Prediction of nitrate-nitrogen leaching in paddy soil using multivariate analysis and artificial neural network

Nitrate-nitrogen leaching from agricultural areas is a major cause for groundwater pollution. Polluted groundwater with high levels of nitrate is hazardous and causes adverse health effects. This research aims to study nitrate–nitrogen leaching into groundwater in paddy soils at Ladang Merdeka Ismail Mulong in Kelantan,Malaysia. The assessment of soil physical properties and groundwater quality, effect of fertilizer on the groundwater nitrate-nitrogen concentrations, identification of contributory factors and prediction of nitrate-nitrogen leaching using computer modeling were the specific objectives of this study. A total of twelve observation wells and sixteen privately-owned wells were selected for groundwater quality monitoring. Their physical properties were measured by ASTM method while the concentrations of nitrate-nitrogen, phosphate and potassium were analyzed according to US-EPA 300.0, 365.2 and 200.2 procedures,respectively. Groundwater nitrate-nitrogen concentration ranged from 0.0 to 3.85 mg NO3--N/l in the obervation wells and 0.0 to 5.08 mg NO3 --N/l in the privateowned wells. These values fall below the permissible limit of 10 mg/l nitratenitrogen. However, the increasing trends of nitrate-nitrogen concentrations in the wells are of concern because it might accumulate over time and pollute the groundwater. The phosphate concentrations in 54.2% of observation wells and 36.7% of privateowned wells exceeded the permissible standard of 0.2 mg/l. The high values of phosphate could create problems related to the taste and odor of the water. The soil texture was classified as clay based on the United States Department of Agriculture (USDA) soil textural classification system. The results of soil bulk density (1.38g/cm3), porosity (38.8%) and soil penetration resistance (1.48 MPascal) confirmed the existence of a hard pan within the soil profile; (1) topsoil (0-30cm), (2) hard pan (30-60 cm) and (3) subsoil (below 60 cm). Nitrate leaching at different soil depths (20, 30 and 40 cm) was monitored using soil suction samplers for two consecutive seasons. The concentrations varied from 1.10- 11.70 mg NO3--N /l and 1.20-3.78 mg NO3--N/l in the first and second season,respectively. The results showed that nitrate-nitrogen concentration in the soil increased with soil depth and higher fertilizer application indicating that fertilizer application influences the leaching process which leads to the accumulation of nitrate-nitrogen in the soil. The total nitrogen loss was 0.93% to 1.30% of the applied nitrogen with the highest leaching rate at the 40 cm soil layer (0.35 kg NO3 --N/ha/d) indicating soil contamination and causes nitrate build-up in the groundwater above permissible limit, thus rendering it unsuitable for human consumption. The complex data matrix (128 x 16) of nitrate-nitrogen parameters was subjected to multivariate analysis mainly principal component analysis (PCA) and discriminant analysis (DA). PCA extracted four principal components from this data set which explained 86.4% of the total variance. Analysis using Alyuda Forecaster software confirmed that the most important contributors were soil physical properties (R2 =0.98). Discriminant analysis was used to evaluate the temporal variation in soil nitrate-nitrogen on leaching process. Discriminant analysis gave four parameters (hydraulic head, evapotranspiration, rainfall and temperature) contributing more than 98% correct assignments in temporal analysis. DA allowed reduction in dimensionality of the large data set which defines the four operating parameters most efficient and economical to be monitored for temporal variations. Four different data sets were used to develop predictive nitrate-nitrogen models in an Artificial Neural Network (ANN) environment. The results showed good agreement between predicted and observed nitrate-nitrogen leaching rate for TD-ANN model with coefficient correlations of R = 0.98 in the testing step. Based on the principal component analysis scores, ANN generated two models, PCS-ANN1and PCSANN2,which gave good predictions with R = 0.97 and 0.94 in their respective testing steps. An inspection of the results showed that ANN gave reliable predictive models with acceptable accuracies. The results of this study indicate that ANN can be reliably used as a tool to predict nitrate-nitrogen leaching rates in paddy soils based on the selected sixteen parameters

Preview PDF FPAS 2012 15R.pdf Download (1MB) | Preview

Actions (login required)

View Item