Predicting soil nutrient contents using Landsat OLI satellite images in rain-fed agricultural lands, northwest of Iran

Soil nutrients are the key factors in soil fertility, which have important roles in plant growth. Determining soil nutrient contents, including macro and micronutrients, is of crucial importance in agricultural productions. Conventional laboratory techniques for determining soil nutrients are expensive and time-consuming. This research was aimed to develop linear regression (LR) models for remote sensing of total nitrogen (TN) (mg/kg), available phosphorous (AP) (mg/kg), available potassium (AK) (mg/kg), and micronutrients such as iron (Fe) (mg/kg), manganese (Mn) (mg/kg), zinc (Zn) (mg/kg), and copper (Cu) (mg/kg) extracted by DTPA in rain-fed agricultural lands in the northwest of Iran. First, 101 soil samples were collected from 0-30 cm of these lands and analyzed for selected nutrient contents. Then a linear regression along with principal component analysis was conducted to correlate soil nutrient contents with reflectance data of different Landsat OLI bands. Finally, the spatial distributions of soil nutrients were drawn. The results showed that there were linear relationships between soil nutrient contents and standardized PC1 (Z(PC1)). The highest significant determination coefficient with an R-2 value of 0.46 and the least relative error (%) value of 11.97% were observed between TN and Z(PC1). The accuracy of the other LR's developed among other soil nutrient contents and remotely sensed data was relatively lower than that obtained for TN. According to the results obtained from this study, although remote sensing techniques may quickly assess soil nutrients, new techniques, technologies, and models may be needed to have a more accurate prediction of soil nutrients.

Automated summary

The study developed linear regression models for remote sensing of soil nutrients in rain-fed agricultural lands in the northwest of Iran. Results showed linear relationships between soil nutrient contents and PC1, with the highest accuracy observed for TN. However, the accuracy of LR models for other soil nutrient contents was relatively lower, indicating the need for new technologies to improve soil nutrient prediction accuracy.