Crop Growth Monitoring System (CGMS)

Introduction

The Crop Growth Monitoring System is the core of the MARS Crop Yield Forecast System (MCYFS) currently used in forecasting activities in Europe by AGRI4CAST action. The role of CGMS is providing reliable and timely spatial information about crop status in Europe, which would be used in different statistical procedures to produce a yield prevision.
CGMS is constituted of three components (Figure 1):

  • Weather monitoring, dedicated to the gathering and processing of meteorological data. It produces meteorological indicators for early alert warning and crop modelling.
  • Crop growth modelling, with the purpose of producing crop status indicators ingesting weather data and static data (crop parameters, soil information, management practices) in crop modelling solutions.
  • Statistical methods, for the evaluation of results, interpreting the relationships between crop indicators and crop yield and assisting in yield forecasting.

Weather Monitoring

Meteorological variables such as daily average, minimum and maximum temperatures, rainfall, etc. may help in the understanding of crop development dynamics and yield along the season. Weather data comes from different sources: (1) direct observations from meteorological stations, (2) meteorological products resulting from weather modelling –ECMWF– or (3) remote sensing observations from meteorological platforms.

Weather data is then processed to generate spatial layers (maps) of all the products comprised in the dataset. The processing and storage of meteorological is as close as possible to the acquisition time. The number of observations during the crop season and time delay between these observations and the product availability date are the main variables determining the ability of the system to produce up to date crop estimations. For that reason the sources for meteorological data fulfil two main requirements:

  • Availability near real time (NRT). The observations should be available (Level 0 information) maximum 1 day after the acquisition time.
  • Short processing time. The processing chain from observations to value-added products should not exceed 2 days, thus enabling to calculate output variables from crop modelling in a time delay of 3 days after the observations.

Crop Modelling

A crop model is a group of algorithms that simulates the functioning of a given crop. Those groups of algorithms mimic the main physiological plant processes –such as light interception, respiration, carbon assimilation, grain production– through a set of assumptions and calibration parameters. Basically, the input datasets of crop models integrate meteorological data (temperature, rainfall, solar radiation, etc.), soil information (soil water capacity, soil depth) and management practices (e.g. irrigation). The outputs are usually indicators of crop development such as the biomass produced, the leaf area produced, the biomass allocated in the storage organs (grain in the case of cereals), etc.

In forecasting activities at national or regional scales crop models play a major role. They provide the basis to assess the influence of weather on crop yield. Rather than producing a specific value describing the actual harvested crop yield, they calculate a set of indicators describing the inter-annual variability of crop biophysical parameters that can be statistically related to official yield figures to produce forecasts.

The operational models currently are integrated in the BioMA (Biophysical Models Application) platform, a new infrastructure for crop modelling already developed within the MARS Unit integrating the existing models (WOFOST, for cereals and tubers; WARM for rice; LINGRA for pastures) in a more efficient environment.

Statistical Methods

The crop yield forecast procedure assesses yield forecasts in ton.ha-1 fresh weight using different statistical methods and software tools. Two different approaches are developed with the aim to predict the yield.

(1) The first one consists in a classic regression approach, in which the focus is on the relationship between a dependent variable - the yield - and one or more independent parameters related to climate/ weather effects,

(2) while the second is based on analogies between the contingent conditions and the past, investigating years that behave similarly with respect to selected events and reporting their measured effects on the actual state in order to predict final consequences.

The yield "predictors" consist in the products previously generated by crop modelling solutions: meteorological impact evaluation (minimum or maximum temperature, rain, radiation level, etc.), crop status assessment (e.g. soil moisture, development stage) and crop growth expectations (e.g. potential yield biomass, potential yield storage).

Further Information
Within CGMS the following crop models are currently implemented: WOFOST for cereals and tubers, the WARM model for rice, and the LINGRA model for pasture.

  • CGMS version 9.2: user manual and technical documentation of the Crop Growth Monitoring System
  • MCYFS: Mars Crop Yield Forecasting System.
  • WOFOST: World Food Studies is a crop growth model for crop yield forecasts
  • LINGRA: Grassland Growth Model to predict growth and development of perennial rye grass across the EU member states at the level of potential production and water-limited production.
  • WARM: Water Accounting Rice Model for paddy rice simulations.
  • BioMA: Biophysical Models Application is a software framework designed and developed for analyzing, parameterizing and running modelling solutions based on biophysical mode.