patchy distribution

The aim of this research was to investigate the spatial relationship between physical and chemical properties of soil and wheat yield in the conditions of competition with weeds in Isfahan weather conditions. 

In the crop year 2019-2020, field experiment was conducted in the form of a grid system with a distance of 2 x 2 meters. At each grid point, soil characteristics, grain yield of wheat and weed density were measured. The spatial distribution of the obtained data was analyzed using the geostatistical technique. 

The results showed that the grain yield has a strong spatial correlation with nitrogen, phosphorus, potassium, pH, and silt traits at the rates of 83.9, 78.3, 79.0, 80.1 and 81.7, respectively were in the range of 2.3, 3.2, 3.2, 3.7 and 4.0 meters. The areas where the wheat yield was lower were often in accordance with the areas where the sand and EC content of the soil were the highest. Alopecurus myosuroides had a strong spatial correlation with nitrogen and clay of soil, and moderate spatial correlation with soil sand; while this weed was found in soils with low amount of phosphor, potassium, pH and silt. Spatial correspondence between weed density and reverse grain yield was 60.7% in the effect range of 4.2 meters. 

The characteristics of the soil and the weed population in the field change from one place to another, and these variations can cause patchy and uneven distribution of yield in the field. Grain yield showed the highest spatial correlation with soil nitrogen.

Recent advances in precision farming technologies have triggered the need for highly flexible modelling methods to estimate, classificate and map weed population patterns for using in site-specific weed management. In this research, a learning vector quantization neural network (LVQNN) model was used to predict and classify the spatial distribution of Acroptilon repens L. density. This method was evaluated on data of A. repens L. density in a fallow field in Shahrood, Semnan province in 2010. Weed density assessments were performed following a 2 m × 2 m grid pattern on the field and a total of 550 sampling units on field. At each node of grid pattern, the numbers of A. repens L. seedlings were counted in the field within a permanent 50 cm by 50 cm quadrat. Some statistical tests, such as comparisions of the means, variance, statistical distribution as well as coefficient of determination in linear regression were used between the observed point sample data and the estimated weed seedling density surfaces to evaluate the performance of the pattern recognition method. Results showed that in training LVQNN, test and total phase P-value was greater than 0.7, 0.8 and 1 percent respectively, indicating that there was no significant (p<0.05) difference between statsitcal parameters such as average, variance, statistical distribution and also coefficient of determination in the observed and the estimated weed seedling density. This results suggest that LVQ neural network can learn weed density model very well. In addition, results indicated that trained LVQ neural network has a high capability in predicting weed density with recognition accuracy of 2.7 percent at unsampled points. The technique showed that the LVQNN could classify and map A. repens L. spatial variability on the field. Our map showed that patchy weed distribution offers large potential for using site-specific weed control on this field.