Location: Southeast Watershed Research
Project Number: 6048-13000-028-061-S
Project Type: Non-Assistance Cooperative Agreement
Start Date: Apr 1, 2024
End Date: Mar 3, 2028
Objective:
To provide ARS the use of lands for the purpose of conducting soil, crop, and biological sample collection.
Approach:
This research applies to objectives 2 and 3 of the project "Shifting the Balance of Water Resources and Interacting Agroecosystem Services toward Sustainable Outcomes in Watersheds of the Southern Coastal Plain." In particular, Sub-Objective 3.A relates to this work: "Characterize field level spatial and temporal variability of biophysical parameters on three farms within the LREW. "
Fine-grained (90m grid sampling) data describing within field variations of soil and water conditions, and crop responses to management practices in the Cooperator's land will increase knowledge about how measurement scale affects uncertainty when “scaling up” outcomes using plot level data. Planned observations include vegetation, terrain, soils, biomass and associated crop yield, and precipitation (or irrigation). At each location on the grid, we will collect samples of soils (shallow and deep) and above ground biomass, following protocols described in detail in the Unit's NP211 project plan. Soil and crop analyses will be completed for each point on the grid. Soil cores are collected at grid locations providing sequential data for spatial effects comparisons at each collection date.
The sample grid spacing was designed to complement satellite data collection efforts from moderate resolution optical satellites, such as Landsat (https://landsat.gsfc.nasa.gov/), to leverage the substantial data archives of freely available imagery for the region. Collection of multispectral data and soil characteristics via UAS and proximal sensors, respectively, will follow protocols described in detail in the Unit's NP211 project plan.
Summary statistics will be calculated for all observed variables, including measures of central tendency, variance and standard deviations, coefficients of variation and variance-covariance matrices. Statistical methods will aim to assess the correlations between regulating, and provisioning ecosystem services. Annual correlations between variables will be evaluated using statistical models. Time-series statistical methods will be used to assess for autocorrelation in the serial datasets. Likewise, methods will be used to account for spatial autocorrelation (e.g. simultaneous autoregression, kriging). In combination with 90-m grid data, contiguous management zones will be delineated based on aggregated soil, biomass, and UAS data. Stepwise regression and path analysis will be conducted to determine the level of association between variables. Statistical and machine learning clustering methods will be used to delineate homogenous clusters of individual and dependent parameter content for all depth increments within each field. Spearman R will be used to evaluate correlations between depths. Changes over time will be reported for each homogeneous cluster to assess the relative contribution of each cluster to total pool sizes within the fields. Geostatistical methods will be used to evaluate sampling strategies to optimize collection efforts. An outcome of this analysis will be estimates of the correlation and error which will serve as a validation step for estimating scaling uncertainties introduced into watershed scale analyses.