Study on spatial variability of soil nutrient is the basis of soil nutrient management in precision agriculture. For study on application potential and characteristics of algebra hyper-curve neural network(AHNN) in delineating soil properties spatial variability and interpolation, total 956 soil samples were taken for alkaline hydrolytic nitrogen measurement from a 50 hectares field using 20m*20m grid sampling. The test data set consisted of 100 random samples extracting from the total 956 samples, and the training data set was extracted from the other samples using 20m*20m, 40*40, 60m*60m, 80m*80m, 100m*100m and 120m*120m grid sampling respectively. Using AHNN model, three training plans were designed including plan AHC1 using spatial coordinates as the only network input, plan AHC2 adding 4 neighboring points’ information as network input and plan AHC3 adding 6 neighboring points’ information as network input. Comparing interpolation precision using AHNN method with Kriging method, following conclusions can be reached: when the number of training samples is bigger, interpolation precisions between Kriging and AHNN were similar; when the number was smaller, the precisions of both methods declined. With comparisons of three indexes of mean absolute error , root mean square error RMSE , mean relative error  and the generated spatial distribution maps using different methods, results shown that it can not simulate the characteristics of soil nutrient spatial variability well using spatial coordinates as the only network input, and the simulation degree can be improved greatly after adding neighboring sample points’ information, considering about the distance effect, as network input. When the number of samples was smaller, interpolation precision can be improve after proper increasing the number of neighboring sample points.

Results also shown that evaluation on interpolation precision using conventional error statistic indexes was not complete, and the spatial distribution maps should be used as an important evaluation indexes.