Journal of Forestry Research

, Volume 31, Issue 1, pp 107–121 | Cite as

BP neural networks and random forest models to detect damage by Dendrolimus punctatus Walker

  • Zhanghua Xu
  • Xuying Huang
  • Lu Lin
  • Qianfeng WangEmail author
  • Jian Liu
  • Kunyong Yu
  • Chongcheng Chen
Original Paper


The construction of a pest detection algorithm is an important step to couple “ground-space” characteristics, which is also the basis for rapid and accurate monitoring and detection of pest damage. In four experimental areas in Sanming City, Jiangle County, Sha County and Yanping District in Fujian Province, sample data on pest damage in 182 sets of Dendrolimus punctatus were collected. The data were randomly divided into a training set and testing set, and five duplicate tests and one eliminating-indicator test were done. Based on the characterization analysis of the host for D. punctatus damage, seven characteristic indicators of ground and remote sensing including leaf area index, standard error of leaf area index (SEL) of pine forest, normalized difference vegetation index (NDVI), wetness from tasseled cap transformation (WET), green band (B2), red band (B3), near-infrared band (B4) of remote sensing image are obtained to construct BP neural networks and random forest models of pest levels. The detection results of these two algorithms were comprehensively compared from the aspects of detection precision, kappa coefficient, receiver operating characteristic curve, and a paired t test. The results showed that the seven indicators all were responsive to pest damage, and NDVI was relatively weak; the average pest damage detection precision of six tests by BP neural networks was 77.29%, the kappa coefficient was 0.6869 and after the RF algorithm, the respective values were 79.30% and 0.7151, showing that the latter is more optimized, but there was no significant difference (p > 0.05); the detection precision, kappa coefficient and AUC of the RF algorithm was higher than the BP neural networks for three pest levels (no damage, moderate damage and severe damage). The detection precision and AUC of BP neural networks were a little higher for mild damage, but the difference was not significant (p > 0.05) except for the kappa coefficient for the no damage level (p < 0.05). An “over-fitting” phenomenon tends to occur in BP neural networks, while RF method is more robust, providing a detection effect that is better than the BP neural networks. Thus, the application of the random forest algorithm for pest damage and multilevel dispersed variables is thus feasible and suggests that attention to the proportionality of sample data from various categories is needed when collecting data.


BP neural networks Detection precision Kappa coefficient Pine moth Random forest ROC curve 



The authors are grateful to the National Natural Science Foundation of China (Grant Nos. 41501361, 41401385, 30871965), the China Postdoctoral Science Foundation (No. 2018M630728), the Open Fund of Fujian Provincial Key Laboratory of Resources and Environment Monitoring and Sustainable Management and Utilization (No. ZD1403), the Open Fund of Fujian Mine Ecological Restoration Engineering Technology Research Center (No. KS2018005) and the Scientific Research Foundation of Fuzhou University (No. XRC1345).


  1. Agatz A, Ashauer R, Sweeney P, Brown CD (2017) Prediction of pest pressure on corn root nodes: the POPP-corn model. J Pest Sci 90(1):161–172Google Scholar
  2. Breiman L (2001) Random forest. Mach Learn 45:5–32Google Scholar
  3. Capodici F, D’Urso G, Maltese A (2013) Investigating the relationship between X-band SAR data from COSMO-SkyMed satellite and NDVI for LAI detection. Remote Sens 5(3):1389–1404Google Scholar
  4. Chen HH, Zhu SY, Cui XF (2003) A study on the forecast model of Dendrolimus punctatus occurrence based on artificial neural network. For Res 16(2):159–165Google Scholar
  5. Chen ZQ, Shi RH, Zhang SP (2013) An artificial neural network approach to estimate evapotranspiration from remote sensing and AmeriFlux data. Front Earth Sci 7(1):103–111Google Scholar
  6. Chen W, Li X, Wang Y, Chen G, Liu S (2014) Forested landslide detection using LiDAR data and the random forest algorithm: a case study of the Three Gorges, China. Remote Sens Environ 152:291–301Google Scholar
  7. Cho MA, Debba P, Mutanga O, Dudeni N, Magadla T, Khuluse SA (2012) Potential utility of the spectral red-edge region of SumbandilaSat imagery for assessing indigenous forest structure and health. Int J Appl Earth Obs Geoinf 16:85–93Google Scholar
  8. Coops NC, Johnson M, Wulder MA, White JC (2006) Assessment of QuickBird high spatial resolution imagery to detect red attack damage due to mountain pine beetle infestation. Remote Sens Environ 103(1):67–80Google Scholar
  9. Cui HJ, Wu HG, Qiao YY, Yan XJ, Chen LH (1997) Modeling Dendrolimus punctatus damage by remote sensing detecting. J Biomath 12(S1):611–616Google Scholar
  10. Fang KN, Wu JB, Zhu JP, Xie BC (2011) A review of technologies on random forests. Stat Inf Forum 26(3):32–38Google Scholar
  11. Haddad JE, Villot-Kadri M, Ismael A, Gallou C, Michel K, Bruyère D, Laperche V, Canioni I, Bousquet B (2013) Artificial neural network for on-site quantitative analysis of soils using laser induced breakdown spectroscopy. Spectrochim Acta B 79–80:51–57Google Scholar
  12. He FD, Zeng MW, Xie HY, Dai HW (2013) Computer simulation of plant disease and insect pests spread based on the cellular automata. Math Model Appl 2(2):42–45Google Scholar
  13. He Y, Bo Y, Chai L, Liu X, Li A (2016) Linking in situ LAI and fine resolution remote sensing data to map reference LAI over cropland and grassland using geostatistical regression method. Int J Appl Earth Obs Geoinf 50:26–38Google Scholar
  14. Jepsen JU, Hagen SB, Høgda KA, Ims RA, Karlsen SR, Tømmervik H, Yoccoz NG (2009) Monitoring the spatio-temporal dynamics of geometrid moth outbreaks in birch forest using MODIS-NDVI data. Remote Sens Environ 113:1939–1947Google Scholar
  15. Kantola T, Lyytikäinen-Saarenmaa P, Coulson RN, Holopainen M, Tchakerian MD, Streett DA (2016) Development of monitoring methods for Hemlock Woolly Adelgid induced tree mortality within a Southern Appalachian landscape with inhibited access. iFor Biogeosci For 9:178–186Google Scholar
  16. Lee S, Park I, Koo BJ, Ryu JH, Choi JK, Woo HJ (2013) Macrobenthos habitat potential mapping using GIS-based artificial neural network models. Mar Pollut Bull 67(1–2):177–186PubMedGoogle Scholar
  17. Li J, Qin G, Wen X, Hu N (2002) Over-fitting in neural network learning algorithms and its solving strategies. J Vib Meas Diagn 22(4):260–264Google Scholar
  18. Li M, Liu M, Liu M, Ju Y (2010) Prediction of pine wilt disease in Jiangsu Province based on web dataset and GIS. Web information systems and mining. Springer, Berlin, pp 146–153Google Scholar
  19. Luo SZ, Wang C (2011) Forest pests and diseases forecasting based on GIS. Adv Mater Res 250–253:2945–2948Google Scholar
  20. Martinez B, Camacho F, Verger A, Garcia-Haro FJ, Gilabert MA (2013) Intercomparison and quality assessment of MERIS, MODIS and SEVIRI FAPAR products over the Iberian Peninsula. Int J Appl Earth Obs Geoinf 21(1):463–476Google Scholar
  21. Park YS, Chung YJ (2006) Hazard rating of pine trees from a forest insect pest using artificial neural networks. For Ecol Manag 222(1–3):222–233Google Scholar
  22. Patil J, Mytri VD (2013) A Prediction model for population dynamics of cotton pest (Thrips tabaci Linde) using multilayer-perceptron neural network. Int J Comput Appl 67(4):19–26Google Scholar
  23. Peixoto MDS, Barros LCD, Bassanezi RC (2014) A model of cellular automata for the fuzzy control of aphids. Appl Math 5(5):1133–1141Google Scholar
  24. Provost F, Hibert C, Malet JP (2017) Automatic classification of endogenous landslide seismicity using the random forest supervised classifier. Geophys Res Lett 44(1):1–8Google Scholar
  25. Pu R, Gong P, Biging GS, Larrieu MR (2003) Extraction of red edge optical parameters from Hyperion data for estimation of forest leaf area index. IEEE Trans Geosci Remote Sens 41(4):916–921Google Scholar
  26. Stenberg P, Rautiainen M, Manninen T, Voipio P, Smolander H (2008) Reduced simple ratio better than NDVI for estimating LAI in Finnish pine and spruce stands. Silva Fennica 38(1):3–14Google Scholar
  27. Svetnik V, Liaw A, Tong C, Culberson JC, Sheridan RP, Feuston BP (2003) Random forest: a classification and regression tool for compound classification and QSAR modeling. J Chem Inf Comput Sci 43(6):1947–1958Google Scholar
  28. Tomassetti B, Lombardi A, Cerasani E, Sabatino AD, Pace L, Ammazzalorso D, Verdecchia M (2013) Mapping of Alternaria and Pleospora concentrations in Central Italy using meteorological forecast and neural network estimator. Aerobiologia 29(1):55–70Google Scholar
  29. Wang Y, Xiong Z (2013) Prediction of the forest health based on BP neural networks. Adv Mater Res 731:4303–4306Google Scholar
  30. Wang L, Huang H, Luo Y (2010) Remote sensing of insect pests in larch forest based on physical model. In: Geoscience and remote sensing symposium. IEEE, pp 3299–3302Google Scholar
  31. Wang L, Ma C, Zhou X, Zi Y, Zhu X, Guo W (2015) Estimation of wheat leaf SPAD value using RF algorithmic model and remote sensing data. Trans Chin Soc Agric Mach 46(1):259–265Google Scholar
  32. Wong MS, Sarker MLR, Nichol J, Lee SC, Chen HW, Wan YL, Chan PW (2013) Modeling BVOC isoprene emissions based on a GIS and remote sensing database. Int J Appl Earth Obs Geoinf 21:66–77Google Scholar
  33. Xu GY, Xu W, Fang SA, Zhang J (2008) The relationship between the occurring damage of Dendrolimus punctatus Walker and forest form. J Hebei Agric Sci 12(10):31–32Google Scholar
  34. Xu ZH, Yu KY, Liu J, Xie SJ, Li XP, Chen FH, Qi XL, Chen GR, Li ZL (2012) A method for extraction of Dendrolimus punctatus damage information suitable for southern hilly areas in China. Acta Agric Univ Jiangxiensis (Natural Sciences Edition) 34(5):933–939Google Scholar
  35. Xu ZH, Liu J, Yu KY, Gong CH, Xie WJ, Tang MY, Lai RW, Li ZL (2013a) Leaf area index and standard error of pine forests estimated with common digital camera. Chin J Eco Agric 21(5):638–644Google Scholar
  36. Xu ZH, Liu J, Yu KY, Gong CH, Xie WJ, Tang MY, Lai RW, Li ZL (2013b) Spectral features analysis of Pinus massoniana with pest of Dendrolimus punctatus Walker and levels detection. Spectrosc Spectr Anal 33(2):428–433Google Scholar
  37. Xu ZH, Li CH, Liu J, Yu KY, Gong CH, Tang MY (2014) Fisher discriminant analysis of Dendrolimus punctatus Walker pest levels. Trans Chin Soc Agric Mach 45(6):275–283Google Scholar
  38. Xu ZH, Huang XY, Lin L, Wang QF, Liu J, Chen CC, Yu KY, Zhou HK, Zhang HF (2018) Dendrolimus punctatus Walker damage detection based on fisher discriminant analysis and random forest. Spectrosc Spectr Anal 38(9):2888–2896Google Scholar
  39. Youssef AM, Pourghasemi HR, Pourtaghi ZS, Ai-Katheeri MM (2015) Landslide susceptibility mapping using random forest, boosted regression tree, classification and regression tree, and general linear models and comparison of their performance at Wadi Tayyah Basin, Asir Region, Saudi Arabia. Landslides 13(5):839–856Google Scholar
  40. Zhang AB, Chen J, Wang ZJ, Li DM, Tian J (2001) The application of BP model and LOGIT model to prediction of forest insect pests. Acta Ecol Sin 21(12):2159–2165Google Scholar
  41. Zhang L, Wang LL, Zhang XD, Liu SR, Sun PS (2014) The basic principle of random forest and its applications in ecology—a case study of Pinus yunnanensis. Acta Ecol Sin 34(3):1–10Google Scholar
  42. Zhang S, Li H, Wang L, Liu D, Zou P, Ping E, Ma T, Huang Q (2016) Research and application of hybrid PSO-BP neural network in fracture acidizing well production prediction. Revista de la Facultad de Ingeniería U.C.V 31(6):166–176Google Scholar

Copyright information

© Northeast Forestry University and Springer-Verlag GmbH Germany, part of Springer Nature 2018

Authors and Affiliations

  • Zhanghua Xu
    • 1
    • 2
    • 3
  • Xuying Huang
    • 1
  • Lu Lin
    • 1
  • Qianfeng Wang
    • 1
    Email author
  • Jian Liu
    • 2
  • Kunyong Yu
    • 2
  • Chongcheng Chen
    • 3
  1. 1.College of Environment and ResourcesFuzhou UniversityFuzhouPeople’s Republic of China
  2. 2.Fujian Provincial Key Laboratory of Resources and Environment Monitoring and Sustainable Management and UtilizationSanmingPeople’s Republic of China
  3. 3.Key Lab of Spatial Data Mining and Information SharingMinistry of EducationFuzhouPeople’s Republic of China

Personalised recommendations