The authors declare that there are no competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Nitrogen leaching as a direct pathway of N loss from agricultural land can negatively affect groundwater and surface water quality. However, a simple and efficient method for nitrogen leaching loss estimation is still inefficient. In this study, an exponential model was developed using the experimental data from a two-year field experiment conducted in the Taihu Lake region of China to simulate the N leaching from the paddy soil. The results showed the leached N was in the range of 5.66 to 8.45 kg N/ha during the whole rice season, which was accounted for 1.7%-2.1% of the applied N. A good agreement between the measured and model predicted results for N leaching loss was observed, suggesting the validity of the established model. The model was further validated with the data of other studies in other regions. The results demonstrated this model is able to simulate the N leaching loss accurately and can provide a beneficial tool for users to predict N leaching loss in paddy soil.

Rice is the most important crop in China, which makes up 43.7% of the national grain production

Leaching is one main pathway of N losses in paddy soil, which normally accounts for 0.1%-4.9% of the applied N

Mathematical modeling has been proven to be a powerful tool in predicting N losses _{4}-N), NO_{3}-N and organic N are the main forms in the leachate of paddy soil while NO_{3}-N is the dominant form in dryland leachate

Rao et al.

Therefore, the objective of this study was to quantify N leaching losses in a paddy soil in TLR for 2 consecutive years. Then, a simple and efficient model as a predictive tool for nitrogen leaching in the TLR was developed and calibrated with additional data. Hopefully, this study can provide an effective tool for N leaching estimation in paddy soil.

The field experiment was conducted at the Yixing city of Jiangsu province near the Taihu Lake (31°16´ N, 119°54´ E) from 2013

The experimental plots were arranged randomly with 3 N treatments and 3 replications with individual plot 44 m^{2}. The 3 N treatments were as follows: N0 (control, 0 kg N/ha), N1 (220 kg N/ha) and N2 (270 kg N/ha), in which N2 was the local treatment. Urea was used as the source of N with 40% basal applied prior to transplanting, 30% top dress at the tillering stage, and the remaining 30% top dress at the ear differentiation stage. Phosphorous (P) and potassium (K) fertilizers were applied basally in the form of superphosphate at a rate of 60 kg P_{2}O_{5}/ha and in the form of KCl at a rate of 45 kg K_{2}O/ha. All fertilizers were surface applied. Local cultivation and field management practices were adopted

Percolation water was collected at 40 cm, 80 cm, and 120 cm soil depth in each N treatment with a vacuum pump at 7-day to 10-day intervals during the rice season as described in Zhao et al ^{o}C in a freezer until analysis. Concentrations of NH_{4}-N, NO_{3}-N and total N (TN) in water samples were analyzed with a continuous-flow N analyzer (Skalar, Netherlands).

The total volume of leached water during the rice season is defined as the rate of surface water vertical percolation (mm/day) × flooded periods (day) × plot area. The average leaching rate of 2 mm/day measured in previous study was adopted in this study

To minimize the influence of root growth and underground water on N leaching loss calculation, we used the TN concentration of percolation water extracted at 120 cm soil depth for N leaching loss estimation.

First, the N leaching loss at each sampling time was calculated by Equation (1):

^{-2} ……..(1)

Where

Then, the N leaching loss ratio of each sampling relative to the cumulative loss was calculated by Equation (2):

Pi = Ni/

….(2)

Where

Then, the cumulative N leaching loss ratio up to the ^{th} sampling was calculated by Equation (3):

Where

Finally, an exponential model was fitted to the N leaching process (Equation 4).

^{-t/k} …….(4)

Where Y is the cumulative N leaching loss ratio, t is the time after basal N fertilizer application (d), a, b and k are constant.

With the increase of urea application, N concentrations in leachates were increased significantly (_{4}-N concentration at 40 cm soil depth was increased from 0.3 mg/L to 8.4 mg/L and 7.3 mg/L respectively when 220 kg N/ha and 270 kg N/ha urea were applied. Accordingly, the peaks of NH_{4}-N concentration at 80 cm and 120 cm soil depths were 2.6 mg/L, 4.2 mg/L and 2.0 mg/L, 4.9 mg/L respectively at 8 days after urea was applied. The NO_{3}-N concentrations at all three soil depths were increased gradually with time, and peaked at 29 days after urea was applied. The highest NO_{3}-N concentration in samples collected at 40 cm, 80 cm and 120 cm soil depths were 7.8 mg/L, 10.2 mg/L, 4.8 mg/L with 220 kg N/ha input and 17.0 mg/L, 17.0 mg/L, 13.4 mg/L with 270 kg N/ha input, respectively. The TN concentrations peaked at 7 days and 29 days after urea application. The highest TN concentrations in 40 cm, 80 cm and 120 cm soil depths were 18 mg/L, 11.2 mg/L, 6.2 mg/L respectively with 220 kg N/ha input and 18.9 mg/L, 19.6 mg/L, 10 mg/L with 270 kg N/ha input. The organic N concentrations at 40 cm, 80 cm and 120 cm soil depths were also increased first and then decreased, which accounted for 25%, 17% and 10% of total N losses on average

The N concentrations were generally decreased with the increase of soil depth. The average NH_{4}-N concentration was decreased by 0.6 mg/L when soil depth was increased from 40 cm to 120 cm. At the same time, the average NO_{3}-N and TN concentrations were decreased about 1.1 mg/L and 2.3 mg/L respectively as the depth increased from 40 cm to 120 cm. The N concentrations of the percolation water collected at 40 cm soil depth varied significantly, while they were stable at 120 cm. The N concentrations in the percolation water showed similar trend in 2013 and 2014 rice season (2013 data not shown).

The estimation of N leached at 120 cm depth were 5.66 to 8.45 kg N/ha during two rice seasons (

Year | Treatment | NH_{4}-N(kg N/ha) |
NO_{3}-N(kg N/ha) |
TN(kg N/ha) | Loss Percentage (%) |

2013 | N0 | 1.03±0.8 | 1.04±0.62 | 2.83±1.81 | - |

N1 | 2.75±1.27 | 3.06±1.38 | 6.79±3.82 | 1.8 | |

N2 | 4.41±2.44 | 3.34±2.07 | 8.45±3.78 | 2.1 | |

2014 | N0 | 0.64±0.23 | 1.22±0.71 | 2.03±1.02 | - |

N1 | 1.59±0.9 | 3.44±1.45 | 5.66±2.76 | 1.7 | |

N2 | 2.88±1.72 | 3.59±2.21 | 7.04±2.81 | 1.9 |

We used the developed model to simulate the N leaching loss from the paddy soil. The N leaching loss data collected in 2013 field experiments was used for model development.

As shown in ^{2}

We validated the model using the 2014 rice season N leaching loss data. A good agreement between the measured and model predicted results was observed (

After validation with our data, we further validated the developed model with N leaching loss data from other studies in nearby regions in Jiangsu Province (

Source | N application rate(kg N/ha) | Experiment year | Soil depth (cm) |

Zhu et al. |
N1 (200)N2 (250)N3 (300) | 1996 | 90 |

Tian et al. |
N1 (180)N2 (255)N3 (330) | 2002-2004 | 90 |

Wang et al. |
N1 (225)N2 (300) | 2001-2003 | 90 |

Zhang et al. |
N1 (360) | 2006 | 100 |

Li et al. |
N1 (220) | 2008 | 100 |

Yu et al. |
N1 (210)N2 (270) | 2009-2010 | 100 |

Zhao et al. |
N1 (300) | 2007-2009 | 100 |

Zhao et al. |
N1 (81)N2 (135)N3 (189)N4 (216)N5 (243)N6 (270) | 2011 | 100 |

Chen et al. |
N1 (200)N2 (270) | 2011 | 120 |

This study | N1 (220)N2 (270) | 2013-2014 | 120 |

We also applied the developed model to simulate the N leaching loss process in other regions (

Nitrogen leaching loss is one of the main pathways for N loss in paddy soil

Compared with other simple models, such as models developed in Chowdary et al.

Furthermore, the findings of this study can help address two questions regarding N leaching process. Firstly, how is the N leaching rate changed during rice growing season? We found that the N leaching rate was high in the early stage of rice growth. Then, the N leaching rate decreased with time. Thus, in order to reduce the N leaching loss, less N should be applied during the initial planting time. Secondly, how many sampling times are necessary to accurately determine the amount of leached N? High frequency of sampling would increase the accuracy of results, but it consumes too much time and labor. By using the developed model, we found the N leaching loss can be calculated only through 4 times sampling which could reduce the expenses of data collection significantly. This can increase the efficiency of leaching N research in TLR and other regions with similar environmental conditions and management practices.

In this study, an exponential model was developed to simulate N leaching loss from the paddy soil in the Taihu Lake region (TLR) of China by using the N leaching data from a two-year field experiment. The model was successfully validated with different data from the same region and from nearby regions to accurately simulate N leaching loss. This model is proven to be simple and efficient to predict N loss from paddy soil.

This study was financially supported by the National Natural Science Foundation of China (No. 51778301), the Key Special Program on the S&T for the Pollution Control (2017ZX07202004), the Postgraduate Research & Practice Innovation Program of Jiangsu Province (KYCX18_1246) and the Special Environmental Research Fund for Public Welfare of the State Environmental Protection Administration of China (201309035).