[[TOC]]

So far, removal of grass biomass on managed grassland occurred globally with only one rule: reducing the leaf biomass when a certain threshold is reached since the last harvest event. Here, new management options on managed grassland representing grazing and mowing regimes are described.

In order to enable a more detailed representation of biomass removal on grasslands, the harvest functions were extended. Firstly, the concept is outlined before giving details on the implemented management options (see also the presentation in the LPJmL_seminar). Additionally, optional links to other applications and models are highlighted.
The harvest options are active when the flag NEW_GRASS is used; otherwise the current trunk version is used. The options are connected with daily allocation of grass biomass.

Options are defined per grid cell and apply to all grasses grown on grassland (C3 and C4, rainfed and irrigated). When regional values are to be met, options have to be distributed to the grid cells of the region.
In general, management options can be used by the following procedure:

• implementing the flag #define WITH_GRASS_SCENARIOS in lpjml.conf
• providing the binary file scenarios.bin in the root directory which gives the chosen option for each grid cell
• providing the information on the scenario file in the input.conf
  #ifdef WITH_GRASS_SCENARIOS
  /home/rolinski/LPJ/Newinput/scenarios.bin
  #endif
• checking the parameters chosen in the header file include/grassland.h
• looking for the output files …

When the flag WITH_GRASS_SCENARIOS is not used, the default case is applied globally. Since the information on the scenario needs only its statement in the scenario.bin file, new options can be easily added. What they need, is the specification of their harvest regime in landuse/daily_grassland.c and landuse/harvest_stand.c

1. Default = GS_DEFAULT = 0
   This option is the current harvest function. When phenology is complete and the biomass increment since the last harvest event exceed 100 g C m^-2, 50% of the leaf biomass is stored in the harvest structure. Growing degree days and phenology are set back to zero.
   Results differ slightly because of a change in the allocation frequency from just before the harvest event to daily.
2. Mowing = GS_MOWING = 1
   Defined harvest events twice a year. The harvest function harvest_grass_mowing is nearly the same as for the default but leaf biomass is reduced to a stubble height of 5 cm (= 25 g C m^-2). At the moment, the events take place at the first of June and December. In the long run, these dates should be adapted to climatic factors such as farmers use. The seasonality, which is computed with the new sowing dates, could give the necessary information. Growing degree days and phenology are set back to zero.
   Parameters to be chosen are
   const int mowingDays[] = {152, 335}; // mowing on fixed dates 1-june or 1-dec
   #define STUBBLE_HEIGHT_MOWING 25; // [g C m^-2] equals 5 cm. The leaf biomass that remains on the paddock after mowing.
3. Extensive grazing = GS_GRAZING_EXT = 2
   This option should reflect low density grazing all year round.
   The demand per cow times the stocking intensity define the biomass removed per day. This amount is removed from the leaf biomass of each grass pft proportionally as long as it is above the threshold for the stubble height. The manure fraction of the removed biomass goes to the fast soil pool and the rest to the harvest assuming this to be divided between atmosphere (60) and the cattle (15). Growing degree days are reduced proportional to the removed biomass.
   Parameters to be chosen are
   #define MANURE 0.25; // [-] 25 % of grazed biomass. The portion going to the fast soil litter pool comes from some literature estimates.
   #define DEMAND_COW_EXT 4000.0; // [g C day^-1] Now the same for extensive and intensive grazing; corresponds to 8.9 kg DW which is in between the observed values of 6 and 12.
   #define STOCKING_DENSITY_EXT 0.5; // [LSU ha^-1] (LiveStockUnit). Stocking densities for both grazing systems are rough estimates. These values have to be tested when doing global runs.
   #define STUBBLE_HEIGHT_GRAZING_EXT 5; // [g C m^-2] Minimal threshold to start extensive grazing
   #define GRAZING_STUBLE 5; // [g C m^-2] Grazing stops when less than this threshold is left in the leaves
4. Intensive grazing = GS_GRAZING_INT = 3
   Here, a lot of cows graze for a short period until there is only some centimeter of grass left and then move to the next paddock (Kurzumtriebsweide). The site itself has then a recovery phase until the flock comes again.
   When the harvest routine is called for the first time (rotation mode = RM_UNDEFINED), the rotation length is determined as (leaf biomass
   • stubble height)/(stocking intensity * demand per cow). This duration is restricted to the maximum value. In case, it is below the minimum value, the harvest is skipped until enough grass biomass is grown. The number of grazing days is rotation length devided by maximum number of paddocks. The actual number of paddocks is rotation length devided by grazing days and the recovery days are grazing days times (number of paddocks - 1). Now, the rotation mode is set to RM_GRAZING.
     Grazing biomass demand is stocking intensity times demand per cow times number of paddocks which is assigned to each grass pft proportional to their contribution to total leaf biomass. If (leaf - grazed biomass) < stubble, the grazed biomass is reduced to (leaf biomass - stubble) leaving cows hungry. Growing degree days are reduced proportionally. Leaf biomass is reduced by grazed biomass which is attributed to the fast soil pool (manure fraction) and the harvest. Number of grazing days are reduced and when they reach zero, the rotation mode is set to RM_RECOVERY.
     In the recovery phase, the recovery days are counted down and when they reach zero, the rotation mode is again RM_UNDEFINED so that the cycle starts again with the actual leaf biomass. Thus, rotation length and all other variables determined are dynamically adjusted to the biomass during the growing period.
     Parameters to be chosen are
     #define MAX_PADDOCKS 16; // Rough estimate from Faustzahlen.
     #define MAX_ROTATION_LENGTH 50; // [days] Rough estimate when usually rotations take 20 to 35 days
     #define MIN_ROTATION_LENGTH 5; // [days] Rough estimate
     #define DEMAND_COW_INT 4000.0; // [g C day^-1] Now the same for extensive and intensive grazing; corresponds to 8.9 kg DW which is in between the observed values of 6 and 12.
     #define STOCKING_DENSITY_INT 1.2; // [LSU ha^-1] (LiveStockUnit).
     #define STUBBLE_HEIGHT_GRAZING_INT 40; // [g C m^-2] equals 7-8 cm. Threshold to start intensive grazing

The implementation of the influence of grazing intensity on grass biomass removal allows for several connections to other models. Two of them can be mentioned here:

• IMAGE
  As vegetation model in the assessment framework of IMAGE, the livestock density and its grass demand could be taken into account dynamically. Depending on the intensity, the management option of extensive or intensive grazing can be chosen and the biomass removal could be realized. At the moment, the livestock density is determined in IMAGE according to the potential harvest of grass for the default option.
• MAgPIE
  With the consideration of detailed feed baskets for certain livestock products from different production systems in MAgPIE by Isabelle, the potential demand and their supply of grass can be synchronized. Comparisons should be possible soon. Further coupling has to be considered carefully.

The options are included in the trunk version. For the implementation of the options, the grassland stand was used and changed to daily allocation.

What you need are

• information on the spatial distribution of the options in the scenario.bin file
• an idea on the parameters to choose for the options

Further ideas are

• use the Bayesian calibration in order to improve the parametrizations
• maybe use other information on livestock densities and give grid cells based parameters per option in the scenario.bin file as additional bands
• use the new opportunity of fire on managed grassland with spitfire on grassland

Susanne Rolinski, Johnny te Roller, Christoph Müller, Jens Heinke

Wiki, Bayesian_uncertainty_analysis, LPJmL_seminar