Sentinel Site Data for Model Improvement – Definitions and Characterization.

Quality data are a necessary pre-requisite for use by crop modelers in order to calibrate, test and improve models for ability to respond to climatic change factors of carbon dioxide, temperature, water, as well as management and genetic factors.  The goal of this presentation is to describe procedures for entering and categorizing quality, intensity, and suitability of data for modeling purposes where the data were obtained on crop growth and process responses measured in various studies.  The AgMIP project is establishing data bases for entry, storage, and retrieval of crop, soils, and weather data to be used for running as well as testing of crop models.  The various data sets are being categorized as platinum, gold, or silver, based on the quantity, quality, and suitability.  For example, platinum indicates highest quality observed data for model intercomparisons, evaluations, and improvement, with a full complement of variables including in-season and end-of-season measurements, along with observed data on soil, weather, and management conditions for use as model inputs.  In-season data could be biomass accumulation, soil water measurement, soil and plant nutrient concentrations.  The second-tier ‘gold sentinel’ sites have the minimum information required for model calibration, operation, and evaluation, but may lack the breadth of variables or intensity of data available.  The third-tier ‘silver sentinel sites’ would be characteristic of many of the yield trials that plant breeders manage, where end-of-season yield and maturity dates are collected, along with basic management, such as planting date, irrigation, and nutrient applications.  Weather and soil survey data, for example, may be from nearby but not actual sites.