Ensemble models for identifying automatically aeolian saltating tracks driven by datasets

doi:10.7522/j.issn.1000-694X.2024.00049

Journal of Desert Research ›› 2024, Vol. 44 ›› Issue (6): 70-78.DOI: 10.7522/j.issn.1000-694X.2024.00049

Previous Articles Next Articles

Ensemble models for identifying automatically aeolian saltating tracks driven by datasets

Hongji Zhou¹(), Fanmin Mei¹(), Mengji Pu¹, Chuan Lin¹, Jin Su², Jinguang Chen³

^1.School of Environmental and Chemical Engineering /, Xi'an Polytechnic University，Xi'an 710600，China
^2.School of Science /, Xi'an Polytechnic University，Xi'an 710600，China
^3.School of Computer Science, Xi'an Polytechnic University，Xi'an 710600，China

Received:2024-03-11 Revised:2024-05-12 Online:2024-11-20 Published:2024-12-06
Contact: Fanmin Mei

Abstract

Abstract:

It is very vital for tracking sand particle to establish automatic identification of saltating tracks. Thus， the four ensemble models， including the Extremely randomized trees， the Random forests， the XGBoost， and the Gradient Boosting Decision Tree driven by the datasets we constructed， were proposed for identifying saltating tracks. Firstly， all the models perform well in spite of the dataset without very good discriminability， suggesting these models own an advantage when dealing with nonlinear relationships. Secondly， the Extremely randomized trees model holds the highest accuracy （0.9035）， precision （0.9030）， recall （0.9035）， F1 score （0.8995）， MCC （0.7378）， and AUC score （0.9179）， and time cost while the XGBoost model has the best balance between the higher scores and lower time cost. It implies that the former is most feasible for identifying offline saltating tracks and that the latter is prospective for tracking sand particle online. Finally， the improved datasets， which incorporate standard deviation of instant horizontal and vertical velocities， significantly enhance the predictive performances of Extremely randomized trees. This study effectively reduces the time cost of manual trajectory verification and broadens the application of machine learning in saltation.

Key words: aeolian saltating, extremely randomized trees, XGBoost, random forest, gradient boosting decision tree

CLC Number:

U418.5+6

Hongji Zhou, Fanmin Mei, Mengji Pu, Chuan Lin, Jin Su, Jinguang Chen. Ensemble models for identifying automatically aeolian saltating tracks driven by datasets[J]. Journal of Desert Research, 2024, 44(6): 70-78.

Figures/Tables 11

Fig.1 The 23rd frame of the M1 section （A）， the recovered saltating tracks from the M1 section via the KF-H-k （B）

Table 1 Parameterized features of saltating track samples

变量	定义
$X 1$	瞬时水平速度的平均值（m·s^-1）
$X 3$	瞬时垂直速度的平均值（m·s^-1）
$X 2$	瞬时合速度的平均值（m·s^-1）
$X 4$	轨迹抛物线方程拟合曲线的决定系数， $R 2$

Table 1 Parameterized features of saltating track samples

变量	定义
$X 1$	瞬时水平速度的平均值（m·s^-1）
$X 3$	瞬时垂直速度的平均值（m·s^-1）
$X 2$	瞬时合速度的平均值（m·s^-1）
$X 4$	轨迹抛物线方程拟合曲线的决定系数， $R 2$

Fig.2 The 27th true ascent trajectory （A）， the 7th true descent trajectory （B） and the 16th false trajectory （C）in video M1 extracted by the KF-H algorithm， respectively； Probability distribution of the determination coefficient （R2 ） for fitting quadratic curves of 5 756 true trajectories （D）

Fig.3 Workflow for model Training

Table 2 Final hyperparameter settings for each model optimized by Tree-structured Parzen Estimator

算法	HP₁	HP₂	HP₃	HP₄	HP₅	HP₆	HP₇	HP₈	HP₉	HP₁₀
随机森林	True	-1	True	180	18			True	“sqrt”	“gini”
梯度提升决策树	True	-1	True	210	3	0.3
XGBoost	True	-1		70	6	0.3	0.5
极度随机树	True	-1	True	186	21	0.1		True	None	“entropy”

Fig.4 Correlation matrix of features and label （the value of each matrix element represents the P-value of the significance test and the corresponding color represents the correlation coefficient）

Fig.5 PCA-based dimensionality reduction visualization

Table 3 Predictive performance of four ensemble models

模型	准确率	精准度	召回率	F1分数	MCC
随机森林	0.8923	0.8939	0.8923	0.8857	0.7056
极度随机树	0.9035	0.9030	0.9035	0.8995	0.7378
梯度提升决策树	0.8976	0.8970	0.8976	0.8929	0.7207
XGBoost	0.8995	0.9012	0.8995	0.8939	0.7267

Fig.6 ROC curves of four algorithms

Fig.7 Average time cost of the ensemble models

Table 4 Performances of the trained Extremely Randomized Trees by optimized datasets

优化数据集	准确率	精确度	召回率	F1分数	MCC	AUC分数
原始数据集	0.9035	0.9030	0.9035	0.8995	0.7378	0.9179
原始数据集+瞬时水平速度的方差	0.9332	0.9327	0.9332	0.9329	0.8241	0.9506
原始数据集+瞬时水平速度的方差+瞬时垂直速度的方差	0.9379	0.9372	0.9379	0.9374	0.8356	0.9697

References 44

1	Bagnold R A.The Physics of Blown Sand and Desert Dunes［M］.Netherlands：Springer，1942.
2	Wang D， Wang Y， Yang B，et al.Statistical analysis of sand grain/bed collision process recorded by high‐speed digital camera［J］.Sedimentology，2008，55（2）：461-470.
3	Jiang C W， Parteli E J R， Dong Z B，et al.Wind-tunnel experiments of aeolian sand transport reveal a bimodal probability distribution function for the particle lift-off velocities［J］.Catena，2022，217：106496.
4	Zhang Y， Wang Y， Jia P.Measuring the kinetic parameters of saltating sand grains using a high-speed digital camera［J］.Science China Physics，Mechanics Astronomy，2014，57：1137-1143.
5	O'Brien P， Neuman C M K.PTV measurement of the spanwise component of aeolian transport in steady state［J］.Aeolian Research，2016，20：126-138.
6	O'Brien P， Neuman C M K.An experimental study of the dynamics of saltation within a three-dimensional framework［J］.Aeolian Research，2018，31：62-71.
7	Yang B， Wang Y， Zhang Y.The 3-D spread of saltation sand over a flat bed surface in aeolian sand transport［J］.Advanced Powder Technology，2009，20（4）：303-309.
8	O'Brien P， McKenna Neuman C.Experimental validation of the near‐bed particle‐borne stress profile in aeolian transport systems［J］.Journal of Geophysical Research：Earth Surface，2019，124（11）：2463-2474.
9	Kang L Q， Zou X Y， Zhao G D，et al.Wind tunnel investigation of horizontal and vertical sand fluxes of ascending and descending sand particles in aeolian sand transport［J］.Earth Surface Processes and Landforms，2016，41（12）：1647-1657.
10	Ho T D， Valance A， Dupont P，et al.Scaling laws in aeolian sand transport［J］.Physical Review Letters，2011，106（9）：094501.
11	Zhang Y， Li M， Wang Y，et al.Reinvestigation of the scaling law of the windblown sand launch velocity with a wind tunnel experiment［J］.Journal of Arid Land，2019，11：664-673.
12	Creyssels M， Dupont P， El Moctar A O，et al.Saltating particles in a turbulent boundary layer：experiment and theory［J］.Journal of Fluid Mechanics，2009，625：47-74.
13	Mei F M， Zhou H J， Su J，et al.A new hybrid algorithm based on Kalman filter-Hungarian algorithm for tracking aeolian saltating particle in the high-speed video［J］.Earth Surface Processes and Landforms， 2024， .
14	Breiman L.Random forests［J］.Machine Learning，2001，45：5-32.
15	Geurts P， Ernst D， Wehenkel L.Extremely randomized trees［J］.Machine Learning，2006，63：3-42.
16	Friedman J H.Greedy function approximation：a gradient boosting machine［J］.Annals of Statistics，2001：1189-1232.
17	Chen T， Guestrin C.Xgboost：a scalable tree boosting system［C］//Proceedings of the 22nd Acm Sigkdd International Conference on Knowledge Discovery and Data Mining.2016：785-794.
18	Boroughani M， Pourhashemi S， Gholami H，et al.Predicting of dust storm source by combining remote sensing，statistic-based predictive models and game theory in the Sistan watershed，southwestern Asia［J］.Journal of Arid Land，2021，13（11）：1103-1121.
19	Boroughani M， Pourhashemi S， Hashemi H，et al.Application of remote sensing techniques and machine learning algorithms in dust source detection and dust source susceptibility mapping［J］.Ecological Informatics，2020，56：101059.
20	Choubin B， Hosseini F S， Rahmati O，et al.Mapping of salty aeolian dust-source potential areas：Ensemble model or benchmark models？［J］.Science of The Total Environment，2023，877：163419.
21	Gholami H， Mohamadifar A， Sorooshian A，et al.Machine-learning algorithms for predicting land susceptibility to dust emissions：the case of the Jazmurian Basin，Iran［J］.Atmospheric Pollution Research，2020，11（8）：1303-1315.
22	Rahmati O， Mohammadi F， Ghiasi S S，et al.Identifying sources of dust aerosol using a new framework based on remote sensing and modelling［J］.Science of The Total Environment，2020，737：139508.
23	Iban M C， Bilgilioglu S S.Snow avalanche susceptibility mapping using novel tree-based machine learning algorithms （XGBoost，NGBoost，and LightGBM） with eXplainable Artificial Intelligence （XAI） approach［J］.Stochastic Environmental Research and Risk Assessment，2023，37（6）：2243-2270.
24	Zafari A， Zurita-Milla R， Izquierdo-Verdiguier E.Land cover classification using extremely randomized trees：a kernel perspective［J］.IEEE Geoscience and Remote Sensing Letters，2019，17（10）：1702-1706.
25	李森，颜长珍.基于ChinaCover数据集的绿洲结构数据制图：以河西内陆河流域为例［J］.中国沙漠，2023，43（3）：230-242.
26	蒋小芳，徐青霞，段翰晨，等.黄河景电灌区土壤盐渍化反演的多模型对比［J］.中国沙漠，2023，43（5）：18-30.
27	吴敏，温小虎，冯起，等.基于随机森林模型的干旱绿洲区张掖盆地地下水水质评价［J］.中国沙漠，2018，38（3）：657-663.
28	张亦然，刘廷玺，童新，等.基于多源遥感和机器学习方法的科尔沁沙地植被覆盖度反演［J］.中国沙漠，2022，42（3）：187-195.
29	Houghton J E， Nichols T E， Griffiths J，et al.Automated classification of estuarine sub‐depositional environment using sediment texture［J］.Journal of Geophysical Research：Earth Surface，2023，128（2）：e2022JF006891.
30	Nichols T E， Worden R H， Houghton J E，et al.Sediment texture and geochemistry as predictors of sub-depositional environment in a modern estuary using machine learning：a framework for investigating clay-coated sand grains［J］.Sedimentary Geology，2023，458：106530.
31	Zheng，D Y， Hou M C， Chen A Q，et al.Application of machine learning in the identification of fluvial-lacustrine lithofacies from well logs：a case study from Sichuan Basin，China［J］.Journal of Petroleum Science and Engineering，2022，215，110610.
32	Bergstra J， Bardenet R， Bengio Y，et al.Algorithms for hyper-parameter optimization［C］//International Conference on Neural Information Processing Systems.2011.
33	Beguería S.Validation and evaluation of predictive models in hazard assessment and risk management［J］.Natural Hazards，2006，37：315-329.
34	Fawcett T.An introduction to ROC analysis［J］.Pattern recognition letters，2006，27（8）：861-874.
35	al Pedregosaet.Scikit-learn：machine learning in python［J］.Journal of Machine Learning Research，2011，12：2825-2830.
36	Canbek G， Sagiroglu S， Temizel T T，et al.Binary classification performance measures/metrics：a comprehensive visualized roadmap to gain new insights［C］//2017 International Conference on Computer Science and Engineering（UBMK）.IEEE，2017：821-826.
37	Chicco D， Jurman G.The advantages of the Matthews correlation coefficient（MCC）over F1 score and accuracy in binary classification evaluation［J］.BMC Genomics，2020，21（1）：1-13.
38	Silla C N， Freitas A A.A survey of hierarchical classification across different application domains［J］.Data Mining and Knowledge Discovery，2011，22：31-72.
39	Akay H.Spatial modeling of snow avalanche susceptibility using hybrid and ensemble machine learning techniques［J］.CATENA，2021，206：105524.
40	Yang J M， He Q， Liu Y.Winter-Spring prediction of snow avalanche susceptibility using optimisation multi-source heterogeneous factors in the Western Tianshan Mountains，China［J］.Remote Sensing，2022，14（6）：1340.
41	Duan T， Anand A， Ding D Y，et al.Ngboost：natural gradient boosting for probabilistic prediction［C］//International Conference on Machine Learning.PMLR，2020：2690-2700.
42	MacQueen J.Some methods for classification and analysis of multivariate observations［C］//Proceedings of the Fifth Berkeley Symposium on Mathematical Statistics and Probability.1967：281-297.
43	Ester M， Kriegel H P， Sander J，et al.A density-based algorithm for discovering clusters in large spatial databases with noise［C］//KDD'96：Proceedings of the Second International Conference on Knowledge Discovery and Data Mining.1996：226-231.
44	Bian J， Tian D， Tang Y，et al.Trajectory data classification：a review［J］.ACM Transactions on Intelligent Systems and Technology（TIST），2019，10（4）：1-34.

[1]	Ting Ning, Dinghai Zhang, Youyi Zhao, Jing Jiang. Relationship between soil moisture and topography and vegetation in the Tengger Desert [J]. Journal of Desert Research, 2024, 44(5): 133-142.
[2]	Fanmin Mei, Qianwen Yang, Wang Li, Jin Su. An overview of analytical models of saltation [J]. Journal of Desert Research, 2024, 44(4): 14-23.
[3]	Teng Zhang, Yunfa Miao, Yaguo Zou, Ziyue Zhang, Guoping Feng. Classification and changes of vegetation in Sugan Lake wetland in the extreme arid region [J]. Journal of Desert Research, 2024, 44(4): 81-90.
[4]	Wu Min, Wen Xiaohu, Feng Qi, Yin Zhengliang, Yang Linshan. Assessment of Groundwater Quality Based on Random Forest Model in Arid Oasis Area [J]. Journal of Desert Research, 2018, 38(3): 657-663.

Ensemble models for identifying automatically aeolian saltating tracks driven by datasets

RichHTML

PDF

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 11

References 44

Related Articles 4

Recommended Articles

Metrics

Comments