Field-calibrated ammonia volatilization models for rice and wheat, validated in APSIM

<title>Abstract</title> <p> Predicting ammonia (NH ₃ ) volatilization from rice and wheat is vital for improving nitrogen-use efficiency (NUE). However, process-based models often require detailed soil-atmosphere data that are not always available. Using data from five field experiments in the Indo-Gangetic Plain (IGPs), we developed and statistically tested two empirically calibrated models for NH ₃ volatilization, then integrated and validated them within the Agricultural Production Systems sIMulator (APSIM) Next Generation platform. A linear model linking cumulative NH ₃ -N loss to nitrogen rate and temperature explained 94.3% of the variation in rice (RMSE = 0.69 kg N ha-1; n = 16), with nitrogen rate as the main factor (+ 0.048 kg N ha ^− 1 per kg N applied) and temperature a smaller yet significant factor (+ 0.60 kg N ha ^− 1 per gradient level). The second model, considering tillage, basal-timing, and nitrogen source/placement for wheat, accounted for 94.2% of variance (RMSE = 1.62 kg N ha-1; n = 39), with timing being the dominant factor while tillage had no significant independent effect. Both models, embedded as APSIM Manager-script components, closely matched their predictions to three decimal places (rice: 14.024 kg N ha ^− 1 ; wheat: 21.640 kg N ha ^− 1 ), simulating daily dynamics through mechanistic pathways where ammonium pools peaked 3–4 days after application. Urease inhibitors and botanical amendments reduced cumulative NH ₃ -N loss by 17–40% compared to urea alone. These results show that simple, field-calibrated linear models can effectively simulate rice and wheat NH3 volatilization over a season, offering accuracy comparable to complex process-based models while capturing realistic daily behavior within an established crop-soil simulation platform. </p>