Space-time upscaling of plot-based research information: frost tillage

Abstract

Most agronomic-environmental research information is based on plot-scale studies conducted for a limited time period, typically two to five years. Generalization is required for larger spatial domains, typically a farm, watershed, or region, and more representative climate periods, at least thirty years. Frost tillage research was conducted at the plot scale at one primary and two ancillary sites. Results indicated that this tillage method is viable and the processes leading to frost-tillable soil conditions are minimally affected by variation of soil properties. The estimation of seasonal probabilities for frost-tillage conditions was determined to be an assessment need for the adoption of the practice. A soil freezing model using basic climate information on minimum and maximum air temperature and snow depth from a dense network was developed and calibrated for sod and bare soil surfaces based on measured soil temperature data from 8 weather station-years. It was independently validated based on data from frost depth tubes. Observed and predicted frost-tillable days for cropped fields were compared and showed good agreement when averaged for bare and sod surface conditions. Soil freezing was simulated for 275 observation sites in the Northeastern USA based on 40-year climate data. Frost tillable days were determined and mapped as the annual number of days in which it can be performed at various recurrence periods. Upscaling methodology used in this study is discussed, especially as it relates to the identification of relevant processes and their stochastic nature.