Compilation of the crop knowledge database
Crop knowledge database
Data about the specific growth potentials and constraints of individual crops are an essential input to any crop growth simulation model. During the first phase of the MARS project, a project was launched to collect all possible data of this type and compile a crop knowledge database. Researchers of different institutes reviewed literature to determine the physiological and cultivation characteristics of the crops in various regions of Europe (Burrill et al., 1995).
The crop knowledge database includes winter wheat, spring wheat, barley, rice, potato, sugar beet, field beans, soybean, oil-seed rape, grain maize and sunflower. It refers to:
- Agronomic data such as the major varieties grown in each region and the earliest and latest dates of sowing and harvest for these varieties, maximum altitude at which a crop is grown, the most common agricultural practices.
- Region specific physiological information such as crop cycle length and relations between phenology and temperature and day length, initial dry matter after emergence (and, indirectly, plant density).
- Non region-specific, detailed physiological information such as energy conversion, partitioning of assimilates over various plant organs, etc. This information was derived from literature. For wheat, information was also derived from field trials executed in Belgium, the United Kingdom and the Netherlands. For other countries and other crops, no detailed field observations were available and consequently calibration of the crop characteristics could not be executed (Boons-Prins et al., 1993).
- Details of optimal, acceptable and harmful conditions for each phenological stage, the event most likely to happen such as the occurrence of a disease, and the impact on the potential production.
The results of the project are presented by Russell and Wilson (1994), Falisse (1992), Narciso et al. (1992), Bignon (1990), Hough (1990) and Russell (1990). Boons-Prins et al. (1993) used these results and constructed parameters to be used as input in the crop simulations, including also information from van Heemst (1988). Later, a ‘Computerised Crop Knowledge base System’ was built for storing, retrieving and extracting relevant information about crops (Russell et al., 1997).
An AgroPheno database contains data from different sources: Boons-Prins et al. (1993), GISAT (2003), Willekens(1998), the ASEMARS project and other sources collected by JRC AGRI4CAST. The AgroPheno database contains phenological data from regional, long term average calendars and field experiments and is manly used for calibration and validation purposes.
Each crop is described by a set of parameters, called a 'crop parametrization'. The LINK_CROP_PARAMETRIZATION object contain a list of the crop parametrizations and additional general attributes such as the crop name, crop model, landcover to which the crop is linked and operability of the crop. Three types of parameters are distinguished: parameters that describe the growth behaviour and parameters, parameters that describe the spatial- and temporal variation of the crop calendar and the spatial variation of crop varieties.
Growth behaviour parameters
These parameters describe the growth behaviour of the crop. The object GLOBAL_CROP_PARAMETERS describes the parameters. The crop specific values of the parameters are copied from the region- and non-region specific physiological information of the crop knowledge database and stored in object LINK_CROP_PARAMETERS.
|parameter types and descriptions|
|A list of all crop parameters is stored in object GLOBAL_CROP_PARAMETERS and has the following attributes:
Parameters are either expressed as a single value (e.g. TSUM1, the temperature sum from emergence to anthesis) or as a function of another variable (e.g. SLATB, the specific leaf area as a function of DVS). The later are expressed as a set of value pairs (x,y) that describe the shape of the function. A special interpolation function (AFGEN) that is part of the crop models perform an interpolation to obtain the function values for arbitrary inputs. The naming convention for the multiple parameters is the name of the parameter, postfixed by ‘_ XX’ . The values of XX range from 01 to 10 or 15, depending on the parameter. The rules for interpolation by the AFGEN function are as follows: If the argument (x-value) is less than the first x-value in the descriptive array return the first y-value. If the argument is between two x-values, return the linear interpolation of the corresponding y-values. If the argument is larger than the last x-value, return the last y-value.
Parameters are also classified as 'crop' parameter or 'model' parameter. Most parameters belong to the first category. This subset is available in object (LINK_CROP_PARAMETERS_CROP). Parameters that are used to configure the crop models are classified as model parameters. This subset is available in object (LINK_CROP_PARAMETERS_MODEL).
The parameter values are stored in the CROP_CALENDARS object. The table stores the crops, grids and seasons that should be simulated. The grids refer to climatic grid cells (see Interpolation of observed weather). Seasons are described in terms of start type (sowing or emergence), end type (maturity or harvest) and the dates on which these events take place. Each calendar has an identifier for the season which is linked to the year of harvest of the particular calendar. When a season fall inside a single calendar year, the identifier is equal to the year (e.g. 1984). When the season includes a year boundary, the identifier specifies both years (e.g. 19831984).
|Types of events during a crop season|
|Seasons are described in terms of events. Possible events are stored in object GLOBAL_EVENTS. The main events that are required for a simulation are the start- en end of a season.
The start type of the simulation is defined by START_EVENT which can take the following values:
The end type of the simulation is determined by END_EVENT and can take the following values:
The date of sowing is fixed and given by the column SOWING. Emergence is calculated using the effective daily temperature (defined by the crop parameters TBASEM, TEFFMX and TSUMEM).
The date of emergence is fixed and given by the column EMERGENCE.
The sowing date is determined by the program. The model starts evaluating sowing conditions 10 days before the earliest sowing date (given by the column SOWING) and will return the earliest sowing date between SOWING and SOWING+WINDOW_DURATION. Emergence is calculated using the effective daily temperature (defined by the crop parameters TBASEM, TEFFMX and TSUMEM).
The simulation runs until maturity is reached. If by this time the date in column MATURITY is not yet reached, the soil water balance continues with a bare soil until the date is reached and crop indicator results are copied for all remaining days until the date is reached. The simulation stops at the date given by column MATURITY and all state variable are reset, even if maturity was not yet reached. Effectively, the behaviour is the same as event HARVEST.
The simulation stops at the date given by column HARVEST and all state variable are reset. If simulated maturity is reached before this date, the crop model stops at simulated maturity and crop indicator results are copied for all remaining days until the HARVEST date is reached. The soil water balance continues after simulated maturity with a bare soil until HARVEST date is reached. Effectively, the behaviour is the same as event MATURITY.
A single crop (i.e. crop parametrization) can extent over large areas. For all grids in this extent the default crop parametrization is the same. The full extent of the crop is usually subdivided into zones, each zone representing a different variety of the crop. The zonation is stored in object CROP_SPATIALIZATIONS. It stores for each grid and crop the variety that should be used to simulate the crop. A variety is described by the default crop parameters from objects LINK_CROP_PARAMETERS_CROP and LINK_CROP_PARAMETERS_MODEL and a limited number of parameters that optimized to the local conditions of the grids of this variety (e.g. TSUMEM, TSUM1 and/or TSUM2).
The locally optimized parameters are stored in the object LINK_CROP_VARIETY_PARAMETERS. If a parameter exist in this object it overwrites the default value of the crop parametrization. Each grid can only have a single variety since the results are stored under the crop number, not the variety number. If more than one varieties need to be simulated (e.g. differently optimized), these should be introduced as additional crop having its own varieties. Values for varieties are derived during the calibration of a crop. A complete list of unique varies is stored in object CROP_VARIETIES.
Crop aggregation areas
In order to forecast yields at administrative levels, simulated crops (identified as IDCROP_PARAMETRIZATION) are aggregated to administrative regions. They are then identified as IDCROP_AGGREGATION.
The CROP_AGGREGATIONS and LINK_CROP_AGGREGATION objects contain a list of available aggregated crops (identified by IDCROP_PARAMETRIZATION) and which simulated crop (identified by IDCROP_PARAMETRIZATION) is used to calculated the aggregate crop. A single simulated crop can be aggregated to different aggregated crops using alternative aggregation areas. The aggregation areas are stored in object STAT_REGION_AREAS.
The following aggregated crop are available over the different regions of interest:
|Region of interest (ROI)||simulated crop (IDCROP_PARAMETRIZATION)||aggregated crop (IDCROP_AGGREGATION)|
|EUR||1||Winter wheat||1||Winter wheat|
|2||Grain maize||2||Grain maize|
|3||Spring barley||3||Spring barley|
|6||Sugar beets||6||Sugar beets|
|8||Field beans||8||Field beans|
|10||Winter rapeseed||10||Winter rapeseed|
|46||Rapeseed and Turnips|
|CHN||1||Winter wheat||1||Winter wheat|
|2||Grain maize||2||Grain maize|
|3||Rice first season||3||Rice first season|
|4||Rice second season||4||Rice second season|
|5||Rice late season||5||Rice late season|
|RUK||1||Winter wheat 2008||1||Winter wheat 2008|
|3||Winter barley 2008|
|2||Winter wheat 2012||2||Winter wheat 2012|
|4||Winter barley 2012|
|3||Spring wheat 2008||5||Spring wheat 2008|
|7||Spring barley 2008|
|4||Spring wheat 2012||6||Spring wheat 2012|
|8||Spring barley 2012|
|5||Grain maize 2008||9||Grain maize 2008|
Crop parameter related information is often available at a limited scale of extent. For example, the sowing dates for a region are often known only for a small sample of fields. Information how representative these samples are, is usually unavailable while this information is needed for accurate scaling up from the site-specific information to a level.
Similarly, values of crop modelling parameters obtained from individual trials will differ from those that would have been obtained if a complete enumeration had been achieved. This is of particular importance wherever the relations are not linear (Vossen and Rijks, 1995).Another problem is that, although the inventories were carefully compiled, information frequently is not available for certain parameters and in certain regions or countries. In such cases a ‘best guess’ was made.