为了治理黄土高原水土流失，国家实施了一系列水土保持措施，其中包括坡改梯措施。梯田作为黄土高原地区重要的一项水土保持措施，具有高效的保水、保土和保肥作用，对防治水土流失具有重要意义。传统梯田统计方法速度慢且精度低，亟需提出一种快速有效的梯田提取方法，科学准确的评估梯田对土壤侵蚀量估算的影响，为黄土高原水土保持工作和生态环境建设提供数据依据。

本文以延河流域为研究区，基于高分一号数据（GF-1），分别采用深度学习法和面向对象分类法提取梯田信息，综合利用GIS、RS、RUSLE模型等技术方法，定量评价梯田的水土保持作用。主要研究结果如下：

（1）基于深度卷积神经网络模型探讨了主干特征网络及损失函数最优提取梯田信息的策略组合。结果表明在PSPNet模型下利用ResNet50主干网络和Lovasz-Softmax损失函数提取梯田效果最优。ResNet50主干网络在训练集、验证集和测试集平均交并比分别比Mobilenet主干网络提高了6.19%、5.51%和13.35%，精度分别提高了5.67%、5.36%和12.37%。ResNet50主干网络提取精度明显优于Mobilenet主干网络。基于ResNet50主干网络分别与Log-softmax、CrossEntropyLoss和Lovasz-Softmax三种损失函数组合，结果表明利用Lovasz-Softmax损失函数在测试集平均交并比分别比Log-softmax和CrossEntropyLoss损失函数高6.23%和3.41%，精度分别提高了2.73%和2.12%。

（2）比较了深度学习方法和面向对象分类法提取梯田边界的精度，结果表明深度学习方法在提取梯田边界时精度更高。基于梯田的光谱、纹理、形态等特征，构建梯田的分类规则，其提取梯田边界的正判率和平均交并比分别为81.25%和78.31%；基于PSPNet模型在梯田信息提取时的正判率和平均交并比相对面向对象分类法分别提高了9.00%和9.01%。因此，利用深度学习方法提取梯田信息更准确。

（3）使用深度学习最优策略组合提取了延河流域梯田信息，定量分析了不同分类结果下梯田对土壤侵蚀量估算的影响。2015年延河流域梯田面积约307.56 km²。利用RUSLE模型定量估算梯田在减少土壤侵蚀量的贡献，结果表明2015年延河流域未考虑梯田的情况下土壤侵蚀量为76.84×10⁶t，考虑梯田作用后的土壤侵蚀量为69.77×10⁶t，梯田信息影响了约9.20%的土壤侵蚀量；考虑梯田后，微度侵蚀、轻度侵蚀和中度侵蚀面积占比分别增加了0.08%、2.36%和1.03%，强烈侵蚀、极强烈侵蚀以及剧烈侵蚀的面积占比分别减少了0.05%、0.83%和2.69%，梯田的建设可以有效减少土壤侵蚀量。此外，计算了不同分类精度对土壤侵蚀量估算的影响。基于深度学习提取的梯田信息与面向对象分类法提取的梯田信息评估的土壤侵蚀量差异达3.29×10⁴t，因此，准确提取梯田信息对评估延河流域的土壤侵蚀情况有着重要影响。

In order to control soil erosion on the Loess Plateau, the state has implemented a series of soil and water conservation measures, including the conversion of slopes into terraces. As an important soil and water conservation measure in the Loess Plateau region, terracing is of great significance to the prevention and control of soil erosion because of its efficient water, soil and fertilizer retention. Traditional terracing statistics methods are slow and low in accuracy, and there is an urgent need to propose a fast and effective terracing extraction method to scientifically and accurately assess the impact of terracing on soil erosion estimation, and provide a data basis for soil and water conservation work and ecological environment construction on the Loess Plateau.

This paper takes the Yanhe River basin as the study area, based on the high-score one data (GF-1), uses the deep learning method and object-oriented classification method to extract terraces information respectively, integrates GIS, RS, RUSLE model and other techniques to quantitatively evaluate the terraces' effects. The main research results are as follows.

(1) Based on the deep convolutional neural network model, a combination of the backbone feature network and three loss functions is explored for optimal extraction of terraces. The results show that the terraces are optimally extracted using the ResNet50 backbone network and the Lovasz-Softmax loss function under the PSPNet model. the average cross-merge ratio of the ResNet50 backbone network is 6.19%, 5.51% and 13.35% higher than that of the Mobilenet backbone network in the training, validation and test sets, respectively, and the accuracy is improved by 5.67%, 5.36% and 12.37% respectively. Therefore, the extraction accuracy of the ResNet50 backbone network is significantly better than that of the Mobilenet backbone network. Based on the combination of ResNet50 backbone network with three loss functions, Log-softmax, CrossEntropyLoss and Lovasz-Softmax, respectively, the terrace extraction accuracy was analysed, and the results showed that using the Lovasz-Softmax loss function was significantly better than Log-softmax and CrossEntropyLoss loss functions by 6.23% and 3.41%, and the accuracy by 2.73% and 2.12%, respectively. The experimental results show that the ResNet50 backbone network using the PSPNet model and the Lovasz-Softmax loss function is the best strategy for extracting terrace information.

(2) The accuracy of the deep learning method and the object-oriented classification method in extracting terrace boundaries were compared, and the results showed that the deep learning method was more accurate in extracting terrace boundaries. Based on the spectral, textural and morphological features of the terraces, the classification rules of the terraces were constructed, and the positive judgment rate and average intersection ratio of the extracted terraces were 81.25% and 78.31% respectively; the positive judgment rate and average intersection ratio of the PSPNet-based model in terrace information extraction were improved by 9.00% and 9.01% respectively compared with the object-oriented classification method. Therefore, the extraction of terrace information using deep learning methods is more accurate.

(3) The terraced fields in the Yanhe River Basin were extracted using a combination of deep learning optimal strategies to quantify the effect of terraces on soil erosion estimation under different classification results. terraces in the Yanhe River Basin covered an area of about 307.56 km² in 2015. The contribution of terraces in reducing soil erosion was quantitatively estimated using the RUSLE model, and the results showed that in 2015 the soil erosion in the Yanhe basin without considering terraces was 76.84×10⁶t, and the soil erosion after considering the role of terraces was 69.77×10⁶t, and the terrace information influenced about 9.20% of soil erosion; after considering terraces, the percentage of slightly eroded, lightly eroded and moderately After considering the terraces, the area share of slight erosion, light erosion and moderate erosion increased by 0.08%, 2.36% and 1.03% respectively, and the area share of strong erosion, very strong erosion and severe erosion decreased by 0.05%, 0.83% and 2.69% respectively, and the construction of terraces could effectively reduce the amount of soil erosion. In addition, the effects of different classification accuracies on soil erosion estimation were calculated. The difference in soil erosion assessed by terrace information extracted based on deep learning and terrace information extracted by object-oriented classification method amounted to 3.29×10⁴t. Therefore, accurate extraction of terrace information has an important impact on the assessment of soil erosion in the Yanhe river basin.

[1] Cerdà A, Jordán A, Zavala L, et al. The contribution of mulches to control high soil erosion rates in vineyards in Eastern Spain[C], EGU General Assembly Conference Abstracts. 2014, 16: 16127.

[2] 陈浩. 黄土高原退耕还林前后流域土壤侵蚀时空变化及驱动因素研究[D]. 西北农林科技大学, 2019.

[3] 刘国彬, 上官周平, 姚文艺,等. 黄土高原生态工程的生态成效[J]. 中国科学院院刊, 2017, 32(001):11-19.

[4] G.B. L. Soil conservation and sustainable agriculture on the Loess Plateau: Challenges and prospects [J]. AMBIO A Journal of the Human Environment, 1999, 28(8): 663-668.

[5] Ping Z, Anbang W, Xinbao Z, et al. Soil conservation and sustainable eco-environment in the Loess Plateau of China[J]. Environmental Earth Sciences, 2013, 68(3): 633-639.

[6] 吕小永. 陕西省黄土高原地区山水林田湖生态保护修复工程实践[J]. 中国资源综合利用, 2018, 36(10): 163-165.

[7] 李宗善, 杨磊, 王国梁, 等. 黄土高原水土流失治理现状、问题及对策[J]. 生态学报, 2019, 39(20): 7398-7409.

[8] 张雨果, 王飞, 孙文义, 等. 基于面向对象的SPOT卫星影像梯田信息提取研究[J]. 水土保持研究, 2016, 23(06): 345-351.

[9] 杨蕾. 基于Spot5遥感影像提取水土保持信息的研究[D]. 西北大学, 2006.

[10] 赵欣, 王晓晶, 赵院, 等. 国产高分一号卫星数据傅里叶变换提取梯田影像可行性分析 [J]. 中国水土保持, 2016, (01): 63-65+73.

[11] Eckert S, Ghebremicael S T, Hurni H, et al. Identification and classification of structural soil conservation measures based on very high resolution stereo satellite data [J]. Journal of environmental management, 2017, 193:592-606.

[12] 刘钊, 廖斐凡, 赵桐. 基于PSPNet的遥感影像城市建成区提取及其优化方法[J]. 国土资源遥感, 2020, 32(04): 84-89.

[13] 张兵. 遥感大数据时代与智能信息提取[J]. 武汉大学学报(信息科学版), 2018,43(12): 1861-1871.

[14] Wang L, Xu X, Dong H, et al. Multi-pixel simultaneous classification of PolSAR image using convolutional neural networks [J]. Sensors, 2018, 18(3): 769.

[15] Lei X, Pan H, Huang X. A dilated CNN model for image classification [J]. IEEE Access, 2019, 7:124087-124095.

[16] Wang J, Zheng Y, Wang M, et al. Object-Scale Adaptive Convolutional Neural Networks for High-Spatial Resolution Remote Sensing Image Classification[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2020, 14:283-299.

[17] 翟睿洁, 赵文武, 贾立志. 基于RUSLE、InVEST和USPED的土壤侵蚀量估算对比研究——以陕北延河流域为例[J]. 农业现代化研究, 2020, 41(06): 1059-1068.

[18] 陈蝶, 卫伟, 陈利顶, 等. 梯田生态系统服务与管理研究进展[J]. 山地学报, 2016, 34(03): 374-384.

[19] 赵艺学. 晋西沟坝地—梯田—坡耕地农业效应的比较[J]. 水土保持学报, 2000, (02): 75-78.

[20] 张元星. 流域水沙变化对水土保持梯田措施的响应研究[D]. 西北农林科技大学, 2014.

[21] 杨磊, 冯青郁, 陈利顶. 黄土高原水土保持工程措施的生态系统服务[J]. 资源科学, 2020,(01): 1-9.

[22] 高健翎, 张建国, 朱莉莉, 等. 黄土丘陵沟壑区水土保持关键措施变化特征[J]. 水土保持通报, 2019, 39(03): 114-118.

[23] 赵东波. 基于高分辨率遥感影像的水土保持措施信息提取和分析[D]. 西北农林科技大学, 2007.

[24] 寇权, 王愿昌, 吴永红, 等. 基于高分辨率卫星影像的梯田监测方法研究[J]. 中国水土保持, 2005, (12): 24-25.

[25] 吴卿. 基于高分辨率遥感影像的水土保持生态建设监测方法研究[D]. 北京林业大学, 2007.

[26] Sofia G, Bailly J-S, Chehata N, et al. Comparison of pleiades and LiDAR digital elevation models for terraces detection in farmlands [J]. IEEE Journal of selected topics in applied Earth observations and remote sensing, 2016, 9(4): 1567-1576.

[27] 王庆. 基于高分辨率遥感影像纹理特征的水土保持措施提取方法研究[D]. 西北大学, 2008.

[28] 于浩, 刘志红, 张晓萍, 等. 基于傅立叶变换的梯田纹理特征提取[J]. 国土资源遥感, 2008, (02): 39-42.

[29] Zhang Y, Shi M, Zhao X, et al. Methods for automatic identification and extraction of terraces from high spatial resolution satellite data (China-GF-1) [J]. International Soil and Water Conservation Research, 2017, 5(1): 17-25.

[30] Diaz-Varela R A, Zarco-Tejada P J, Angileri V, et al. Automatic identification of agricultural terraces through object-oriented analysis of very high resolution DSMs and multispectral imagery obtained from an unmanned aerial vehicle[J]. Journal of Environmental Management, 2014, 134:117-126.

[31] 侯群群. 面向对象的遥感影像水土保持措施信息提取研究[D]. 西北农林科技大学, 2013.

[32] 党恬敏, 穆兴民, 孙文义, 等. 高分辨率遥感影像梯田快速提取方法研究进展[J]. 人民黄河, 2017, 39(03): 85-89+94.

[33] 薛牡丹. 基于面向对象分析的无人机影像梯田田面提取研究[D]. 西北农林科技大学, 2018.

[34] 胡勇. 面向无人机影像和坡度数据的梯田田块提取方法研究[D]. 西北农林科技大学, 2018.

[35] 常立娜. 深度学习文献综述[J]. 开放学习研究, 2018, 23(02): 30-35.

[36] 梁泽毓. 基于深度学习的多源遥感水体信息提取方法及其应用研究[D]. 安徽大学, 2019.

[37] Kim Y. Convolutional neural networks for sentence classification[J]. Eprint ArXiv, 2014.

[38] 段琳琳. 基于深度学习的遥感影像分类研究[D]. 河南大学, 2018.

[39] Ma X, Fu A, Wang J, et al. Hyperspectral image classification based on deep deconvolution network with skip architecture[J]. IEEE Transactions on Geoscience and Remote Sensing, 2018, 56(8): 4781-4791.

[40] 范荣双, 陈洋, 徐启恒, 等. 基于深度学习的高分辨率遥感影像建筑物提取方法[J]. 测绘学报, 2019, 48(01): 34-41.

[41] 冯丽英. 基于深度学习技术的高分辨率遥感影像建设用地信息提取研究[D]. 浙江大学, 2017.

[42] Vakalopoulou M, Karantzalos K, Komodakis N, et al. Building detection in very high resolution multispectral data with deep learning features[C]//2015 IEEE International Geoscience and Remote Sensing Symposium (IGARSS). IEEE, 2015: 1873-1876.

[43] 钟传琦. 基于深度学习的高分辨率遥感影像道路自动提取研究[D]. 电子科技大学, 2018.

[44] Xu Y, Xie Z, Feng Y, et al. Road extraction from high-resolution remote sensing imagery using deep learning[J]. Remote Sensing, 2018, 10(9): 1461.

[45] Li P, Zang Y, Wang C, et al. Road network extraction via deep learning and line integral convolution[C]//2016 IEEE International Geoscience and Remote Sensing Symposium (IGARSS). IEEE, 2016: 1599-1602.

[46] 朱祺夫, 赵俊三, 陈磊士, 等. 基于深度学习的遥感影像城市建筑用地提取[J]. 软件导刊, 2018, 17(10): 18-21.

[47] Chen S W, Tao C S. PolSAR image classification using polarimetric-feature-driven deep convolutional neural network[J]. IEEE Geoscience and Remote Sensing Letters, 2018, 15(4): 627-631.

[48] Zhang W, Tang P, Zhao L. Fast and accurate land-cover classification on medium-resolution remote-sensing images using segmentation models[J]. International Journal of Remote Sensing, 2021, 42(9): 3277-3301.

[49] Ding P, Zhang Y, Jia P, et al. A comparison: Different DCNN models for intelligent object detection in remote sensing images[J]. Neural Processing Letters, 2019, 49(3): 1369-1379.

[50] Hinton G E, Krizhevsky A, Sutskver I. Imagenet classification with deep convolutional neural networks[J]. Advances in neural information processing systems, 2012, 25:1106-1114.

[51] Badrinarayanan V, Kendall A, Cipolla R. Segnet: A deep convolutional encoder-decoder architecture for image segmentation[J]. IEEE transactions on pattern analysis and machine intelligence, 2017, 39(12): 2481-2495.

[52] Zhao H, Shi J, Qi X, et al. Pyramid scene parsing network[C]. Proceedings of the IEEE conference on computer vision and pattern recognition. 2017: 2881-2890.

[53] 刘同星, 马明霞, 路长伦, 等. 基于无人机影像的冬小麦深度学习分类[J]. 中国农业资源与区划, 2020, 41(04): 150-158.

[54] 邵琥翔, 丁凤, 杨健, 等. 基于深度学习的黑臭水体遥感信息提取研究[J]. 长江科学院院报, 2021, 1-10.

[55] 刘钊, 赵桐, 廖斐凡, 等. 基于语义分割网络的高分遥感影像城市建成区提取方法研究与对比分析[J]. 国土资源遥感, 2021, 33(01): 45-53.

[56] 张宇婷, 肖海兵, 聂小东, 等. 基于文献计量分析的近30年国内外土壤侵蚀研究进展 [J]. 土壤学报, 2020, 57(04): 797-810.

[57] 张晶. SWAT模型在辽西地区土壤侵蚀模拟的应用[J]. 水土保持应用技术, 2020(03): 8-10.

[58] 陈童尧, 贾燕锋, 王佳楠, 等. 基于InVEST模型的祁连山国家级自然保护区土壤保持现状与功能[J]. 干旱区研究, 2020, 37(01): 150-159.

[59] 王森, 王海燕, 谢永生, 等. 延安市退耕还林前后土壤保持生态服务功能评价[J]. 水土保持研究, 2019, 26(01): 280-286.

[60] 党虹, 葛丽娟, 赵传燕, 等. 基于InVEST模型的称钩河流域土壤保持功能研究[J]. 兰州大学学报(自然科学版), 2018, 54(05): 633-639.

[61] 薛卓彬. 基于InVEST模型的延河流域生态系统服务功能评估[D]. 西北大学, 2017.

[62] 陈学林, 文仕知, 杨丽丽, 等. 基于USLE模型的长株潭城市群生态绿心区水土流失研究[J]. 水土保持通报, 2020, 40(03): 118-124.

[63] 黄硕文,李健,张欣佳,邓联文,张金萍.河南省近十年来土壤侵蚀时空变化分析[J].农业资源与环境学报,2021,38(02):232-240.

[64] 司家济. 基于ArcGIS和USLE模型的颍河流域土壤侵蚀研究[J]. 人民珠江, 2020, 41(06): 93-98.

[65] 陈豪, 丁文广, Zafar T B. 基于USLE模型的祁连山国家公园土壤水力侵蚀评价[J]. 中国水土保持科学, 2020, 18(04): 38-44.

[66] Alewell C, Borrelli P, Meusburger K, et al. Using the USLE: Chances, challenges and limitations of soil erosion modelling[J]. International soil and water conservation research, 2019, 7(3): 203-225.

[67] 刘斌涛, 宋春风, 史展,等. 西南土石山区土壤流失方程坡度因子修正算法研究[J]. 中国水土保持, 2015, 000(008):49-51.

[68] 李亚平, 卢小平, 张航, 等. 基于GIS和RUSLE的淮河流域土壤侵蚀研究——以信阳市商城县为例[J]. 国土资源遥感, 2019, 31(04): 243-249.

[69] Hao C, Oguchi T, Pan W U. Assessment for soil loss by using a scheme of alterative sub-models based on the RUSLE in a Karst Basin of Southwest China[J]. Journal of integrative agriculture, 2017, 16(2): 377-388.

[70] 潘美慧, 伍永秋, 任斐鹏, 等. 基于USLE的东江流域土壤侵蚀量估算[J]. 自然资源学报, 2010, 25(12): 2154-2164.

[71] 薛亚永, 王晓峰. 黄土高原森林草原区退耕还林还草土壤保持效应评估[J]. 干旱地区农业研究, 2017, 35(05): 122-128.

[72] 刘桐恺, 陈元哲, 董天翔. 基于GIS/RS的灾区土壤保持功能恢复效应评价[J]. 水土保持研究, 2019, 26(02): 78-85.

[73] Zhu X, Zhang R, Sun X. Spatiotemporal dynamics of soil erosion in the ecotone between the Loess Plateau and Western Qinling Mountains based on RUSLE modeling, GIS, and remote sensing[J]. Arabian Journal of Geosciences, 2021, 14(1): 1-12.

[74] 高海东, 李占斌, 李鹏, 等. 梯田建设和淤地坝淤积对土壤侵蚀影响的定量分析[J]. 地理学报, 2012, 67(05): 599-608.

[75] 张宏鸣, 杨勤科, 王猛, 等. 梯田区侵蚀地形因子随DEM分辨率变化的特征分析[J]. 农业机械学报, 2017, 48(10): 172-179.

[76] 王翊人, 赵牡丹, 张倩, 等. 梯田作为地形因子和工程措施对土壤侵蚀定量评价影响的对比研究[J]. 水土保持通报, 2017, 37(02): 148-152+57.

[77] 孙虎. 陕西延河流域地貌组合类型的模糊聚类划分[J]. 陕西师范大学学报(自然科学版), 1996, 24(04):79-84.

[78] 凯楠. 基于时空分辨率融合的延河流域植被季节变化分析[D]; 西北大学, 2018.

[79] McCool D K, Foster G R, Mutchler C K, et al. Revised slope length factor for the Universal Soil Loss Equation[J]. Transactions of the ASAE, 1989, 32(5): 1571-1576.

[80] Baoyuan L, Nearing M A, Risse L M. Slope gradient effects on soil loss for steep slopes[J]. Transactions of the ASAE, 1994, 37(6): 1835-1840.

[81] 蔡崇法, 丁树文, 史志华, 等. 应用USLE模型与地理信息系统IDRISI预测小流域土壤侵蚀量的研究[J]. 水土保持学报, 2000(02): 19-24.

[82] 李双才, 罗利芳, 张科利, 等. 黄土沟壑丘陵区退耕对土壤侵蚀影响的模拟研究[J]. 水土保持学报, 2004(01): 74-77+81.

[83] 罗舒元, 朱清科, 辛云玲, 等. 陕北吴起县植被覆盖度时空特征及其影响机制[J]. 中国水土保持科学, 2020, 18(03): 146-154.

[84] 王万忠，焦菊英. 中国的土壤侵蚀因子定量评价研[J]. 水土保持通报, 1996(05): 1-20.

[85] Renard K G, Foster G R, Weesies G A, et al. Predicting soil erosion by water: A guide to conservation planning with the Revised Universal Soil Loss Equation (RUSLE)[J]. Agriculture handbook, 1996, 703: 25-28.

[86] 刘文超,刘纪远,匡文慧.陕北地区退耕还林还草工程土壤保护效应的时空特征[J].地理学报,2019,74(09):1835-1852.

[87] 李妙莹. 基于InVEST模型的黄土高原人工灌草系统土壤保持功能评估[D], 兰州大学, 2019.

[88] Gao L, Chen P Y, Yu S. Demonstration of convolution kernel operation on resistive cross-point array[J]. IEEE Electron Device Letters, 2016, 37(7): 870-873.

[89] Simonyan K, Zisserman A. Very deep convolutional networks for large-scale image recognition[J]. arXiv preprint arXiv:1409.1556, 2014.

[90] Szegedy C, Ioffe S, Vanhoucke V, et al. Inception-ResNet and the impact of residual connections on learning[J]. arXiv preprint arXiv:1602.07261, 2016.

[91] He K, Zhang X, Ren S, et al. Deep residual learning for image recognition[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2016: 770-778.

[92] 陈启浩. 面向对象的多源遥感数据分类技术研究与实现[D], 中国地质大学, 2007.

[93] Sonka M, Hlavac V, Boyle R. Image Processing, Analysis and Machine Vision || Segmentation[J]. 1993, 10.1007/978-1-4899-3216-7(Chapter 5):112-191.

[94] 齐义娜. 面向对象的高分辨率遥感影像信息提取与尺度效应分析[D], 东北师范大学, 2009.

[95] 黄慧萍. 面向对象影像分析中的尺度问题研究[D], 中国科学院研究生院（遥感应用研究所）, 2003.

[96] 张春鹏. 基于高分辨率遥感影像的城市环境结构信息提取研究[D], 华东师范大学, 2007.

[97] 岳安志, 杨建宇, 超 张, 等. 基于多尺度形态学滤波的高分辨率遥感影像分割[J]. 农业工程学报, 2013, 29(25): 89-95.

[98] 张宏鸣, 杨勤科, 刘晴蕊, 等. 基于GIS的区域坡度坡长因子提取算法[J]. 计算机工程, 2010, 36(09): 246-248.

[99] 廖义善, 卓慕宁, 谢金波, 等. USLE/RUSLE模型中植被因子变化特征及其影响因素[J]. 生态学报, 2017, 37(06): 1987-1993.

[100]张岩, 刘宝元, 史培军, 等. 黄土高原土壤侵蚀作物覆盖因子计算[J]. 生态学报, 2001 (07): 1050-1056.

[101]Yan Z, Bao-Yuan L, Qing-Chun Z, et al. Effect of different vegetation types on soil erosion by water[J]. Journal of Integrative Plant Biology, 2003, 45(10): 1204.

[102]Tang Q, Xu Y, Bennett S J, et al. Assessment of soil erosion using RUSLE and GIS: a case study of the Yangou watershed in the Loess Plateau, China[J]. Environmental Earth Sciences, 2015, 73(4): 1715-1724.

[103]吴发启, 张玉斌, 王健. 黄土高原水平梯田的蓄水保土效益分析[J]. 中国水土保持科学, 2004(01): 34-37.

[104]齐清, 王天明, 寇晓军, 等. 基于GIS的黄土高原小流域土壤侵蚀定量评价[J]. 水土保持研究, 2009, 16(03): 1-5+37.

[105]包玉斌. 基于InVEST模型的陕北黄土高原生态服务功能时空变化研究[D]; 西北大学, 2015.