Comparison of Agricultural Crop Type Classifications with Different Machine Learning Algorithms (RF-SVM-ANN-XGBoost) by Generating Ground Truth Data from Farmer Declaration Parcels

Abstract

In large-scale agricultural crop classification studies (Turkey, Adana, Çukurova Plain, 2500 km²), collecting sufficient and accurate ground truth data is costly, time-consuming, and unsustainable. This study utilized parcels registered in the Farmer Registration System (FRS) as ground truth data. By analyzing time series EVI curves, discrepancies were identified between declared and actual crops. Erroneous parcels were eliminated, and the corrected data were used in the classification process.Using multi-temporal Sentinel-2 images from 2021, this study compared the performance of Random Forests (RF), Artificial Neural Networks (ANN), Support Vector Machines (SVM), and Extreme Gradient Boosting (XGBoost) algorithms for classifying crops like citrus, cotton, maize, peanut, sunflower, watermelon, wheat, and double-crop combinations (e.g., wheat-cotton, wheat-maize). The classification utilized 121 features (11 images × 10 Sentinel-2 bands + EVI). XGBoost achieved the highest overall accuracy (92.14%), followed by RF (89.15%), SVM (86.14%), and ANN (85.48%).The EVI index proved critical, particularly in separating spectral curves of double crops. While single crops like cotton, maize, and wheat yielded high classification accuracy, double crops with overlapping phenological stages had lower accuracy. The study highlighted that crops at distinct phenological stages performed well across algorithms, whereas crops with similar stages struggled to achieve high accuracy.This method of using corrected farmer-declared parcels (FDP) as ground truth data demonstrated high classification performance across all algorithms, proving its reliability. The findings emphasize that FDP can effectively replace traditional field data collection, reducing costs and improving efficiency. This classification approach supports agricultural production monitoring, yield estimation, water resource analysis, and sustainable policy-making, serving as a robust tool for agricultural evaluation

Keywords

• ANN
• RF
• SVM
• XGBoost
• FDP

References

1. Şimşek, F.F. (2023). Optik ve radar görüntüleri ile aşırı gradyan artırma algoritması kullanılarak tarımsal ürün desen tespiti. Geomatik Dergisi, 9(1),54–68 https://doi.org/10.29128/geomatik.1332997
2. Matton, N., Canto, G.S., Waldner, F., Valero, S., Morin, D., Inglada, J., Arias, M., Bontemps, S., Koetz, B., & Defourny, P. (2015). An automated method for annual cropland mapping along the season for various globally-distributed agro systems using high spatial and temporal resolution time series. Remote Sensing, 7 (10), 13208-13232.
3. Qiong, H., Wen-bin, W., Qian, S., Miao, L., Di, C., Qiang-yi, Y., & Hua-jun, T. (2017). How do temporal and spectral features matter in crop classification in Heilongjiang Province, China Journal of Integrative Agriculture, 16(2), 324–336.https://doi:10.1016/S2095 3119(15)61321-1
4. Zhang, C., Zhang, H., Du, J., & Zhang, L. (2018). Automated paddy rice extent extraction with time stacks of sentinel data: a case study in Jianghan plain, Hubei, China. 7th International Conference on Agro-geoinformatics (Agro-geoinformatics), 1-6 https://doi:10.1109/AgroGeoinformatics.2018.8476119
5. Altun, M., & Turker, M. (Year). Integration of Sentinel-1 and Landsat-8 images for crop detection: The case study of Manisa, Turkey. Advanced Remote Sensing, 2(1), 23-33
6. Waldner, F., Canto, G.S., Defourny, P. (2015). Automated annual cropland mapping using knowledge-based temporal features. ISPRS Journal of Photogrammetry and Remote Sensing,110:1-13. https://doi:10.1016/j.isprsjprs.2015.09.013
7. Csillik, O., Belgiu, M., Asner, G.P., & Kelly, M. (2016). Object-based time-constrained dynamic time warping classification of crops using sentinel-2. Remote Sensing, 11(10),1257. https://doi.org/10.3390/rs11101257
8. King, L.M., Adusei, B., Stehman, S., Potapov, P.V., Song, X., Krylov, A., Bella, C.M., Loveland, T.R., Johnson, D.M., & Hansen, M.C., (2017). A multi-resolution approach to national-scale cultivated area estimation of soybean. Remote Sensing of Environment, 195, 13-29. https://doi.org/10.1016/j.rse.2017.03.047

9. Wardlow, B.D., & Egbert, S.L. (2008). Large-area crop mapping using time-series MODIS 250m NDVI data: An assessment for the U.S. Central Great Plains. Remote Sensing of Environment 112, 1096-1116 https://doi.org/10.1016/j.rse.2007.07.019
10. Wang, S., Azzari, G., & Lobell, D. (2019). Crop type mapping without field-level labels: Random forest transfer and unsupervised clustering techniques. Remote Sensing of Environment, 222,303-317. https://doi.org/10.1016/j.rse.2018.12.026
11. Kang, J., Zhang, H., Yang, H., & Zhang, L. (2018). Support vector machine classification of crop lands using sentinel-2 imagery. 2018 7th International Conference on Agro-geoinformatics (Agro-geoinformatics), Hangzhou, China, pp. 1-6. https://doi: 10.1109/Agro-Geoinformatics.2018.8476101
12. Gomez, G., Shi, Z., Zhu, Y., Yang, X., & Hao, Y. (2020). Land use/cover classification in an arid desert-oasis mosaic landscape of China using remote sensed imagery: Performance assessment of four machine learning algorithms. Global Ecology and Conservation, 22, e00971. https://doi.org/10.1016/j.gecco.2020.e00971
13. Zhang, H., Kang, J., Xu, X., & Zhang, L. (2020). Accessing the temporal and spectral features in crop type mapping using multi-temporal Sentinel-2 imagery: A case study of Yi'an County, Heilongjiang province, China. Computer Electronic Agriculture, 176, 105618. https://doi.org/10.1016/j.compag.2020.105618
14. Li, Q., Tian, J., & Tian, Q. (2023). Deep learning application for crop classification via multi-temporal remote sensing images. Agriculture, 13(4),906.https://doi.org/10.3390/agriculture13040906
15. Vuolo, F., Neuwirth, M., Immitzer, M., Atzberger, C., & Ng, W. (2018). How much does multi-temporal Sentinel-2 data improve crop type classification? Int. J. Appl. Earth Obs. Geoinformation, 72, 122-130 https://doi.org/10.1016/j.jag.2018.06.007
16. Arvor, D., Jonathan, M., Simoes, M., Dubreuil, V., & Durieux, L. (2011). Classification of MODIS EVI time series for crop mapping in the state of Mato Grosso, Brazil. International Journal of Remote Sensing, 32(22), 7847-7871. https://doi.org/10.1080/01431161.2010.531783
17. Belgiu, M., & Csillik, O. (2018). Sentinel-2 cropland mapping using pixel-based and object-based time-weighted dynamic time warping analysis. Remote Sensing of Environment, 204, 509-523. https://doi.org/10.1016/j.rse.2017.10.005
18. Zheng, H., Du, P., Chen, J., Xia, J., Li, E., Xu, Z., Li, X., & Yokoya, N. (2017). Performance evaluation of downscaling sentinel-2 imagery for land use and land cover classification by spectral-spatial features. Remote Sensing, 9(12),1274. https://doi.org/10.3390/rs9121274
19. Müller, U. &, Wilm, U. (2017). Sen2Cor configure ration and user manual. Ref. S2-PDGS-MPC-L2A-SUM-V2.4, 1, 9-12
20. Müller, U., Wilm, U., Louis, J., Richter, R., Gascon, F., & Niezette, M. (2013). Sentinel-2 level 2a prototype processor: architecture, algorithms and first results. ESA Living Planet Symposium vol. 722, p. 98
21. Zhu, Z., & Woodcock, C.E. (2012). Object-based cloud and cloud shadow detection in Landsat images for tropical forest monitoring. Remote Sensing of Enviroment, 118,83-94. https://doi.org/10.1016/j.rse.2011.10.028
22. Unel, F. B., Kusak, L., & Yakar, M. (2023). GeoValueIndex map of public property assets generating via Analytic Hierarchy Process and Geographic Information System for Mass Appraisal: GeoValueIndex. Aestimum, 82, 51-69. https://doi.org/10.3390/rs13153031
23. Morsy, S., & Hadi, M. (2022).Investigation of phenological stages of wheat plant using vegetation index. Mersin Photogrammetry Journal, 2 (1) , 24-28.
24. Kaya, Y., & Polat, N. (2020). Impact of land use/land cover on land surface temperature and its relationship with spectral indices in Dakahlia Governorate, Egypt . International Journal of Engineering and Geosciences, 7 (3) , 272-282. https://doi.org/10.26833/ijeg.978961
25. Maxwell, A.E., Warner, T.A., & Fang, F. (2018). Implementation of machine-learning classification in remote sensing: an applied review. International Journal of Remote Sensing, 39(9),2784-2817. https://doi.org/10.1080/01431161.2018.1433343
26. Mountrakis, G., Im, J., & Ogole, C. (2011). Support vector machines in remote sensing: a review. ISPRS Journal of Photogrammetry and Remote Sensing, 66, 247-259 https://dx.doi.org/10.1016/j.isprsjprs.2010.11.001
27. Huang, C., Davis, L,S., & Townshend, J.R.G. (2002). An assessment of support vector machines for land cover classification. International Journal of Remote Sensing, 23: 725-749.http://dx.doi.org/10.1080/01431160110040323
28. Mathur, A., & Foody, G.M. (2008). Crop classification by support vector machine with intelligently selected training data for an operational application. International Journal of Remote Sensing, 29, 2227-2240.https://doi.org/10.1080/01431160701395203
29. Duro, D. C., Franklin, S.E., & Dubé, M.G. (2012). A comparison of pixel-based and object-based image analysis with selected machine learning algorithms for the classification of agricultural landscapes using SPOT-5 HRG imagery. Remote Sensing of Environment, 118, 259-272. http://dx.doi.org/10.1016/j.rse.2011.11.020.
30. 30.Ustuner, M., & Sanli, F.B. (2021). Crop classification from multi-temporal PolSAR data with regularized greedy forest. Advanced Remote Sensing, 1(1), 10-15
31. 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.https://doi.org/10.1016/j.isprsjprs.2016.01.011
32. Avcı, C., Budak, M., Yağmur, N., Balçık, F. (2023). Comparison between random forest and support vector machine algorithms for LULC classification. International Journal of Engineering and Geosciences, 8(1), 1-10. https://doi.org/10.26833/ijeg.98760533.
33. Tatsumi, M., Yamashiki, Y., Torres, M.A.C., & Taipe, C.L.R. (2015). Crop classification of upland fields using Random forest of time series Landsat 7 ETM+ data. Computers and Electronics in Agriculture, 115, 171-179. https://doi.org/10.1016/j.compag.2015.05.01
34. Zhong, L., Gong. P., Biging, G.S. (2014). Efficient corn and soybean mapping with temporal extendability: A multi-year experiment using Landsat imagery. Remote Sensing of Environment, 140, 1-13. https://doi.org/10.1016/j.rse.2013.08.023
35. Kusak, L., Unel, F. B., Alptekin, A., Celik, M. O., & Yakar, M. (2021). Apriori association rule and K-means clustering algorithms for interpretation of pre-event landslide areas and landslide inventory mapping. Open Geosciences, 13(1), 1226-1244
36. Terzi Türk, S., & Balçık, F. (2023). Rastgele orman algoritması ve Sentinel-2 MSI ile fındık ekili alanların belirlenmesi: Piraziz Örneği. Geomatik, 8(2), 91-98. https://doi.org/10.29128/geomatik.1127925
37. Zhang, H. K., & Roy, D.P. (2014). Using the 500 m MODIS land cover product to derive a consistent continental scale 30 m Landsat land cover classification. Remote Sensing of Environment, 197, 15-34. https://doi.org/10.1016/j.rse.2017.05.02
38. Chen, T.Q., & Guestrin, C. (2016). Xgboost: a scalable tree boosting system. proceedings of the 22nd acm sigkdd international conference on knowledge discovery and data mining, San Francisco, 13-17 August 2016, 785-794. https://doi.org/10.1145/2939672.2939785
39. Farid, D.M., Maruf, G.M., & Rahman, C .M. (2013). A new approach of boosting using decision tree classifier for classifying noisy data. 2013 International Conference on Informatics, Electronics and Vision (ICIEV), Dhaka, Bangladesh, 2013, pp. 1-4, https://dx.doi.org/10.1016/j.isprsjprs.2010.11.001