Evaluation of Different Machine Learning Algorithms for Land Cover Mapping: A Case Study of Izmir Province

Abstract

Detection of land use and land cover (LULC) based on detailed, current, and accurate information plays an important role in natural resource management and spatial planning processes. However, there are some limitations such as data processing and storage requirements that hinder monitoring of land use dynamics at the regional scale. GEE is an open-source, free cloud platform that enables processing of geographic data at the global scale. The aim of this research is to obtain the LCLU map of Izmir province using different machine learning algorithms on GEE and to compare the results of the classification algorithms used. Sentinel-2 multi-band satellite images with a spatial resolution of 10m for the year 2022 and various UA indices were used in the study. The broad-scale LCLU classes used in the study were determined as 'Agricultural Area', 'Forest Area', 'Human Surfaces', 'Open Surfaces' and 'Water Surfaces'. Classification and Regression Tree (CART), Support Vector Machine (SVM), and Random Forest (RF) machine learning algorithms were used in the study, and the Producer's Accuracy (PA), User's Accuracy (UA), Overall Accuracy (OA), and Kappa Coefficient of each classifier were calculated. As a result, the RF classification algorithm with a GD of 97.2% and a Kappa value of 95.7% had the highest classification accuracy. The SVM algorithm with a GD of 96.1% and a Kappa value of 94.9% was the second highest accuracy algorithm. The GD of the CART algorithm was calculated as 93.3% and the Kappa value was 91.4%. Therefore, it was found that the RF method produced better results than the CART and SVM methods. On the other hand, it is seen that the overlap of reflection between open surfaces and human surfaces and bare agricultural areas especially in classification models makes it difficult to distinguish these classes.

Keywords

• Land Cover and Land Use
• Google Earth Engine
• Machine Learning
• Izmir

Arazi Örtüsü Haritalamasında Farklı Makine Öğrenmesi Algoritmalarının Değerlendirilmesi: İzmir İli Örneği

Öz

Doğal kaynak yönetimi ve mekânsal planlama süreçlerinde ayrıntılı, güncel ve doğru bilgilere dayanan arazi örtüsü ve arazi kullanımı (AÖAK) durumunun tespiti önemli rol oynamaktadır. Ancak, bölgesel ölçekte arazi kullanım dinamiklerinin izlenmesini engelleyen veri işleme süreci ve depolama gereksinimi gibi bazı sınırlılıklar vardır. GEE, küresel ölçekte coğrafi verilerin işlenmesine olanak tanıyan açık kaynak kodlu, ücretsiz bir bulut platformdur. Bu araştırmanın amacı GEE üzerinde farklı makine öğrenmesi algoritmaları ile İzmir ili AÖAK haritasını elde etmek ve kullanılan sınıflandırma algoritmaların sonuçlarını karşılaştırmaktır. Araştırmada 2022 yılına ait 10m mekânsal çözünürlüğe sahip Sentinel-2 çok bantlı uydu görüntüleri ile çeşitli UA indeksleri kullanılmıştır. Araştırmada kullanılan geniş ölçekteki AÖAK sınıfları ‘Tarım Alanı’, ‘Orman Alanı’, ‘Beşeri Yüzeyler’, ‘Açık Yüzeyler’ ve ‘Su Yüzeyleri’ şeklinde belirlenmiştir. Çalışmada Sınıflandırma ve Regresyon Ağacı (SRA), Destek Vektör Makinesi (DVM), Rastgele Orman (RO) makine öğrenmesi algoritmaları kullanılmış ve her bir sınıflandırıcının Üretici Doğruluğu (ÜD), Kullanıcı Doğruluğu (KD) ve Genel Doğruluğu (GD) ile Kappa Katsayısı hesaplanmıştır. Sonuç olarak %97,2 GD ve Kappa değeri %95,7 olan RO sınıflandırma algoritması, en yüksek sınıflandırma doğruluğuna sahiptir. %96,1 GD ve %94,9 Kappa değeri ile DVM algoritması ikinci en yüksek sınıflandırma doğruluğuna sahip algoritma olmuştur. SRA algoritmasının GD %93,3, Kappa değeri ise %91.4 olarak hesaplanmıştır. Sonuç olarak RO yöntemi SRA ve DVM yöntemlerine göre daha iyi sonuç verdiği tespit edilmiştir. Diğer yandan sınıflandırma modellerinde özellikle açık yüzeyler ile beşeri yüzeyler ve çıplak tarım alanları arasındaki yansıma örtüşmesi bu sınıfların ayırt edilmesini güçleştirdiği görülmektedir.

Anahtar Kelimeler

• Arazi Örtüsü ve Arazi Kullanımı
• Google Earth Engine
• Makine Öğrenmesi

References

1. Acar U., Yılmaz O. S., Çelen M., Ateş A. M., Gülgen F. & Balık Şanlı F. (2021). Determination of mucilage in the Sea of Marmara using remote sensing techniques with Google Earth Engine. International Journal of Environment and Geoinformatics, 8(4),423-434. https://doi.org/10.30897/ijegeo.957284 , Aghlmand, M., Kalkan, K., Onur, M. İ., Öztürk, G. & Ulutak, E. (2021). Google Earth Engine ile arazi kullanımı haritalarının üretimi. Niğde Ömer Halisdemir Üniversitesi Mühendislik Bilimleri Dergisi, 10(1), 38-47. https://doi.org/10.28948/ngumuh.795977
2. Al-Amri S.S, Kalyankar N.V & Khamitkar S.D. (2010). A comparative study of removal noise from remote sensing image. International Journal of Computer Science, 7(1). https://doi.org/10.48550/arXiv.1002.1148
3. Aplin, P. (2003). Using remotely sensed data. In Clifford, N.J. & Valentine, G., (Eds.) Key Methods In Geography. Sage, 291–308.
4. Belgiu, M., & Dragut, L. (2016). Random forest in remote sensing: A review of applications and future directions. ISPRS Journal of Photogrammetry and Remote Sensing, 114, 24–31.
5. Blaschke, T. (2010). Object based image analysis for remote sensing. ISPRS Journal of Photogrammetry and Remote Sensing, 65(1), 2-16. https://doi.org/10.1016/j.isprsjprs.2009.06.004
6. Breiman, L., Friedman, J.H., Olshen, R.A., & Stone, C.J. (1984). Classification and regression trees, Wadsworth.
7. Breiman, L. (2001). Random forests. Machine Learning, 45, 5–32. https://doi.org/10.1023/A:1010933404324
8. Campbell, J.B. (1996). Introduction to remote sensing. Second Edition. Virginia Polytechnic Institute and State University, The Guildford Pres.

9. Chang, K. T., Merghadi, A., Yunus, A. P., Pham, B. T., & Dou, J. (2019). Evaluating scale effects of topographic variables in landslide susceptibility models using GIS-based machine learning techniques. Scientific Reports, 9 (1), 1–21. https://doi.org/10.1038/s41598- 019-48773-2
10. Chen, B., Xiao, X., Li, X., Pan, L., Doughty, R., Ma, J., Dong, J., Qin, Y., Zhao, B., Wu, Z., Sun, R., Lan, G., Xie, G., Clinton, N., & Giri, C. (2017). A mangrove forest map of China in 2015: Analysis of time series Landsat 7/8 and Sentinel-1A imagery in Google Earth Engine cloud computing platform. ISPRS Journal of Photogrammetry and Remote Sensing, 131, 104–120. https://doi.org/10.1016/j.isprsjprs.2017.07.011.
11. Chen, B., Jin, Y., & Brown, P. (2019). Automatic mapping of planting year for tree crops with Landsat satellite time series stacks. ISPRS Journal of Photogrammetry and Remote Sensing, 151, 176–188. https://doi.org/10.1016/j.isprsjprs.2019.03.012.
12. Congalton, R.G. & Green, K. (2019). Assessing the cccuracy of remotely sensed data: principles and practices, (3rd ed.). CRC Press. https://doi.org/10.1201/9780429052729
13. Çölkesen, İ., Kavzoğlu, T., & Yomralıoğlu, T. (2015). Uzaktan algılanmış görüntülerde optimum bantların seçiminde destek vektör makinelerinin kullanımı. TUFUAB VIII. Teknik Sempozyumu, 21-23 Mayıs 2015, Konya.
14. Danacıoğlu, Ş. (2019). Arazi kullanımı/arazi örtüsü ve uzaktan algılama. D.D. Yavaşlı ve M.K. Ölgen (Ed.), Coğrafyada Uzaktan Algılama içinde (s. 161-198). İstanbul: Kriter Yayınevi.
15. Debella-Gilo M., & Gjertsen, A. K. (2021). Mapping seasonal agricultural land use types using deep learning on Sentinel-2 image time series. Remote Sensing, 13(2), 289. https://doi.org/10.3390/rs13020289
16. Deng,Y., Wu, C., Li, M. & Chen,R. (2015). RNDSI: A ratio normalized difference soil index for remote sensing of urban/suburban environments, International Journal of Applied Earth Observation and Geoinformation, 39, 40-48. https://doi.org/10.1016/j.jag.2015.02.010.
17. Diek, S., Fornallaz, F. Schaepman, M.E. & De Jong, R.. (2017). Barest pixel composite for agricultural areas using Landsat time series. Remote Sensing, 9(12), 1245. https://doi.org/10.3390/rs9121245
18. Erinç, S., (1996). Klimatoloji ve metotları. İstanbul Üniversitesi Coğrafya Enstitüsü Yayınları No: 35.
19. Erlat, E., (2003). İzmir’in hava tipleri klimatolojisi. Ege Üniversitesi Edebiyat Fakültesi Yayınları No:121.
20. Goldblatt, R., You, W., Hanson, G., & Khandelwal, A.K. (2016). Detecting the boundaries of urban areas in India: A dataset for pixel- based ımage classification in Google Earth Engine. Remote Sensing, 8, 634. https://doi.org/10.3390/rs8080634
21. Gómez, C., White, J.C., & Wulder, M.A. (2016). Optical remotely sensed time series data for land cover classification: A review. ISPRS Journal of Photogrammetry and Remote Sensing, 116, 55–72. https://doi.org/10.1016/j.isprsjprs.2016.03.008
22. Gorelick, N., Hancher, M.; Dixon, M.; Ilyushchenko, S.; Thau, D.; Moore, R. (2017). Google Earth Engine: Planetaryscale geospatial analysis for everyone. Remote Sensing of Environment, 202, 18–27. https://doi.org/10.1016/j.rse.2017.06.031
23. Hay, A.M. (1979). Sampling designs to test land use map accuracy. Photogrammetric Engineering and Remote Sensing, 45(4), 529-533.
24. He, C., Shi, P., Xie, D. & Zhao, Y. (2010). Improving the normalized difference built-up index to map urban built-up areas using a semiautomatic segmentation approach. Remote Sensing Letters, 1, 213–221. https://doi.org/10.1080/01431161.2010.481681
25. Howden, S.M., Soussana, J.-F., Tubiello, F.N., Chhetri, N., Dunlop, M., & Meinke, H. (2007). Adapting agriculture to climate change. Proceedings of the National Academy of Sciences USA, 104(50), 19691-19696. https://doi.org/10.1073/pnas.0701890104
26. Hsu, C.W.; Chang, C.C. & Lin C.J. A. (2003). A Practical guide to support vector classification. University of National Taiwan, Department of Computer Science and Information Engineering., 1–12. https://www.csie.ntu.edu.tw/~cjlin/papers/guide/guide.pdf
27. Huete, A.R. (1988). A soil-adjusted vegetation index (SAVI). Remote Sensing of Environment, 25, 295-309. https://doi.org/10.1016/0034-4257(88)90106-X
28. Huete, A., Didan, K., Miura, T., Rodriguez, E. P., Gao, X., & Ferreira, L. G. (2002). Overview of the radiometric and biophysical performance of the MODIS vegetation indices. Remote sensing of environment, 83(1-2), 195-213. https://doi.org/10.1016/S0034-4257(02)00096-2
29. Hurni, H.,Tato, K., & Zeleke G. (2005). The implications of changes in population, land use, and land management for surface runoff in the upper Nile basin area of Ethiopia Mountain. Research and Development, 25 (2), 147-154. https://doi.org/10.1659/0276-4741
30. Ibrahim, S. (2023). Improving land use/cover classification accuracy from random forest feature ımportance selection based on synergistic use of Sentinel data and digital elevation model in agriculturally dominated landscape. Agriculture, 13(1), 98. https://doi.org/10.3390/agriculture13010098
31. İzmir Valiliği Çevre, Şehircilik ve İklim Değişikliği İl Müdürlüğü. (2022). İÇDR, İzmir ili 2021 yılı çevre durum raporu, https://webdosya.csb.gov.tr/db/ced/icerikler/izmir-ilcdr-2021-20220811104124.pdf
32. Jamal, S. & Ahmad, W.S. (2020). Assessing land use land cover dynamics of wetland ecosystems using Landsat satellite data. SN Applied Sciences, 2, 1891. https://doi.org/10.1007/s42452-020-03685-z
33. Jordan, C.F., (1969). Derivation of leaf-area index from quality of light on the forest floor. Ecology, 50 (4), 663–666. https://doi.org/10.2307/1936256
34. Kavzoğlu, T., Şahin, E.K., & Çölkesen, İ. (2012). Heyelan duyarlılığının incelenmesinde regresyon ağaçlarının kullanımı: Trabzon örneği, Harita Dergisi, 147, 21-33.
35. Kayode A. Adepoju & Samuel A. A. (2020). Improving accuracy of Landsat-8 OLI classification using image composite and multisource data with Google Earth Engine, Remote Sensing Letters, 11(2), 107-116. https://doi.org/10.1080/2150704X.2019.1690792
36. Kawamura, M., S. Jayamanna & Y. Tsujiko (1997). Quantitative evaluation of urbanization in developing countries using satellite data. Doboku Gakkai Ronbunshu. 580, 45–54.
37. Li, Y., Li, C., Li, M. & Liu, Z. (2019). Influence of variable selection and forest type on forest aboveground biomass estimation using machine learning algorithms. Forests, 10, 1073. https://doi.org/10.3390/f10121073
38. Lillesand, T.M., Kiefer, R.W. , & Chipman, J.W. (2018). Uzaktan algılama ve görüntü yorumalama. (K.Ş. Kavak, Çev.) Palme Yayınevi (Orijinal çalışma basım tarihi 2015).
39. Loukika, K. N., Keesara, V. R., & Sridhar, V. (2021). Analysisof land use and land cover using machine learningalgorithms on google earth engine for Munneru riverbasin, India. Sustainability, 13(24). https://doi.org/10.3390/su13241375
40. Mahdianpari, M., Salehi, B., Mohammadimanesh, F., Homayouni, S., & Gill, E., (2019). The first wetland inventory map of newfoundland at a spatial resolution of 10 m using sentinel-1 and sentinel-2 data on the google earth engine cloud computing platform. Remote Sensing, 11, 43. https://doi.org/10.3390/rs11010043
41. McFeeters, S.K. (1996). The use of the normalized difference water ındex (NDWI) in the delineation of open water features. International Journal of Remote Sensing, 17, 1425–1432. https://doi.org/10.1080/01431169608948714
42. Meteoroloji Genel Müdürlüğü. (2022). İllere ait mevsim normalleri (19912020). https://www.mgm.gov.tr/veridegerlendirme/il-ve-ilceler-istatistik.aspx?m=IZMIR
43. Meyer, W. B. & Turner, B. L. (1992). Human population growth and global land-use/cover change. Annual Review of Ecology and Systematics, 23(1), 39–61. https://doi.org/10.1146/annurev.es.23.110192.00035
44. Midekisa, A.; Holl, F.; Savory, D.J.; Andrade-Pacheco, R.; Gething, P.W.; Bennett, A. & Sturrock, H.J.W. (2017). Mapping land cover change over continental Africa using Landsat and Google Earth Engine cloud computing. PLoS One, 12, 1–15, https://doi.org/10.1371/journal.pone.0184926
45. Mitran T., Meena R.S. & Chakraborty A. (2021). Geospatial technologies for crops and soils: an overview. In: Mitran T., Meena R.S., & Chakraborty A. (Eds.), Geospatial Technologies For Crops And Soils. Springer, Singapore. https://doi.org/10.1007/978-981-15-6864-0_1
46. Mountrakis, G., Im, J., & Ogelo, C. (2011). Support vector machines in remote sensing: A review. ISPRS Journal of Photogrammetry and Remote Sensing, 66, 247-259. http://dx.doi.org/10.1016/j.isprsjprs.2010.11.001
47. Mutanga, O., Dube, T. & Galal, O. (2017). Remote sensing of crop health for food security in Africa: Potentials and constraints, Remote Sensing Applications: Society and Environment, 8, 231-239. https://doi.org/10.1016/j.rsase.2017.10.004
48. Nguyen C.T., Chidthaisong, A., Kieu Diem, P., & Huo, L-Z. (2021). A modified bare soil ındex to ıdentify bare land features during agricultural fallow-period in Southeast Asia using Landsat 8. Land, 10(3):231. https://doi.org/10.3390/land10030231
49. Pal M. (2005). Random forest classifier for remote sensing classification, International Journal Of Remote Sensing, 26(1), 217-222. https://doi.org/10.1080/01431160412331269698
50. Pal, M., & Mather, PM., (2005). Support vector machines for classification in remote sensing, International Journal of Remote Sensing, 26, 1007–1011. https://doi.org/10.1080/01431160512331314083
51. Pettorelli, N., Laurance, W.F., O’Brien, T.G., Wegmann, M., Nagendra, H. & Turner, W. (2014). Satellite remote sensing for applied ecologists: opportunities and challenges. Journal of Applied Ecology, 51, 839-848. https://doi.org/0.1111/1365-2664.12261
52. Plourde, L., & Congalton, R. G. (2003). Sampling method and sample placement: How do they affect the accuracy of remotely sensed maps. Photogrammetric Engineering & Remote Sensing, 69(3). https://www.asprs.org/wp-content/uploads/pers/2003journal/march/2003_mar_289-297.pdf
53. Rembold, F., Meroni, M., Urbano, F., Csak, G., Kerdiles, H., Perez-Hoyos, A., Lemoine, G., Leo, O., & Negre, T. (2019). ASAP: A new global early warning system to detect anomaly hot spots of agricultural production for food security analysis. Agricultural Systems, 168, 247–257. https://doi.org/10.1016/j.agsy.2018.07.002
54. Rogan, J. & Chen, DM. (2004). Remote sensing technology for mapping and monitoring land-cover and land-use change. Progress in Planning, 61, 301–325. https://doi.org/10.1016/S0305-9006(03)00066-7
55. Rikimaru, A.; Roy, P.S. & Miyatake, S. (2002). Tropical forest cover density mapping. Int Society for Tropical Ecology, 43, 39-47. https://tropecol.org/pdf/open/PDF_43_1/43104.pdf
56. Rasul A, Balzter H, Ibrahim GRF, Hameed HM, Wheeler J, Adamu B, Ibrahim S, & Najmaddin PM. (2018). Applying built-up and bare-soil ındices from Landsat 8 to cities in dry climates. Land, 7(3):81. https://doi.org/10.3390/land7030081
57. Rouse, J. W., R. H. Haas, D. W. Deering, & Schell, J. A. (1973). Monitoring the vernal advancement and retrogradation (green wave effect) of natural vegetation, 44–47, College Station, TX: Remote Sensing Center, Texas A&M University.
58. Saah, D., Johnson, G., Ashmall, B., Tondapu, G., Tenneson, K., Patterson, M., Poortinga, A., Markert, K., Quyen, N.H., & San Aung, K. (2019). Collect Earth: An online tool for systematic reference data collection in land cover and use applications. Environmental Modelling & Software, 118, 166–171. https://doi.org/10.1016/j.envsoft.2019.05.004
59. Shafizadeh-Moghadam, H., Khazaei, M., Alavipanah, S. K., & Weng Q. (2021). Google Earth Engine for large-scale land use and land cover mapping: an object-based classification approach using spectral, textural and topographical factors, GIScience & Remote Sensing, 58(6), 914-928. https://doi.org/10.1080/15481603.2021.1947623
60. Shao, Y. & Lunetta, R.S. (2012). Comparison of support vector machine, neural network, and cart algorithms for the land-cover classification using limited training data points. ISPRS Journal of Photogrammetry and Remote Sensing, 70, 78–87. https://doi.org/10.1016/j.isprsjprs.2012.04.001.
61. Shen, L., & Li, C. (2010). Water body extraction from Landsat ETM+ imagery using adaboost algorithm, 2010 18th International Conference on Geoinformatics, 2010. https://doi.org/10.1109/GEOINFORMATICS.2010.5567762.
62. Smith, C. & McDonald, G. (1998). Assessing the sustainability of agriculture at the planning stage. Journal of Environmental Management, 52, 15–37. https://doi.org/10.1006/jema.1997.0162
63. Song, XP., Hansen, M.C., Stehman, S.V., Potapov, PV., Tyukavina, A. Vermote EF. & Townshend, JR. (2018). Global land change from 1982 to 2016. Nature, 560, 639–643. https://doi.org/10.1038/s41586-018-0411-9
64. Stathakis, D., Perakis, K., & Savin, I. (2012). Efficient segmentation of urban areas by the VIBI. International Journal of Remote Sensing, 33(20):6361–6377. https://doi.org/10.1080/01431161.2012.687842
65. Tağıl, Ş. (2015). Effect Of Topographic Habitat Characteristics On The Spatial Distribution Of Landuse Landcover İn The Kapidag Peninsula Turkey. Journal of Applied Science, 15(6), 850–861. https://doi.org/10.3923/jas.2015.850.861
66. Tamiminia, H., Salehi, B., Mahdianpari, M., Quackenbush, L., Adeli, S., & Brisco, B. (2020). Google Earth Engine for geo-big data applications: A meta-analysis and systematic review. ISPRS Journal of Photogrammetry and Remote Sensing, 164, 152–170. https://doi.org/10.1016/j.isprsjprs.2020.04.001
67. Tarım ve Orman Bakanlığı. (2022). CORINE CBS, https://corinecbs.tarimorman.gov.tr.
68. Tso, B. & Mather P. (2009). Classification methods for remotely sensed data. CRC Press.
69. TÜİK, Türkiye İstatistik Kurumu (2022). Adrese dayalı nüfus kayıt sistemi. https://data.tuik.gov.tr/Bulten/Index?p=45500
70. Wang, Q., Shi, W.; Li, Z.; Atkinson, P.M. (2016). Fusion of sentinel-2 images. Remote Sensing of Environment, 187, 241–252, https://doi.org/10.1016/j.rse.2016.10.030.
71. Waqar M.M., Mirza J.F., Mumtaz, R., & Hussain, E. (2012). Development of new ındices for extraction of builtup area & bare soil from Landsat data. Open Access Scientific Reports 1: 136. https://doi.org/10.4172/scientificreports.136
72. Villa, P. (2012). Mapping urban growth using soil and vegetation ındex and LANDSAT data: The Milan (Italy) city area case study, Landscape and Urban Planning, 107, 245-254. https://doi.org/10.1016/j.landurbplan.2012.06.014
73. Xu, H. (2006). Modification of normalized difference water index (NDWI) to enhance open water features in remotely sensed imagery. International Journal of Remote Sensing, 27, 3025–3033. https://doi.org/10.1080/01431160600589179
74. Varade, D., Sure, A., & Dikshit, O., (2018). Potential of Landsat-8 and Sentinel-2A composite for land use land cover analysis. Geocarto International, 34(14),1552-1567. https://doi.org/10.1080/10106049.2018.1497096
75. Yılmaz, O.S., Oruç, M.S., Ateş, A.M., & Gülgen, F. (2021). Orman yangın şiddetinin Google Earth Engine ve coğrafi bilgi sistemleri kullanarak analizi: Hatay-Belen örneği. Journal of the Institute of Science and Technology, 11(2), 1519-1532. https://doi.org/10.21597/jist.817900
76. Zha, Y., Gao, J., & Ni, S. (2003). Use of normalized difference built-up index in automatically mapping urban areas from TM imagery. International Journal of Remote Sensing, 24, 583–594. https://doi.org/10.1080/01431160304987