| The soil and nutrient loss in the Yanhe River Basin of the loess hilly and gully region was serious.There wre few studies on the multi-temporal characteristics of non-point source nitrogen and phosphorus loss about year,month and day in the current literature.This study used the coupling method of indoor artificial simulated rainfall and SWAT model to reveal the runoff,sediment and nitrogen loss rules on bare loess soil under the rainfall intensity of 45-120mm/h,simulate annual,monthly and daily runoff and NH4+-N changes from 1984 to 2012,and evaluated the scale effect relationship about runoff and non-point source NH4+-N load at different time scales,the main conclusions were as follows:(1)The linear fitting coefficient of soil infiltration rate,runoff and rainfall intensity on loessial soil were all greater than 0.8 under three bare slopes of 10°,15°and 25°,and there was good positive correlation.When the slope is 25°,the nutrient loss had a strong linear correlation with slope runoff,and it had a significant power function relationship with sediment yield;the initial concentration of NO3--N was high,then decreased with the rainfall duration,it had obvious initial flushing effect;the initial loss concentration of NH4+-N increased from 0.326mg/L under rainfall intensity of 90mm/h to 0.384mg/L of120mm/h,but its concentration decreased continuously with rainfall duration;total concentration of TN loss when the rainfall intensity was 90,105 and 120 mm/h,it was0.6056,0.8011 and 1.3076 mg/L,respectively,which increased with the increase of rainfall intensity.At 15°slope,nitrogen loss was dominated by particulate form under different rainfall intensities,with an average of about 72%.However,the proportion of particulate nitrogen decreased first and then increased with the increase of rainfall intensity.(2)Based on the SWAT model,the parameters about the annual,monthly,daily runoff and annual and monthly NH4+-N loads in the catchment area of the Ganguyi station in Yanhe River Basin were determined and verified.The calibration and verification periods of annual and monthly runoff were 1984-1993 and 1994-2000 respectively,the annual Nash coefficient was 0.71,the verification was 0.66;the monthly Nash coefficient was 0.78,and the verification was 0.78.The calibration period of daily runoff was 1991-1995,and its Nash coefficient was 0.57,the verification period was 1996-2000,its Nash coefficient was0.58.The calibration evaluation of NH4+-N was based on the monthly load of 1985-1987,its determination coefficient was 0.55 and the deviation percentage was-5.91%.The verification of monthly NH4+-N load was 1988-1989,and its determination coefficient was0.74,the percentage of deviation was 19.92%,the annual NH4+-N load from 1985 to 1989was used as annual verification data,its determination coefficient was 0.75 and the deviation percentage was 4.04%.Both the runoff and the NH4+-N satisfied the accuracy requirements,so the established SWAT model had good applicability in this basin.(3)There was a significant positive correlation between runoff and rainfall at annual,seasonal,monthly and daily scales from 1984 to 2012.Among them,the correlation coefficient between annual runoff and annual rainfall was as high as 0.92;the seasonal distribution of runoff was significantly different,which was as follows:summer>autumn>spring>winter,where summer runoff accounted for 44.20%of the whole year.The correlation coefficient between monthly runoff and monthly rainfall was 0.87,the runoff first increased and then decreased,showing a“single peak”trend.The correlation coefficient between daily runoff and daily rainfall was 0.72,which decreased slowly over the years;it was founded that when the runoff was greater than 20m3/s,the trend of the duration curves of each stage was extremely high,but when the runoff was 1-20m3/s,the changing trend was obviously different,which the diachronic curve analysis of daily runoff;the significance of correlations between runoff and rainfall in the Yanhe river basin on multi-temporal scale were:summer>autumn>year>month>spring>day>winter;in simulation,the trend of runoff was similar at each time scale,but the time differentiation degree was different,among which the daily runoff differentiation was highest,followed by the monthly runoff,while the annual runoff was lowest,the three time variation coefficients were:2.41,1.17 and 0.18,there the larger the time scale was,the smaller the degree of time differentiation was.(4)The maximum loss of non-point source NH4+-N was 235.1 tons from 1984 to 2012,the minimum was 54.15 tons.The NH4+-N loss was most affected by runoff,the corresponding Pearson correlation coefficient was 0.86.The NH4+-N loss in different typical years was quite different:high flow year>median flow year>low flow year.The seasonal loss of NH4+-N had a certain synchronization with the rainfall process,which showed that the larger the runoff,the more NH4+-N loss;while the monthly load had a strong correlation with rainfall and runoff,and was affected by runoff more than rainfall,and its loss was larger in July and August and smaller in December.The fractal dimension of non-point source NH4+-N load in the year,season,month,flood season and non-flood period was between 0 and 1 with obvious fractal characteristics;the time variability of NH4+-N was between 0.66 and 1.02,the maximum variability in the flood season was 1.02,followed by the annual scale of 0.95,the variability of monthly NH4+-N load was 0.9,and the rainstorm events was 0.85,the NH4+-N loss variation was close in autumn and winter,and the change of NH4+-N loss in the non-flood period was smallest with the time variability of 0.66. |