Pasquale Steduto

Multimodel ensembles of wheat growth: many models are better than one.

Crop models of crop growth are increasingly used to quantify the impact of global changes due to climate or crop management. Therefore, accuracy of simulation results is a major concern. Studies with ensembles of crop models can give valuable information about model accuracy and uncertainty, but such studies are difficult to organize and have only recently begun. We report on the largest ensemble study to date, of 27 wheat models tested in four contrasting locations for their accuracy in simulating multiple crop growth and yield variables. The relative error averaged over models was 24-38% for the different end-of-season variables including grain yield (GY) and grain protein concentration (GPC). There was little relation between error of a model for GY or GPC and error for in-season variables. Thus, most models did not arrive at accurate simulations of GY and GPC by accurately simulating preceding growth dynamics. Ensemble simulations, taking either the mean (e-mean) or median (e-median) of simulated values, gave better estimates than any individual model when all variables were considered. Compared to individual models, e-median ranked first in simulating measured GY and third in GPC. The error of e-mean and e-median declined with an increasing number of ensemble members, with little decrease beyond 10 models. We conclude that multimodel ensembles can be used to create new estimators with improved accuracy and consistency in simulating growth dynamics. We argue that these results are applicable to other crop species, and hypothesize that they apply more generally to ecological system models.

Concepts and Applications of AquaCrop: The FAO Crop Water Productivity Model

Predicting attainable yield under water-limiting conditions is an important goal in arid, semi-arid and drought-prone environments. To address this task, FAO has developed a model, AquaCrop, which simulates attainable yields of the major herbaceous crops in response to water. Compared to other models, AquaCrop has a significantly smaller number of parameters and attempts to strike a balance between simplicity, accuracy, and robustness. Root zone water content is simulated by keeping track of incoming and outgoing water fluxes. Instead of leaf area index, AquaCrop uses canopy ground cover. Canopy expansion, stomatal conductance, canopy senescence, and harvest index are the key physiological processes which respond to water stress. Low and high temperature stresses on pollination and harvestable yield are considered, as is cold temperature stress on biomass production. Evapotranspiration is simulated separately as crop transpiration and soil evaporation and the daily transpiration is used to calculate the biomass gain via the normalized biomass water productivity. The normalization is for atmospheric evaporative demand and carbon dioxide concentration, to make the model applicable to diverse locations and seasons, including future climate scenarios. AquaCrop accommodates fertility levels and water management systems, including rainfed, supplemental, deficit, and full irrigation. Simulations are routinely in thermal time, but can be carried out in calendar time. Future versions will incorporate salt balance and capillary raise. AquaCrop is aimed at users in extension services, consulting firms, governmental agencies, NGOs, farmers associations and irrigation districts, as well as economists and policy analysts in need of crop models for planning and assessing water needs and use of projects and regions.

Sentinel-1 for wheat mapping and soil moisture retrieval

The main objective of this study is to assess the use of Sentinel-1 (S-1) data for surface soil moisture (SSM) retrieval and wheat mapping (WM) at high spatial resolution (e.g. 100-500m), which constitute valuable information for improving crop yield forecast at large scale. A knowledge based classification method and a SSM retrieval algorithm, developed in view of the European Space Agency Sentinel-1 mission, have been applied to a time series of S-1A data collected from October 2014 to April 2015 over a well-documented agricultural site in southern Italy. In particular, observations of SSM content recorded by a network of ground stations deployed in an experimental farm have been used to test the accuracy of the retrieved SSM values. First results indicate an rms error between 5% and 6%. However, the range of observed SSM values is still quite limited and, therefore, longer time series are needed to investigate the retrieval performance over the full range of SSM values.