Apparent soil electrical conductivity and gamma-ray spectrometry to map particle size fraction in micro-irrigated citrus orchards in California.

Abstract

In specialty crops, water and nutrient management may be optimized using accurate, high-resolution soil maps, especially in resource-constrained farmland, such as California. We evaluated the use of soil apparent electrical conductivity (EC_a) and gamma-ray spectrometry (GRS) to map particle size fraction across three micro-irrigated non-saline citrus orchards in California. Our research showed that EC_a was a reliable predictor of soil texture, particularly sand and silt contents, with Pearson correlation coefficients (r) as high as -0.92 and 0.94, respectively, at the field level. Locally-adjusted analysis of covariance (ANOCOVA) regressions using EC_a data returned accurate sand, silt, and clay content estimations with mean absolute errors (MAE) below 0.06, even when calibrated with a limited dataset (n=5 per field). On the other hand, we observed mixed results with GRS. We observed negative correlations between GRS total counts and sand content over the entire dataset (r = -0.55). However, one site (Strathmore) showed a field-scale positive correlation (r = 0.88). Clay content significantly correlated with gamma-ray total counts (TC) over the entire dataset (r = 0.37) but not at the field scale. Additional soil data analyses using GRS radionuclide ratios and soil laboratory analyses using diffuse reflectance infrared Fourier transform spectroscopy and acid ammonium oxalate extractable elements indicated unique geochemical and mineralogical characteristics in Strathmore, suggesting that factors such as soil mineralogy influenced the GRS measurements. This inconsistency prevented the development of a multi-field GRS-based soil texture ANOCOVA model. These findings confirm that EC_a is highly effective for soil texture mapping in non-saline soils using linear modeling, while GRS may require field-specific calibration due to variations in local mineralogy. Integrating multi-sensor data is a viable means for reducing ground-truthing requirements and related costs, and improving the quality and accuracy of soil maps in agriculture.