Importance Analysis of Vegetation Change Factors in East Africa Based on Machine Learning

Zhang Xiumei; Ma Bo; Zhang Yijie

doi:10.13961/j.cnki.stbctb.2023.06.028

您当前的位置：

首页 >

文章列表页 >

Importance Analysis of Vegetation Change Factors in East Africa Based on Machine Learning

- Importance Analysis of Vegetation Change Factors in East Africa Based on Machine Learning
- Bulletin of Soiland Water Conservation Vol. 43, Issue 6, Pages: 227-236(2023)
- 作者机构：
  
  1. 西北农林科技大学水土保持研究所, 陕西杨凌,712100
  2. 中国科学院水利部水土保持研究所, 陕西杨凌,712100
- 作者简介：
- 基金信息：
- DOI：10.13961/j.cnki.stbctb.2023.06.028
  CLC： F301.24
- Published：2023
- 稿件说明：
移动端阅览
Zhang Xiumei, Ma Bo, Zhang Yijie. Importance Analysis of Vegetation Change Factors in East Africa Based on Machine Learning[J]. Bulletin of Soiland Water Conservation, 2023, 43(6): 227-236.
DOI：

Zhang Xiumei, Ma Bo, Zhang Yijie. Importance Analysis of Vegetation Change Factors in East Africa Based on Machine Learning[J]. Bulletin of Soiland Water Conservation, 2023, 43(6): 227-236. DOI： 10.13961/j.cnki.stbctb.2023.06.028.

摘要

[目的

]

基于机器学习算法对东非植被变化进行因子重要性分析，测度不同算法在各情况下的精度差异及适用性，为保护、恢复和促进可持续森林管理、水土流失综合防治提供科学依据。[方法

]

以东非9个国家2001—2020年的归一化植被指数（normalized difference vegetation index，NDVI）变化为研究对象，选取影响东非植被变化的2个气候因子及5个人类活动因子作为自变量，利用随机森林（random forest，RF）、BP神经网络（BP neural networks，BP）、支持向量机（support vector machines，SVM）、遗传算法（genetic algorithm，GA）、径向基神经网络（radial basis function，RBF）、卷积神经网络（convolutional neural networks，CNN）6种机器学习算法建立NDVI预测模型，以决定系数（R

）、平均绝对误差（MAE，mean absolute error）、平均相对误差（MRE，mean relative error）3个指标评价评估6种机器学习算法预测NDVI变化的潜力，并基于所得的最优模型即对选取的7个因子进行重要性分析。[结果

]

精度验证结果表明，研究区内在全因子的情况下，CNN算法的回归精度最差；经逐轮删除一个综合表现不佳的算法后，RF算法建立的模型在东非NDVI变化分析中回归精度较高；基于随机森林算法的不同因子变量对NDVI变化的重要性表明，年降水量、N

O排放量、CH

排放量、牲畜数量4个变量对NDVI变化回归的结果影响较大。[结论

]

随机森林算法的回归能力在东非NDVI模拟中具有相对优势，降水量是影响植被变化最重要的气候因子，同时，温室气体的排放对于东非植被的变化也具有一定的影响。东非各国应提高植被变化对气候环境、社会经济和政治制度相互依存关系的认识和理解，并制定适当的政策以促进可持续森林管理、防治荒漠化。

Abstract

[Objective] A factor importance analysis of vegetation changes in East Africa based on different machine learning algorithms was conducted to measure the accuracy and applicability of the different algorithms in order to provide a scientific basis for protecting

restoring

and promoting sustainable forest management and comprehensive prevention and control of soil erosion. [Methods] Changes in normalized difference vegetation index (NDVI) for nine countries in East Africa from 2001 to 2020 were determined. The independent treatment variables were two climatic factors and five human activity factors affecting vegetation changes in East Africa. Six machine learning algorithms were used to establish NDVI prediction models:random forest (RF)

BP neural networks (BP)

support vector machines (SVM)

genetic algorithm (GA)

radial basis function (RBF)

and convolutional neural networks (CNN). Coefficient of determination (R2)

mean absolute error (MAE)

and mean relative error (MRE) were used as error indicators to evaluate the potential of the six machine learning algorithms for predicting NDVI changes. Based on the optimal model (RF)

the importance of the selected seven factors was determined. [Results] The accuracy verification results showed that the regression accuracy of the CNN algorithm was the worst for the full factor case in the study area. After deleting an algorithm with poor comprehensive performance in each round of testing

the model established by the RF algorithm had the highest regression accuracy for NDVI change analysis in East Africa. The importance of different factor variables to NDVI change based on RF showed that annual precipitation

N2O emission

CH4 emission

and livestock number had the greatest influence on the results of the NDVI change regression. [Conclusion] RF had a comparative advantage for NDVI simulation in East Africa. Precipitation was the most important climatic factor affecting vegetation changes. At the same time

greenhouse gas emissions also had an important impact on vegetation changes in East Africa. East African countries should raise awareness and understanding of the interdependence of vegetation changes on climate

environment

socio-economic conditions

and political systems

and develop appropriate policies to promote sustainable forest management and to combat desertification.

关键词

Keywords

references

Kalisa W, Igbawua T, Henchiri M, et al.Assessment of climate impact on vegetation dynamics over East Africa from 1982 to 2015[J].Scientific reports, 2019,9(1):16865.

杜佳梦,包刚,佟斯琴,等.1982-2015年蒙古国植被覆盖变化及其与气候变化和人类活动的关系[J].草业学报,2021,30(2):1-13.

Ma Bo, Wang Shanshan, Mupenzi C, et al.Quantitative contributions of climate change and human activities to vegetation changes in the Upper White Nile River [J].Remote Sensing, 2021,13(18):3648.

郭敏杰.基于NDVI的黄土高原地区植被覆盖度对气候变化响应及定量分析[D].北京:中国科学院研究生院(教育部水土保持与生态环境研究中心),2014.

孙艳萍.黄土高原水蚀风蚀交错带植被覆盖动态变化及其与气候因子的关系[D].陕西杨凌:西北农林科技大学,2012.

Gao Mengdi, Piao Shilong, Chen Anping, et al.Divergent changes in the elevational gradient of vegetation activities over the last 30 years [J].Nature Communications, 2019,10(1):2970.

Fangang Yu, Price K P, Ellis J, et al.Response of seasonal vegetation development to climatic variations in Eastern Central Asia [J].Remote Sensing of Environment, 2003,87(1):42-54.

Meng Yanrong, Yang Mingxia, Liu Shan, et al.Quantitative assessment of the importance of bio-physical drivers of land cover change based on a random forest method [J].Ecological Informatics, 2021,61:101204.

Barrett A B, Duivenvoorden S, Salakpi E E, et al.Forecasting vegetation condition for drought early warning systems in pastoral communities in Kenya [J].Remote Sensing of Environment, 2020,248:111886.

Shi Yu, Jin Ning, Ma Xuanlong, et al.Attribution of climate and human activities to vegetation change in China using machine learning techniques [J].Agricultural and Forest Meteorology, 2020,294:108146.

Huang Shengzhi, Zheng Xudong, Ma Lan, et al.Quantitative contribution of climate change and human activities to vegetation cover variations based on GA-SVM model [J].Journal of Hydrology, 2020,584:124687.

Wu Shupu, Gao Xin, Lei Jiaqiang, et al.Spatial and temporal changes in the normalized difference vegetation index and their driving factors in the desert/grassland biome transition zone of the Sahel Region of Africa [J].Remote Sensing, 2020,12(24):4119.

Hawinkel P, Thiery W, Lhermitte S, et al.Vegetation response to precipitation variability in East Africa controlled by biogeographical factors [J].Journal of Geophysical Research:Biogeosciences, 2016,121(9):2422-2444.

Hoscilo A, Balzter H, Bartholomé E, et al.A conceptual model for assessing rainfall and vegetation trends in sub-Saharan Africa from satellite data [J].International Journal of Climatology, 2015,35(12):3582-3592.

Ayugi B, Tan G, Niu R, et al.Evaluation of meteorological drought and flood scenarios over Kenya, East Africa [J].Atmosphere, 2020,11(3):307.

Payet R, Obura D.The negative impacts of human activities in the Eastern African region:An international waters perspective [J].AMBIO:A Journal of the Human Environment, 2004,33(1):24-33.

Pricope N G, Husak G, Lopez C D, et al.The climate-population nexus in the East African Horn:Emerging degradation trends in rangeland and pastoral livelihood zones [J].Global Environmental Change, 2013,23(6):1525-1541.

Agutu N O, Awange J L, Zerihun A, et al.Assessing multi-satellite remote sensing, reanalysis, and land surface models' products in characterizing agricultural drought in East Africa [J].Remote Sensing of Environment, 2017,194:287-302.

Breiman L.Random forests [J].Machine learning, 2001,45:5-32.

Zhong Shifa, Zhang Kai, Bagheri M, et al.Machine learning:new ideas and tools in environmental science and engineering [J].Environmental Science & Technology, 2021,55(19):12741-12754.

Zipser D, Andersen R A.A back-propagation programmed network that simulates response properties of a subset of posterior parietal neurons [J].Nature, 1988,331(6158):679-684.

Xue Zhenghai, Feng Wenkai, Li Botao, et al.Landslide susceptibility mapping based on the coupling of two correlation methods and the BP neural network model:a case study of the Baihetan Reservoir area, China [J].Frontiers in Environmental Science, 2022,10:1039985.

Giorgos M, Jungho I, Caesar O.Support vector machines in remote sensing:a review [J].ISPRS Journal of Photogrammetry and Remote Sensing, 2011,66(3):247-259.

Corinna C, Vladimir V.Support-vector networks [J].Machine Learning, 1995,20:273-297.

Rodriguez G V, Sanchez C M, Chica O M, et al.Machine learning predictive models for mineral prospectivity:an evaluation of neural networks, random forest, regression trees and support vector machines [J].Ore Geology Reviews, 2015,71:804-818.

Dao S D, Abhary K, Marian R.Maximising performance of genetic algorithm solver in Matlab [J].Engineering Letters, 2016,24(1).

Wu Shengshen, Zhao Gaofeng, Wu Bisheng.Real-time prediction of the mechanical behavior of suction caisson during installation process using GA-BP neural network [J].Engineering Applications of Artificial Intelligence, 2022,116:105475.

Whirter J G M C, Broomhead D S, Shepherd T J.A systolic array for nonlinear adaptive filtering and pattern recognition [J].Journal of VLSI signal processing systems for signal, image and video technology, 1991,3(1-2):69-75.

Li Tang, He Kaiyue, Chen Yang, et al.Integral BLF-based adaptive neural constrained regulation for switched systems with unknown bounds on control gain [J].IEEE Transactions on Neural Networks and Learning Systems, 2022.

Petsche T, Dickinson B W.Trellis codes, receptive fields, and fault tolerant, self-repairing neural networks [J].IEEE Transactions on Neural Networks, 1990,1(2):154-166.

Kim H.Convolution neural network based degenerated mode decomposition via near-field images from linear and circular polarizers [J].Optics and Lasers in Engineering, 2023,160:107310.

Athanasopoulos G, Hyndman R J.Modelling and forecasting Australian domestic tourism [J].Tourism Management, 2007,29(1):19-31.

Yirga A, Addisu Legesse S, Mekuriaw A.Carbon stock and mitigation potentials of Zeghie natural forest for climate change disaster reduction, Blue Nile Basin, Ethiopia [J].Earth Systems and Environment, 2020,4(1):27-41.

Van Ittersum M K, Van Bussel L G J, Wolf J, et al.Can sub-Saharan Africa feed itself? [J].Proceedings of the National Academy of Sciences, 2016,113(52):14964-14969.

章明,张培松,刘洪斌,等.基于SPOT VEGETATION数据的海南岛年际植被变化研究[J].西南大学学报(自然科学版),2009,31(3):148-153.

卜灵心,来全,刘心怡.不同机器学习算法在草原草地生物量估算上的适应性研究[J].草地学报,2022,30(11):3156-3164.

Hamimeche M, Niculescu S, Billey A, et al.Identification and mapping of Algerian island vegetation using high-resolution images (Pléiades and SPOT 6/7) and random forest modeling [J].Environmental Monitoring and Assessment, 2021,193(9):617.

Gage E A, Cooper D J.Urban forest structure and land cover composition effects on land surface temperature in a semi-arid suburban area [J].Urban Forestry & Urban Greening, 2017,28:28-35.

Ali I, Cawkwell F, Dwyer E, et al.Satellite remote sensing of grasslands:from observation to management [J].Journal of Plant Ecology, 2016,9(6):649-671.

Barrett B, Nitze I, Green S, et al.Assessment of multi-temporal, multi-sensor radar and ancillary spatial data for grasslands monitoring in Ireland using machine learning approaches[J].Remote sensing of environment, 2014,152:109-124.

陈柯兵,李圣伟,何奇锴,等.关键水文信息对三峡水库调度决策的重要性分析[J].水电与新能源,2022,36(11):45-49.

张子慧,吴世新,赵子飞,等.基于机器学习算法的草地地上生物量估测:以祁连山草地为例[J].生态学报,2022,42(22):8953-8963.

何坤龙,赵伟,刘晓辉,等.云雾覆盖下地表温度重建机器学习模型的训练集敏感性分析[J].遥感学报,2021,25(8):1722-1734.

Nicholson S E.An analysis of recent rainfall conditions in Eastern Africa [J].International Journal of Climatology, 2016,36(1):526-532.

Hulme M, Doherty R, Ngara T, et al.African climate change:1900-2100[J].Climate research, 2001,17(2):145-168.

Pfeiffer M, Kumar D, Martens C, et al.Climate change will cause non-analog vegetation states in Africa and commit vegetation to long-term change [J].Biogeosciences, 2020.

Serneels S, Linderman M, Lambin E F.A multilevel analysis of the impact of land use on interannual land-cover change in East Africa [J].Ecosystems, 2007,10(3):402-418.

Elagouz M H, Abou-Shleel S M, Belal A A, et al.Detection of land use/cover change in Egyptian Nile Delta using remote sensing [J].The Egyptian Journal of Remote Sensing and Space Science, 2019,1(23):57-62.

Hosea M M, Stefan J, Sopan D P, et al.Relative contribution of land use change and climate variability on discharge of upper Mara River, Kenya [J].Journal of Hydrology:Regional Studies, 2016,5:244-260.

Views

537

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Evaluation of landslide susceptibility at Guilin City， Guangxi Zhuang Autonomous Region based on hyperparameter optimization and shapely additive explanations

Landslide susceptibility evaluation and driving force analysis for Jiuzhaigou scenic area based on explainable machine learning

Estimation of soil moisture and organic matter content in saline-alkaline farmland of Hetao Plain based on interpretable machine learning

NDVI Variations and Its Response to Extreme Climate in Xinjiang Uygur Autonomous Region During 2001-2017

Remote Sensing Estimation of Forest Volume in Typical Karst Mountainous Areas

Related Author

Zeng Mingyue

Su Huiwei

Liu Zihan

Zhao Peng

Pei Zengli

Lin Yongming

Wang Daojie

He Songtang

Related Institution

Guangxi Key Laboratory of Culture and Tourism SmartTechnology， Guilin Tourism University

College of Juncao Science and Ecology， Fujian Agriculture and Forestry;University

Key Laboratory for Forest Ecosystem Process and;Management of Fujian Province

Key Laboratory of Mountain Hazards and;Engineering Safety， Institute of Mountain Hazards and Environment， Chinese Academy of Sciences and Ministry of;Water Resources

University of the Chinese Academy of Sciences

⁰