Location: Agroecosystem Management ResearchTitle: Detecting variability in saline-sodic soil using remote sensing methods in the Northern Great Plain
|MALO, DOUGLAS - South Dakota State University|
|CLAY, DAVID - South Dakota State University|
Submitted to: Meeting Abstract
Publication Type: Abstract Only
Publication Acceptance Date: 6/6/2021
Publication Date: N/A
Interpretive Summary: This study clearly showed using geospatial statistics particularly, local Moran’s I, semivariogram modelling of soil attributes, and NDVI data, could help to quantify spatial heterogeneity in saline-sodic soils. Thus, a better understanding of the spatial pattern of the measured soil variables in saline sodic soils can easily be captured. It also showed soil series variation for all the measured soil attributes and demonstrates the need to further explore and examine other soil attributes not covered in this study. Integrating high resolution imagery could be an area of future research in saline-sodic soil.
Technical Abstract: Quantifying spatial heterogeneity in saline-sodic soil using remote sensing methods is critical to make sound management decisions. Soil spatial variability in the Northern Great Plains of USA is related to topographic, vegetation, time, parent material, climate and anthropogenic (management and land use change). The objective of this study was to describe the spatial variability of selected saline sodic soil properties using remote sensing methods at a landscape scale. The study was conducted in a 55.4 ha field at Pierpont, SD. The field was planted with corn (Zea mays) in 2014. A total of 169 grid points with the area of 62 x 62 m grid were laid out in the field in 2014 for sample collection. Soil pH and electrical conductivity (EC) of the samples collected from each grid points were measured. Sodium Adsorption Ratio (SAR) were derived from cation analysis. Mollic depth and lime depth were measured at each grid points. Semivariograms fit for exponential, spherical, and Gaussian models were tested. Spatial class was developed using nugget to sill ratio. Global Moran’s I and local Moran’s I statistics were performed. The exponential model was the optimum fit for mollic depth, lime depth, pH, EC, and SAR with nugget to sill ratio of 0, 0, 45, 17, and 49, respectively. EC and SAR showed moderate spatial dependence whereas the other parameters showed strong spatial dependence. At the V1, V4, and V6 growth stages the exponential model was the optimum fit for NDVI with a value of nugget to sill ratio of 23, 0, and 25, respectively. At all plant growth stages the NDVI had showed strong spatial dependence. Analyses of variance of all the parameters measured were significantly different at P < 0.05. Mollic depth, lime depth, and EC showed slight positive spatial autocorrelation with Moran’s statistic value of 0.193, 0.106, and 0.337. So, the null hypothesis of random distribution was rejected for these variables. Whereas the Global Moran’s I statistics value and the z-score of SAR was very small and statistically non-significant indicating no spatial dependency patterns of local spatial autocorrelation were assessed from a generated map using Local Moran’s I. Semivariogram modelling and Moran’s I of soil attributes and NDVI data can help to quantify spatial heterogeneity in saline-sodic soils. Thus, a better understanding of the spatial pattern of the measured soil variables in saline sodic soils can easily be captured. It also showed soil series variation for all the measured soil attributes and demonstrates the need to further explore and examine other soil attributes not covered in this study. Integrating high resolution imagery and indices could be an area of future research in saline-sodic soil.