Location: Agroecosystems Management Research2013 Annual Report
1a. Objectives (from AD-416):
1. Explore the use of data mining and analysis techniques to identify important Dynamic Soil Properties (DSPs) and influencing factors, such as climate, soil, and management practice; 2. Enhance existing models by collaborating with modelers to add subroutines for DSPs and/or develop new pedotransfer functions (PTFs); and 3. Define the most efficient methodology for combining modeling and sampling for rapid DSP database development and model validation.
1b. Approach (from AD-416):
1. Data mining and analysis. Both manual and computerized data mining of the existing literature will be performed to identify appropriate minimum data set components and trends for agricultural systems, by region and management. Where data is limited or variance unobtainable, un-weighted meta-analysis will be performed and the sensitivity of the analysis will be assessed with Monte Carlo simulations. Finally, meta-analysis will provide overall trends in DSPs, by soil, climate, management system, or other factors, as deemed appropriate. 2. Development of PTFs or model subroutines. Using statistical models and in collaboration with modelers, new PTFs and/or model subroutines will be developed to estimate DSPs based on other soil properties in combination with management and climate (or other unforeseen factors.) 3. Recommendation of sampling intensity. Using information on variance obtained from both data mining and model sensitivity analysis, recommendations for sampling design and intensity will be made. The anticipated method will be landscape transects with varying sampling frequency, depending on property variability. Analyzed samples will be used to populate the DSP database, validate models, and provide references for agroecological sites.
3. Progress Report:
During FY 2014, we will use data mining and analysis techniques data of the existing literature to identify appropriate minimum data set components and trends for agricultural systems, by region and management. Where data is limited or variance unobtainable, un-weighted meta-analysis will be performed and the sensitivity of the analysis will be assessed with Monte Carlo simulations.