Improving wheat nitrogen use efficiency through a management zone-based optimization strategy by integrating crop model and remote sensing

Context

Precision nitrogen (N) management (PNM) is gaining attention for its ability to optimize wheat N use efficiency while reducing environmental impacts. Although both remote sensing platforms and crop modeling have been widely explored as key tools for PNM, limited research has integrated these approaches to dynamically optimize both basal and topdressing N rates on a large scale.

Objective

This study aimed to 1) estimate site- and season-specific target yields and basal N rates; 2) develop a new topdressing N strategy based on management zones (MZs) by integrating crop models with remote sensing data at the regional scale; and 3) evaluate its effectiveness in comparison with farmers’ conventional management (FCM) and the economic optimal N rate (EONR) strategies.

Methods

To achieve these objectives, county-level wheat yield data, vegetation indices, and a sustainable environmental index from Jiangsu Province were used to establish Random Forest (RF) and Quantile Random Forest models for estimating site- and season-specific target yields and basal N rates. MZs were delineated based on regional environmental heterogeneity, serving both for model calibration and determining the leaf area index (LAI) through remote sensing imagery. The WheatGrow model, calibrated by adjusting genetic coefficients, was employed to simulate wheat yield and phenology, showing strong potential for accurate simulation. Topdressing N rates were determined by calculating the LAI difference at the booting stage between topdressing and no-topdressing conditions, using a weather data fusion method. Field validation trials were conducted during the 2022–2023 wheat growing season across four MZs in Jiangsu Province.

Results

The results demonstrated that the RF model performed the best in yield prediction (R² = 0.94–0.95) and target yield determination. Higher target yields were observed in regions where the N recommendations were higher. Approximately 75 % of the N recommendations from the proposed PNM strategy were within 20 % of the measured EONR values. Field trials further revealed that the proposed PNM strategy improved partial factor productivity by 26–58 % compared to the FCM strategy while maintaining comparable yields and economic benefits.

Conclusion

In conclusion, the proposed PNM strategy offers a promising and scalable tool for variable-rate fertilization, with significant potential to enhance N use efficiency and promote sustainable agricultural development on a regional scale.