The field flux measurement was conducted at the Station of Rural Environmental, Research Center for Eco-Environmental Sciences (SRE-RCEES; 38∘71′ N, 115∘15′ E), located in Wangdu County, Hebei Province, China. The station is surrounded by vast agricultural fields, with winter wheat and summer maize rotation. More detailed descriptions of the measurement site can be found in our previous studies (Liu et al., 2017; Song et al., 2022a, b).

During the summer maize season, two different treatments were designed in the experiment fields: CK (control, normal flood irrigation but no fertilization for decades) and NP (fertilizer deep placement and normal flood irrigation, same as local farmers). In the experimental fields, summer maize was sown on 17 June 2021 and harvested on 27 September 2021. Only one fertilizer application event was conducted during the maize season. A typically used compound fertilizer (N:P2O5:K2O=28%:6%:6%) was mechanized and buried to a depth of 8–10 cm when sowing maize seeds. The fertilizer application rate (FAR) of the NP treatments was 300 kg N ha-1, which is the typical amount local farmers used. According to the fertilization event and the characteristics of HONO flux, the whole observation was divided into three periods: (1) pre-fertilization period (PFP, before 18 June); (2) high HONO emission period (HEP, from 18 June to 10 July); and (3) low HONO emission period (LEP, after 10 July).

HONO exchange fluxes were measured by an OTDC system, which is updated based on our previous design (Xue et al., 2019a) (i.d. of 32 cm, 80 cm in height). Eight chambers (six experimental chambers, denoted “Exp-chambers”, and two reference chambers, denoted “Ref-chambers”) were divided into two groups (NP and CK) to obtain HONO fluxes from soils with different treatments (see Sect. 2.1). As shown in Fig. 1, each group contains three replicated Exp-chambers and one Ref-chamber that are flushed by the same air pumped by diaphragm pumps (N838KNE, Germany) from the top of the metal sample tube (i.d. of 4 cm, 2 m in height, with the inner wall coated with Teflon film). The layout and construction of the OTDC system and other equipment are shown in Fig. 1. Note that both fluxes from the NP and CK plots include heterogeneous HONO formation on the ground surface, which leads to an overestimation of soil HONO flux (Xue et al., 2019a). By comparing between flux from the NP and CK groups, we can distinguish the relative importance of soil emissions and NO2 heterogeneous reactions and determine the net effect of fertilizer use. However, the current OTDC still has some problems, such as drying of the soil by flushing gas from the bottom, the greenhouse effect in the chambers, and the formation of HONO originating from soil NOx emissions, etc., which will result in the overestimation of the HONO flux. These issues should be considered in the future.

HONO was continuously sampled by stripping coils with the absorption solution of ultrapure water. The sampling interval was normally set at 12 h for all four chambers from the CK group during the whole measurement period. For the NP group, the sampling interval was set at 2 h for one Exp-chamber and one Ref-chamber during PFP and HEP and 12 h for the other two Exp-chambers in parallel. The 2 h interval measurements can provide detailed information about the diurnal variation. All samples were timely analyzed by an ion chromatography system (IC6200, WAYEE, China) (Xue et al., 2019b).

The soil HONO exchange flux (FHONO, ng N m-2 s-1) can be obtained by the difference of HONO concentrations in the Exp-chambers and Ref-chamber (Xue et al., 2019a): 1FHONO=HONOexp-HONOref×Fflush×MN×PR×T×S, where HONOexp, HONOref, Fflush, MN, P, R, T, and S are the HONO concentrations (ppbv) in the Exp-chambers and Ref-chambers, the flushing flow (3.33×10-4 m3 s-1), the molar mass of N (ng mol-1), the atmospheric pressure (Pa), the ideal gas constant (8.314 J mol-1 K-1), the atmospheric thermodynamic temperature (K), and the area of the soil covered by the chamber (m2), respectively.

The emission factor of HONO (EFHONO, %) relative to the amount of applied nitrogen can be calculated based on the following formula: 2EFHONO=ENP-ECKFAR×100%, where ENP and ECK are the cumulative HONO emissions (kg N ha-1) from the fertilized plots and the control plots, respectively. FAR is 300 kg N ha-1.

Meteorological parameters, including temperature, relative humidity (RH), pressure, wind speed (WS), wind direction (WD), precipitation, and solar radiation (SR), were recorded by an auto weather station (Vaisala WXT520, Finland). The photolysis frequency of NO2 (J(NO2)) was measured by a 2-π J(NO2) filter radiometer (MetCon, Germany). However, J(NO2) was not measured during 18–27 August. Instead, it was estimated via a high correlation between SR and J(NO2) (Fig. S1 in the Supplement, J(NO2)=-5.20×10-9 m4 W-2 s-2×SR+1.17×10-5 m2 W-1 s-1×SR; R2=0.91).

The soil temperature at ∼ 5 cm depth was auto-monitored by a sensor. The moisture of topsoil (0–5 cm) was expressed as water-filled pore space (WFPS), which was calculated by dividing the volumetric water content by the total soil porosity. The volumetric water content was measured twice a day (09:00 and 21:00 LT (local time); times for all instances in the text are in local time) by a soil humidity sensor (Stevens Hydra Probe II, USA) when collecting HONO samples, and total soil porosity was calculated according to the following relationship: soil porosity =(1-soil bulk density/2.65), assuming a particle density of 2.65 g cm-3, which is the most commonly used value for mineral soils (Linn and Doran, 1984). The topsoil samples were taken once a week by a ring sampler (5 cm diameter × 5 cm height), and the topsoil bulk density was determined gravimetrically by oven drying at 105 ∘C for 12 h.

To analyze the soil NH4+-N and NO3--N concentrations, soil samples were collected every 3 d for both NP and CK fields. Each sample was collected in four points and homogeneously mixed. Then 20 g of the mixed soil was extracted with 100 mL of 1 mol L-1 KCl solution, shaken in a rotary shaker (140 rpm for 1 h and filtered into sampling bottles. Samples were frozen until analyzed by a colorimetric continuous-flow analyzer (Seal Analytical AutoAnalyzer 3, USA).

Mean tests, variance tests, and correlation analysis for HONO fluxes and other parameters were performed using the statistical software SPSS Statistic 24 (SPSS Inc., Chicago, USA). The graphing software Origin 2018 (Origin Lab Corporation, Northampton, MA, USA) and ArcGIS 10.5 (ESRI Inc., California, USA) created figures.

The variations of key meteorology (air temperature, RH, pressure, rainfall, and J(NO2)) and soil parameters (soil NH4+-N and NO3--N concentrations, temperature, and WFPS) during the measurement period are shown in Figs. 2 and S2, respectively. During the whole maize growing season (from June to September), average air temperature, RH, and soil temperature were 24.8 ∘C, 72.3 %, and 23.7 ∘C, respectively. Except for the value less than 1.0×10-4 s-1, the average J(NO2) during the campaign was 3.12×10-3 s-1. The accumulative rainfall was 514 mm during this period, significantly higher than that in previous years (195–302 mm in the summer maize season during 2008–2011) (Zhang et al., 2014). The measured soil WFPS ranged from 33 % to 82 % and quickly increased following irrigation or precipitation, with a mean value of 59.1 %. In the NP plots, the soil NH4+-N and NO3--N concentrations increased significantly after applying nitrogen fertilizer (with averages of 48.0 and 112 mg kg-1 within 20 d after fertilization). In contrast, they remained at much lower levels throughout the whole maize season in the CK plots (with averages of 4.63 and 4.45 mg kg-1).

Figure 3 shows the time series of HONO fluxes at a 12 h interval (daytime: 09:00–21:00, nighttime: 21:00–09:00) from the NP and CK plots (FHONO-NP and FHONO-CK, average of three in-parallel duplications) during the whole maize season of 2021. HONO emission after fertilization showed a distinctly diurnal variation; that is, it was significantly higher in the daytime than at night, which will be discussed in the next section.

During the PFP period, the average FHONO-NP and FHONO-CK values were 0.55±0.35 and -0.51±0.13 ng N m-2 s-1 from the NP and CK plots, respectively (Table S1 in the Supplement). The higher level of FHONO-NP than FHONO-CK might be ascribed to the residual effect of fertilizer in the NP plots, considering that CK plots have not been fertilized for years. After fertilization, as shown in Fig. 3, FHONO-NP gradually increased and peaked on the 15th day. The measured FHONO-NP then trended downward but still maintained at a high level of 100 ng N m-2 s-1 within 3 weeks after fertilization. In contrast, FHONO-CK always fluctuated around zero. During HEP, the average FHONO-CK value was -0.36±0.04 ng N m-2 s-1, which is similar to the PFP period (Table S1). In comparison, the mean value of FHONO-NP during HEP was 97.7±8.6 ng N m-2 s-1, revealing the large potential of fertilized soils in HONO emissions. A heavy rainstorm (with a cumulative rainfall of more than 100 mm) occurred during 11–13 July, significantly limiting HONO release from the soil. During LEP, the average FHONO-NP and FHONO-CK values were 1.03 and -2.88 ng N m-2 s-1, respectively. Overall, except the data from NP plots during HEP, the other fluxes (including fluxes from CK plots during all periods and fluxes from NP plots during PFP and LEP) are similar to observations at other sites with no nitrogen fertilization application, such as grass (von der Heyden et al., 2022) or forest (Ramsay et al., 2018; Sörgel et al., 2015; Zhou et al., 2011) regions.

During the whole maize growing season, negative values of FHONO-CK were frequently observed at night, accounting for ∼ 70 % of the total investigated data. Numerous studies on HONO flux measurements from unfertilized fields also reported this phenomenon, which is ascribed to nocturnal HONO deposition (Laufs et al., 2017; Ren et al., 2011; Tang et al., 2020), indicating the complexity of the role of the ground surface in nocturnal HONO production and deposition (Ramsay et al., 2018; VandenBoer et al., 2013, 2015; von der Heyden et al., 2022). Nevertheless, it is worth noting that the high soil moisture after irrigation for the unfertilized CK plots may reduce the conversion from liquid-phase HNO2 to gas-phase HONO in soils and accelerate the absorption of atmospheric HONO at the soil surface, thereby contributing to the negative HONO fluxes.

Figure 4 shows the time series of FHONO-NP with an interval of 2 h, which shows similar levels and trends to 12 h interval measurements. Diurnal variations of HONOexp, HONOref, and FHONO-NP from the NP plot during PFP and HEP are shown in Fig. 5. During PFP, FHONO-NP exhibited a distinct diurnal variation (Fig. 5a), with a maximum of 0.38 ng N m-2 s-1 at noon and a minimum of 0.07 ng N m-2 s-1 in the early morning (Fig. 5a).

During HEP, HONOexp increased by 1–2 orders of magnitude compared to that during the PFP period. As shown in Fig. 5b, the FHONO-NP during HEP also showed bell-shaped diurnal variations but with a much higher peak of 152 ng N m-2 s-1 at noon and a minimum of 40 ng N m-2 s-1 in the early morning. Similar diurnal variation trends of HONO flux have been reported in previous studies (Tang et al., 2019, 2020; von der Heyden et al., 2022; Xue et al., 2019a; Zhou et al., 2011), while some studies also found morning peaks of the diurnal HONO flux (Laufs et al., 2017; Meng et al., 2022; Ren et al., 2011; von der Heyden et al., 2022). Laufs et al. (2017) and Ren et al. (2011) also conducted flux measurements in farmland, and they found high correlations between HONO flux and the product of NO2 concentrations ([NO2]) and J(NO2) or solar radiation. A similar good correlation was found during the PFP period of our present study (Fig. S3a, R2=0.89). These findings suggest that photosensitized heterogeneous reactions of NO2 on the soil surfaces may be the main sources of the observed HONO flux under typical non-fertilization conditions. However, after fertilizer application, HONO flux from the NP plot showed a weaker correlation with the product [NO2]×J(NO2) during HEP (Fig. S3b, R2=0.43). Moreover, the fluxes from the CK plots (no fertilization, representative for NOy-to-HONO conversion on the ground surface; see Sect. 3.2) are about 2 orders of magnitude lower than those from the NP plots during HEP. Therefore, the NO2 or other NOy reactions are not the main drivers of the observed HONO fluxes during HEP.

The observed HONO flux, measurement methods, and FAR in previous field measurements are shown in Table 1. With the increase of FAR, the emission of nitrogenous gas tends to show nonlinear growth (Grant et al., 2006; Ma et al., 2010; McSwiney and Robertson, 2005; Shcherbak et al., 2014; Xue et al., 2022b). Except for previous measurements at this agricultural site (Tang et al., 2019; Xue et al., 2022b, 2019a), fluxes from all other measurements are much lower than that in this study, which should be due to the difference in FAR. In this study, the maximum HONO flux at 2 h interval was 372 ng N m-2 s-1, which was in the range of the maximums of HONO flux in previous field measurements (up to 1515 ng N m-2 s-1) (Tang et al., 2019, 2020; Xue et al., 2019a). In three previous studies in the summer maize field at the SRE-RCEES station, the peaks occurred within 1 week after fertilization but their levels were variable depending on the FAR (Tang et al., 2019; Xue et al., 2022b, 2019a). However, the peak of HONO flux occurred approximately 2 weeks after fertilization during this campaign. As mentioned in Sect. 2.1, fertilizers were buried to a depth of 8–10 cm in this study, while they were spread on the soil surface in previous studies (Tang et al., 2019; Xue et al., 2022b, 2019a). Therefore, the relatively late occurrence of the HONO flux peak is probably caused by the slow dissolution due to the deep placement of nitrogen fertilizer. In addition to the impact on the occurrence time of peak flux, fertilizer application methods also lead to differences in levels of HONO emission fluxes. Compared with our study of fertilizer deep placement, Xue et al. (2019a) observed a peak HONO emission of 1515 ng N m-2 s-1 under the SF method, which is 4 times the peak in this study (372 ng N m-2 s-1), although their fertilizer rates are similar (300 kg N ha-1 in this study and 330 kg N ha-1 in Xue et al., 2019a). Our recent study found high exponential correlations between maximum HONO flux (Fmax) and FAR; that is, Fmax=4.7×exp⁡(FAR/57.3)+15.3 (R2=0.998, SF method, 0–350 kg N ha-1) (Xue et al., 2022b). With the FAR used in this study (300 kg N ha-1), the predicted Fmax is ∼ 890 ng N m-2 s-1, which is much higher than the observation, indicating that compared to the SF method, the DF method can reduce soil HONO emissions, possibly through minimizing the mineral nitrogen content in topsoil (Ke et al., 2018; Liu et al., 2015; Weber et al., 2015). Therefore, the DF method is expected to be able to help reduce reactive nitrogen emissions from agricultural activities but needs more future systematic assessments.