Achieving sustainable fertilization is critical for balancing crop productivity with environmental stewardship and resource efficiency. However, conventional fertilization methods rely on fixed schedules and generalized routines, resulting in inefficient nitrogen use and environmental risks owing to over- or under-application. Sustainable fertilization requires adaptive strategies that optimize resource use while preserving long-term soil health and productivity. Reinforcement learning (RL) offers a promising alternative by continuously adapting fertilization strategies based on real-time data, such as soil conditions, crop growth stages, and weather patterns. This study introduced the time-aware, idle-biased, Lagrangian-based, and resource-constrained approach with proximal policy optimization (TILARC-PPO), a novel RL framework designed to adaptively optimize fertilization. TILARC-PPO integrates (1) idle-biased action selection to prevent unnecessary fertilization, (2) time-awareness to optimize decision timing, and (3) Lagrangian-based resource constraints to dynamically regulate nitrogen applications. Experimental results show that TILARC-PPO maintains a comparable grain yield with only a slight reduction of 7.93%, while reducing nitrogen consumption by 32% when compared to expert fertilization. Additionally, it achieved the highest nitrogen use efficiency (30.8 kg grain per kg N), surpassing both the expert-based and vanilla proximal policy optimization (PPO) approaches. TILARC-PPO further improved training stability and policy convergence by learning effective fertilization strategies within 300,000 timesteps. These findings highlight TILARC-PPO as a scalable, intelligent solution for sustainable precision agriculture, aligned with global efforts to enhance resource efficiency, maintain soil health, and promote sustainable food production.

1. [1] P. Krasilnikov, M. A. Taboada, and Amanullah, “Fertilizer use, soil health and agricultural sustainability,” Agriculture, Vol.12, No.4, Article No.462, 2022. https://doi.org/10.3390/agriculture12040462
2. [2] S. Nath, “A vision of precision agriculture: Balance between agricultural sustainability and environmental stewardship,” Agronomy J., Vol.116, No.3, pp. 1126-1143, 2024. https://doi.org/10.1002/agj2.21405
3. [3] Food and Agriculture Organization of the United Nations, “The future of food and agriculture: Trends and challenges,” 2017. http://www.fao.org/3/i6583e/i6583e.pdf [Accessed March 2, 2025]
4. [4] M. Li and R. S. Yost, “Management-oriented modeling: Optimizing nitrogen management with artificial intelligence,” Agricultural Systems, Vol.65, No.1, pp. 1-27, 2000. https://doi.org/10.1016/S0308-521X(00)00023-8
5. [5] D. Wright et al., “Field corn production guide: SS-AGR-85/AG202, rev. 8/2022,” EDIS, Vol.2022, No.4, 2022. https://doi.org/10.32473/edis-ag202-2022
6. [6] S. Sela, H. M. van Es, B. N. Moebius-Clune, R. Marjerison, and G. Kneubuhler, “Dynamic model-based recommendations increase the precision and sustainability of N fertilization in midwestern US maize production,” Computers and Electronics in Agriculture, Vol.153, pp. 256-265, 2018. https://doi.org/10.1016/j.compag.2018.08.010
7. [7] G. Mandrini, C. M. Pittelkow, S. V. Archontoulis, T. Mieno, and N. F. Martin, “Understanding differences between static and dynamic nitrogen fertilizer tools using simulation modeling,” Agricultural Systems, Vol.194, Article No.103275, 2021. https://doi.org/10.1016/j.agsy.2021.103275
8. [8] K. J. Boote, J. W. Jones, G. Hoogenboom, and J. W. White, “The role of crop systems simulation in agriculture and environment,” Int. J. of Agricultural and Environmental Information Systems, Vol.1, No.1, pp. 41-54, 2010.
9. [9] D. Holzworth et al., “APSIM next generation: Overcoming challenges in modernising a farming systems model,” Environmental Modelling & Software, Vol.103, pp. 43-51, 2018. https://doi.org/10.1016/j.envsoft.2018.02.002
10. [10] G. Hoogenboom et al., “The DSSAT crop modeling ecosystem,” K. Boote (Ed.), “Advances in Crop Modelling for a Sustainable Agriculture,” pp. 173-216, Burleigh Dodds Science Publishing, 2019. https://doi.org/10.19103/AS.2019.0061.10
11. [11] A. R. Kemanian et al., “The cycles agroecosystem model: Fundamentals, testing, and applications,” SSRN, 2022. https://doi.org/10.2139/ssrn.4188402
12. [12] P. Steduto, T. C. Hsiao, D. Raes, and E. Fereres, “AquaCrop—The FAO crop model to simulate yield response to water: I. Concepts and underlying principles,” Agronomy J., Vol.101, No.3, pp. 426-437, 2009. https://doi.org/10.2134/agronj2008.0139s
13. [13] A. de Wit et al., “25 years of the WOFOST cropping systems model,” Agricultural Systems, Vol.168, pp. 154-167, 2019. https://doi.org/10.1016/j.agsy.2018.06.018
14. [14] R. Gautron, O.-A. Maillard, P. Preux, M. Corbeels, and R. Sabbadin, “Reinforcement learning for crop management support: Review, prospects and challenges,” Computers and Electronics in Agriculture, Vol.200, Article No.107182, 2022. https://doi.org/10.1016/j.compag.2022.107182
15. [15] H. Overweg, H. N. C. Berghuijs, and I. N. Athanasiadis, “CropGym: A reinforcement learning environment for crop management,” arXiv:2104.04326, 2021. https://doi.org/10.48550/arXiv.2104.04326
16. [16] M. G. J. Kallenberg, H. Overweg, R. van Bree, and I. N. Athanasiadis, “Nitrogen management with reinforcement learning and crop growth models,” Environmental Data Science, Vol.2, Article No.e34, 2023. https://doi.org/10.1017/eds.2023.28
17. [17] M. Turchetta et al., “Learning long-term crop management strategies with CyclesGym,” Proc. of the 36th Int. Conf. on Neural Information Processing Systems, pp. 11396-11409, 2022.
18. [18] T. D. Kelly, T. Foster, and D. M. Schultz, “Assessing the value of deep reinforcement learning for irrigation scheduling,” Smart Agricultural Technology, Vol.7, Article No.100403, 2024. https://doi.org/10.1016/j.atech.2024.100403
19. [19] M. Madondo et al., “A SWAT-based reinforcement learning framework for crop management,” AAAI Conf. on Artificial Intelligence, 2023.
20. [20] M. Alkaff, A. Basuhail, and Y. Sari, “Optimizing water use in maize irrigation with reinforcement learning,” Mathematics, Vol.13, No.4, Article No.595, 2025. https://doi.org/10.3390/math13040595
21. [21] R. Gautron et al., “gym-DSSAT: A crop model turned into a reinforcement learning environment,” arXiv:2207.03270, 2022. https://doi.org/10.48550/arXiv.2207.03270
22. [22] K. Yamashita and T. Hamagami, “Reinforcement learning for POMDP environments using state representation with reservoir computing,” J. Adv. Comput. Intell. Intell. Inform., Vol.26, No.4, pp. 562-569, 2022. https://doi.org/10.20965/jaciii.2022.p0562
23. [23] F. Pardo, A. Tavakoli, V. Levdik, and P. Kormushev, “Time limits in reinforcement learning,” Proc. of the 35th Int. Conf. on Machine Learning, pp. 4045-4054, 2018.
24. [24] R. S. Sutton and A. G. Barto, “Reinforcement learning,” MIT Press, 1998.
25. [25] J. Schulman, F. Wolski, P. Dhariwal, A. Radford, and O. Klimov, “Proximal policy optimization algorithms,” arXiv:1707.06347, 2017. https://doi.org/10.48550/arXiv.1707.06347
26. [26] P. Hintjens, “ZeroMQ: Messaging for many applications,” O’Reilly Media, 2013.
27. [27] G. Brockman et al., “OpenAI Gym,” arXiv:1606.01540, 2016. https://doi.org/10.48550/arXiv.1606.01540
28. [28] V. Mnih et al., “Playing Atari with deep reinforcement learning,” arXiv:1312.5602, 2013. https://doi.org/10.48550/arXiv.1312.5602
29. [29] V. Mnih et al., “Asynchronous methods for deep reinforcement learning,” arXiv:1602.01783, 2016. https://doi.org/10.48550/arXiv.1602.01783
30. [30] A. Bhatia, P. Varakantham, and A. Kumar, “Resource constrained deep reinforcement learning,” Proc. of the Int. Conf. on Automated Planning and Scheduling, Vol.29, pp. 610-620, 2019. https://doi.org/10.1609/icaps.v29i1.3528
31. [31] L. A. Hunt and K. J. Boote, “Data for model operation, calibration, and evaluation,” G. Y. Tsuji, G. Hoogenboom, and P. K. Thornton (Eds.), “Understanding Options for Agricultural Production,” pp. 9-39, 1998. https://doi.org/10.1007/978-94-017-3624-4_2
32. [32] C. W. Richardson, “Weather simulation for crop management models,” Trans. of the ASAE, Vol.28, No.5, pp. 1602-1606, 1985. https://doi.org/10.13031/2013.32484
33. [33] A. Soltani and G. Hoogenboom, “A statistical comparison of the stochastic weather generators WGEN and SIMMETEO,” Climate Research, Vol.24, pp. 215-230, 2003. https://doi.org/10.3354/cr024215
34. [34] T. F. Morris et al., “Strengths and limitations of nitrogen rate recommendations for corn and opportunities for improvement,” Agronomy J., Vol.110, No.1, pp. 1-37, 2018. https://doi.org/10.2134/agronj2017.02.0112
35. [35] B. Vanlauwe et al., “Agronomic use efficiency of N fertilizer in maize-based systems in sub-Saharan Africa within the context of integrated soil fertility management,” Plant and Soil, Vol.339, pp. 35-50, 2011. https://doi.org/10.1007/s11104-010-0462-7