 麻豆成人午夜福利视频| 国产精品熟女久久久久浪| 丰满少妇做爰视频| 卡戴珊不雅视频在线播放| 女人十人毛片免费观看3o分钟| 综合色丁香网| 亚洲av电影在线观看一区二区三区| 色哟哟·www| 久久人人爽人人爽人人片va| 日韩中文字幕视频在线看片 | 香蕉精品网在线| 久久精品久久久久久久性| 午夜激情福利司机影院| 熟女人妻精品中文字幕| a 毛片基地| 国产人妻一区二区三区在| 熟女人妻精品中文字幕| 最近2019中文字幕mv第一页| 国产片特级美女逼逼视频| 在线 av 中文字幕| 下体分泌物呈黄色| 下体分泌物呈黄色| 国产美女午夜福利| 国产美女午夜福利| 人妻系列 视频| 国产精品国产三级国产专区5o| 少妇熟女欧美另类| 少妇熟女欧美另类| 欧美丝袜亚洲另类| 妹子高潮喷水视频| 亚洲精品中文字幕在线视频 | 青青草视频在线视频观看| 一级毛片久久久久久久久女| 成人漫画全彩无遮挡| 综合色丁香网| 91精品国产国语对白视频| 亚洲,一卡二卡三卡| 久久99热这里只频精品6学生| 亚洲色图综合在线观看| 亚洲成人手机| 国语对白做爰xxxⅹ性视频网站| av不卡在线播放| 大香蕉97超碰在线| 老司机影院成人| 精品人妻熟女av久视频| 日韩 亚洲 欧美在线| 日韩一区二区三区影片| 超碰97精品在线观看| 蜜桃久久精品国产亚洲av| 欧美高清性xxxxhd video| 久久精品国产a三级三级三级| 新久久久久国产一级毛片| 国产色婷婷99| 99久久精品一区二区三区| 国产精品国产av在线观看| 亚洲国产精品999| 伊人久久精品亚洲午夜| 国产黄片美女视频| 午夜福利在线观看免费完整高清在| 亚洲第一区二区三区不卡| 大话2 男鬼变身卡| 中文字幕av成人在线电影| 在线观看免费日韩欧美大片 | 性色av一级| 婷婷色av中文字幕| 成年av动漫网址| videos熟女内射| 精品一区二区三卡| 国产又色又爽无遮挡免| 十八禁网站网址无遮挡 | 在线精品无人区一区二区三 | 欧美日韩视频高清一区二区三区二| 啦啦啦中文免费视频观看日本| 六月丁香七月| 一本一本综合久久| 秋霞在线观看毛片| 极品教师在线视频| 亚洲精品乱码久久久v下载方式| videos熟女内射| 国产欧美日韩精品一区二区| 午夜免费男女啪啪视频观看| 91精品一卡2卡3卡4卡| h日本视频在线播放| 色网站视频免费| 国产高潮美女av| 2022亚洲国产成人精品| 最黄视频免费看| 成人国产av品久久久| 一边亲一边摸免费视频| 亚洲精品国产成人久久av| 观看美女的网站| 天天躁日日操中文字幕| 免费观看的影片在线观看| 欧美精品亚洲一区二区| 婷婷色av中文字幕| 亚洲精品一区蜜桃| 美女中出高潮动态图| 91精品一卡2卡3卡4卡| 看十八女毛片水多多多| 一级毛片黄色毛片免费观看视频| av网站免费在线观看视频| 久久国产乱子免费精品| 十分钟在线观看高清视频www | 日韩欧美 国产精品| av在线观看视频网站免费| 色婷婷av一区二区三区视频| 国产亚洲最大av| 欧美zozozo另类| 丝袜脚勾引网站| 美女视频免费永久观看网站| 在线观看一区二区三区激情| 国产成人午夜福利电影在线观看| 亚洲最大成人中文| 日本-黄色视频高清免费观看| 亚洲精品视频女| 日韩不卡一区二区三区视频在线| 中文精品一卡2卡3卡4更新| 国产精品女同一区二区软件| 丰满少妇做爰视频| 婷婷色av中文字幕| 久久国产精品男人的天堂亚洲 | 91狼人影院| 亚洲欧美一区二区三区黑人 | 国产久久久一区二区三区| 亚洲三级黄色毛片| 国产在线视频一区二区| 人妻夜夜爽99麻豆av| 国产精品偷伦视频观看了| 亚洲国产av新网站| 一级毛片电影观看| 最后的刺客免费高清国语| 久久精品夜色国产| 欧美极品一区二区三区四区| 少妇精品久久久久久久| 亚洲国产精品999| 国产熟女欧美一区二区| 2022亚洲国产成人精品| 一级毛片 在线播放| 国国产精品蜜臀av免费| 男男h啪啪无遮挡| 久久久色成人| 亚洲精品国产av蜜桃| 在线观看免费视频网站a站| 日韩,欧美,国产一区二区三区| 永久免费av网站大全| 欧美日韩视频精品一区| 免费不卡的大黄色大毛片视频在线观看| 在线观看国产h片| 我要看黄色一级片免费的| 免费看日本二区| 久久婷婷青草| 亚洲欧美日韩东京热| 国产精品一区二区性色av| 日本午夜av视频| 一级a做视频免费观看| 美女福利国产在线 | 久久久a久久爽久久v久久| 高清视频免费观看一区二区| 成年美女黄网站色视频大全免费 | 久久久久性生活片| 日日啪夜夜爽| 欧美日本视频| 天堂中文最新版在线下载| 日韩不卡一区二区三区视频在线| 一级a做视频免费观看| 一区二区三区乱码不卡18| 国产无遮挡羞羞视频在线观看| 免费少妇av软件| 亚洲经典国产精华液单| av在线观看视频网站免费| 国产精品人妻久久久影院| 亚洲欧美日韩卡通动漫| 亚洲国产精品成人久久小说| 在线观看人妻少妇| www.av在线官网国产| 中文字幕av成人在线电影| 美女国产视频在线观看| 亚洲欧美日韩卡通动漫| 天堂俺去俺来也www色官网| 老师上课跳d突然被开到最大视频| 日韩欧美一区视频在线观看 | 国产精品一区二区三区四区免费观看| 亚洲精品aⅴ在线观看| 日韩伦理黄色片| 99九九线精品视频在线观看视频| 我要看黄色一级片免费的| 亚洲精品乱码久久久v下载方式| 18+在线观看网站| 久久人妻熟女aⅴ| 又大又黄又爽视频免费| 99九九线精品视频在线观看视频| 中文字幕av成人在线电影| 草草在线视频免费看| 久久99热这里只频精品6学生| 狂野欧美激情性bbbbbb| 一个人免费看片子| 五月开心婷婷网| 久久99蜜桃精品久久| 亚洲精品色激情综合| 能在线免费看毛片的网站| 狂野欧美激情性bbbbbb| 亚洲三级黄色毛片| 成人亚洲精品一区在线观看 | 在线观看三级黄色| 久久午夜福利片| 99久久人妻综合| 亚洲国产日韩一区二区| 午夜福利在线在线| 少妇的逼好多水| 好男人视频免费观看在线| 久久久久久久国产电影| 亚洲无线观看免费| 久久热精品热| 久久久国产一区二区| 亚洲综合精品二区| 中文字幕免费在线视频6| 99国产精品免费福利视频| 亚洲精品乱码久久久久久按摩| 日韩三级伦理在线观看| 天天躁日日操中文字幕| 国产一级毛片在线| 亚洲性久久影院| 婷婷色麻豆天堂久久| 午夜激情久久久久久久| 午夜福利影视在线免费观看| 99热这里只有精品一区| 亚洲,欧美,日韩| 久久99蜜桃精品久久| 国产高清国产精品国产三级 | 夜夜看夜夜爽夜夜摸| 26uuu在线亚洲综合色| 久久久久久久精品精品| 性色avwww在线观看| 老司机影院成人| 国产精品久久久久久久久免| 国产久久久一区二区三区| 国产欧美亚洲国产| 哪个播放器可以免费观看大片| 国产黄片视频在线免费观看| 久久精品国产亚洲网站| 日韩欧美一区视频在线观看 | 一个人看的www免费观看视频| 欧美bdsm另类| 日韩成人av中文字幕在线观看| 亚洲精品第二区| 国产伦精品一区二区三区四那| 男女免费视频国产| 十分钟在线观看高清视频www | 国产精品av视频在线免费观看| 天堂8中文在线网| 麻豆精品久久久久久蜜桃| 蜜桃在线观看..| 国产精品偷伦视频观看了| 久久久久人妻精品一区果冻| 国产美女午夜福利| 另类亚洲欧美激情| 日韩av在线免费看完整版不卡| 一边亲一边摸免费视频| 五月天丁香电影| 国产免费又黄又爽又色| 永久免费av网站大全| 亚洲av福利一区| 2021少妇久久久久久久久久久| 国产大屁股一区二区在线视频| 黑丝袜美女国产一区| 久久久久久伊人网av| 美女内射精品一级片tv| 熟妇人妻不卡中文字幕| 一个人看的www免费观看视频| 男女下面进入的视频免费午夜| 免费观看性生交大片5| 嘟嘟电影网在线观看| 色5月婷婷丁香| 国产 一区精品| 免费大片18禁| 婷婷色av中文字幕| 伊人久久精品亚洲午夜| 国产 一区精品| 亚洲aⅴ乱码一区二区在线播放| 在线亚洲精品国产二区图片欧美 | 久久精品久久精品一区二区三区| 男人狂女人下面高潮的视频| 亚洲,欧美,日韩| 五月伊人婷婷丁香| 91精品伊人久久大香线蕉| 亚洲图色成人| 国产精品一区二区在线观看99| 亚洲av成人精品一二三区| 女人十人毛片免费观看3o分钟| 熟妇人妻不卡中文字幕| 国产av精品麻豆| 日韩三级伦理在线观看| 国产老妇伦熟女老妇高清| a级一级毛片免费在线观看| 汤姆久久久久久久影院中文字幕| 国产亚洲最大av| 日本黄色片子视频| 久久精品国产亚洲av涩爱| 少妇丰满av| av卡一久久| 亚洲美女搞黄在线观看| 久久国产精品大桥未久av | 日韩制服骚丝袜av| 精品久久久久久久末码| 日日啪夜夜爽| 男女免费视频国产| 日韩av免费高清视频| 少妇 在线观看| 久久久久久久久久成人| 日韩伦理黄色片| 一级毛片我不卡| 夜夜爽夜夜爽视频| 亚洲激情五月婷婷啪啪| 在线观看美女被高潮喷水网站| 偷拍熟女少妇极品色| 日本黄色片子视频| 大香蕉久久网| 久久99热这里只有精品18| 日韩免费高清中文字幕av| 国产色婷婷99| 91午夜精品亚洲一区二区三区| 成人漫画全彩无遮挡| 蜜桃亚洲精品一区二区三区| 国产白丝娇喘喷水9色精品| 亚洲av男天堂| 中文乱码字字幕精品一区二区三区| 成人午夜精彩视频在线观看| 国产精品福利在线免费观看| 亚洲aⅴ乱码一区二区在线播放| av黄色大香蕉| 中文天堂在线官网| 美女xxoo啪啪120秒动态图| 晚上一个人看的免费电影| 看十八女毛片水多多多| 免费不卡的大黄色大毛片视频在线观看| 日本欧美国产在线视频| 中国美白少妇内射xxxbb| 黄片wwwwww| 亚洲国产最新在线播放| 久久这里有精品视频免费| 黄片无遮挡物在线观看| 欧美xxxx黑人xx丫x性爽| 国产色爽女视频免费观看| 亚洲av欧美aⅴ国产| 偷拍熟女少妇极品色| 亚洲av不卡在线观看| 成年美女黄网站色视频大全免费 | 日韩电影二区| 自拍偷自拍亚洲精品老妇| 女人十人毛片免费观看3o分钟| 乱码一卡2卡4卡精品| 日韩av不卡免费在线播放| 观看美女的网站| 国产日韩欧美亚洲二区| 天堂中文最新版在线下载| 偷拍熟女少妇极品色| 直男gayav资源| 亚洲av日韩在线播放| 久久综合国产亚洲精品| 新久久久久国产一级毛片| 久久热精品热| 校园人妻丝袜中文字幕| 精品亚洲乱码少妇综合久久| 男的添女的下面高潮视频| 久久97久久精品| 亚洲国产毛片av蜜桃av| 国产永久视频网站| 亚洲色图综合在线观看| 久久久久久久亚洲中文字幕| 汤姆久久久久久久影院中文字幕| 99久久综合免费| 青春草视频在线免费观看| 亚洲欧美日韩无卡精品| 婷婷色综合www| 尾随美女入室| 亚洲伊人久久精品综合| 国产 精品1| 日韩国内少妇激情av| 亚州av有码| 妹子高潮喷水视频| 欧美日韩综合久久久久久| 黑人猛操日本美女一级片| 香蕉精品网在线| 亚洲av日韩在线播放| 天美传媒精品一区二区| 尤物成人国产欧美一区二区三区| 亚洲av欧美aⅴ国产| 91精品国产国语对白视频| 久久婷婷青草| 少妇丰满av| 欧美精品亚洲一区二区| 五月开心婷婷网| 国产亚洲午夜精品一区二区久久| 欧美成人精品欧美一级黄| 日本黄大片高清| 久久99热这里只频精品6学生| 国产色婷婷99| 日韩中字成人| 亚洲国产精品成人久久小说| 亚洲久久久国产精品| 中文精品一卡2卡3卡4更新| 欧美成人精品欧美一级黄| av.在线天堂| 日本爱情动作片www.在线观看| 欧美一级a爱片免费观看看| 亚洲熟女精品中文字幕| 五月天丁香电影| 色综合色国产| av一本久久久久| 久久久久久久久久久丰满| 亚洲精品久久久久久婷婷小说| 欧美成人一区二区免费高清观看| 欧美变态另类bdsm刘玥| 99热网站在线观看| 少妇熟女欧美另类| 在线天堂最新版资源| 综合色丁香网| 联通29元200g的流量卡| 日韩精品有码人妻一区| 久久久久国产精品人妻一区二区| 欧美区成人在线视频| 一级毛片久久久久久久久女| 成人综合一区亚洲| 最新中文字幕久久久久| 亚洲精品第二区| 欧美区成人在线视频| 亚洲精品久久午夜乱码| 亚洲丝袜综合中文字幕| www.色视频.com| 日本黄色片子视频| 人妻夜夜爽99麻豆av| 亚洲aⅴ乱码一区二区在线播放| 人人妻人人看人人澡| 色吧在线观看| 精品人妻偷拍中文字幕| 日韩成人av中文字幕在线观看| 国产色婷婷99| 国产精品秋霞免费鲁丝片| 99热这里只有是精品在线观看| 精品熟女少妇av免费看| 免费观看无遮挡的男女| 免费看不卡的av| 国产精品人妻久久久影院| 熟妇人妻不卡中文字幕| 日日啪夜夜撸| 亚洲精品日本国产第一区| 精品酒店卫生间| 成人特级av手机在线观看| 中文字幕人妻熟人妻熟丝袜美| 少妇人妻一区二区三区视频| 亚洲国产色片| 九九在线视频观看精品| 最近中文字幕高清免费大全6| 少妇的逼水好多| 中国三级夫妇交换| 18禁动态无遮挡网站| 久久热精品热| 大香蕉97超碰在线| 少妇高潮的动态图| 免费大片18禁| 精品一品国产午夜福利视频| 欧美变态另类bdsm刘玥| 免费不卡的大黄色大毛片视频在线观看| 国产精品伦人一区二区| 亚洲av国产av综合av卡| 男人和女人高潮做爰伦理| 国产白丝娇喘喷水9色精品| 综合色丁香网| 欧美变态另类bdsm刘玥| 国产精品人妻久久久影院| 亚洲av在线观看美女高潮| 精品人妻视频免费看| 亚洲欧美日韩另类电影网站 | 菩萨蛮人人尽说江南好唐韦庄| 蜜桃亚洲精品一区二区三区| 天天躁夜夜躁狠狠久久av| 99久国产av精品国产电影| 大陆偷拍与自拍| 啦啦啦中文免费视频观看日本| 久久av网站| 久久ye,这里只有精品| 成年女人在线观看亚洲视频| 十分钟在线观看高清视频www | 日本av手机在线免费观看| 在线播放无遮挡| 97在线视频观看| 国产日韩欧美在线精品| 十分钟在线观看高清视频www | 日日摸夜夜添夜夜爱| 久久久久久久久久成人| 91久久精品国产一区二区三区| 青春草视频在线免费观看| 精品午夜福利在线看| 在线观看人妻少妇| 亚洲国产av新网站| 欧美极品一区二区三区四区| 精华霜和精华液先用哪个| 久久久久性生活片| 午夜福利高清视频| 人妻少妇偷人精品九色| 亚洲欧美一区二区三区黑人 | 亚洲av欧美aⅴ国产| 国内精品宾馆在线| 免费观看在线日韩| 插阴视频在线观看视频| 久久国产精品大桥未久av | 亚洲,一卡二卡三卡| 日韩国内少妇激情av| 青春草亚洲视频在线观看| 日韩在线高清观看一区二区三区| 国产精品爽爽va在线观看网站| 午夜免费观看性视频| 2018国产大陆天天弄谢| 国产高清三级在线| 一本色道久久久久久精品综合| 极品少妇高潮喷水抽搐| 国产成人91sexporn| 久久精品国产a三级三级三级| 美女主播在线视频| 久久久久久久大尺度免费视频| 成人黄色视频免费在线看| 免费看av在线观看网站| 一区二区三区精品91| 久久久久性生活片| 免费人妻精品一区二区三区视频| 一级黄片播放器| 婷婷色麻豆天堂久久| 国产淫片久久久久久久久| 纵有疾风起免费观看全集完整版| 亚洲精品日本国产第一区| 三级国产精品欧美在线观看| 国产精品熟女久久久久浪| 免费观看av网站的网址| 亚洲欧美清纯卡通| 午夜福利在线观看免费完整高清在| 亚洲四区av| 我的老师免费观看完整版| 国产片特级美女逼逼视频| 国产永久视频网站| 免费黄色在线免费观看| 欧美精品一区二区大全| 国产片特级美女逼逼视频| 纵有疾风起免费观看全集完整版| 国产精品久久久久久精品电影小说 | 色综合色国产| 久久韩国三级中文字幕| 亚洲色图av天堂| 卡戴珊不雅视频在线播放| 亚洲经典国产精华液单| 日韩欧美精品免费久久| 欧美三级亚洲精品| 五月天丁香电影| 22中文网久久字幕| 韩国av在线不卡| 亚洲欧美日韩无卡精品| 特大巨黑吊av在线直播| 国内揄拍国产精品人妻在线| 国产乱人偷精品视频| 日韩大片免费观看网站| 夜夜爽夜夜爽视频| 国国产精品蜜臀av免费| 中文资源天堂在线| 免费观看性生交大片5| 看十八女毛片水多多多| 一级毛片我不卡| 亚洲欧美精品自产自拍| 国产人妻一区二区三区在| 亚洲精品日韩在线中文字幕| 亚洲欧美日韩另类电影网站 | 黑人猛操日本美女一级片| 日韩成人伦理影院| 久久 成人 亚洲| 99视频精品全部免费 在线| 日本欧美国产在线视频|