CMCC
DGCSMFDynamic Gridded Crop Simulation Modelling Framework
Engine checking…
Model DSSAT
Runs on …
Coverage Global

Input sets

DSSAT input files made with Generate FileX in 01 Generate. Run DSSAT on a set here: the stored files run as they were generated.
Input setCreatedTypeCropRegionGridSeasonsCellsStatus
No input sets yet. In 01 Generate, set up the inputs and press Generate FileX.

Simulation runs

RunStartedCropRegionGridSeasonsCellsStatusMean yield
No runs yet. Generate an input set in 01 Generate, then run DSSAT on it above.
Model check · DSSAT reference experiment (maize, Gainesville 1982, with observed yields)

Crop

Experiment

Nitrogen × irrigation trial used to test DSSAT releases. Observed yields come with the experiment, so you can see the model against field data.

The run happens on the server: DSSAT-CSM 4.8 compiled from the official source, the same engine used for HPC job arrays.

Simulated vs observed grain yield real DSSAT output

#TreatmentSimulated kg/haObserved kg/haBiomass kg/haIrrigation mmN uptake kg/ha
Press “Run DSSAT” to simulate.
Runs–DSSAT seasons
Model time–on the server
Mean abs. error–vs observed yield
Agreement (d)–Willmott index
$ waiting for the engine…

1 · Region

2 · Climate, soil and period

3 · Simulation type and crop

Cultivar, crop model, sowing, fertilizer, irrigation, tillage, amendments, simulation options, automatic management and soil are set per grid cell, in the panel under the map.

4 · Management from literature

Click a cell to select it; hold Shift and drag to select many, or pick a region above.

Cell inputs

The crops grown one after the other on the same field, for one cycle of the rotation. DSSAT runs them in sequence mode: soil water, mineral nitrogen, organic matter and residues of one crop are the starting conditions of the next, so the effect of the previous crop is simulated. The cycle repeats over the run's seasons (e.g. a 2-year cycle from 2005: years 1, 2 = 2005, 2006, then 2007, 2008 …). A winter crop sown in autumn of year 1 is harvested in year 2. Each crop's fertilizer is split ⅓ at sowing, ⅔ after 30 days; simulation options come from the Simulation options tab.

Mineral fertilizer applications, in days after sowing. Materials and methods are DSSAT's own codes. Without a schedule, the cell's total N is split ⅓ at sowing and ⅔ after 30 days (urea).

Irrigation events in days after sowing.

Tillage operations in days from sowing (negative = before sowing).

Manure, compost, residues and other organic amendments, in days from sowing (negative = before sowing).

DSSAT simulation methods for the chosen cells. Changing the crop model changes the cultivar list; cells whose cultivar the new model does not have take its default cultivar.

With automatic sowing, DSSAT sows on the first day inside the window when soil water and temperature are within the limits. Automatic irrigation (refill or fixed amount) is in the Irrigation tab. DSSAT 4.8 has automatic fertilization and automatic residue application switched off, so fertilizer and organic amendments are scheduled.

Writes every cell's experiment file (FileX, or .SQX for rotations), daily weather (.WTH), soil profile (.SOL) and calibrated cultivars (.CUL). You can look at and download the files, then run DSSAT on them in 03 Simulate.

1 · Cells to calibrate

The grid, inputs and selection from 01 Generate. Calibrate on the selected cells, or on all cells when none is selected.

4 · Run

2 · What to adjust

Each ticked coefficient gets evenly spaced test values between Min and Max. Defaults are ±20% around the cultivar's own value, inside DSSAT's allowed range.

UseCoefficientWhat it controlsCultivarDSSAT rangeMinMaxLevelsTest values

3 · Observations and fitting

Score of a combination = 0.4·d + 0.2·r + 0.4·(1 − nRMSE), on the calibration years; d = Willmott agreement, r = correlation, nRMSE = RMSE / mean observed. The best score wins; validation years are only reported.

Observed-yield datasets: source, scale, years, moisture, how they are used

How calibration works
  1. Prepare the grid in 01 Generate: region, weather, soils (SoilGrids per cell), management per cell (from literature or entered).
  2. Choose what to adjust: cultivar coefficients (each controls one part of crop development or growth) and, if management is uncertain, sowing-date shifts and nitrogen factors. Every combination of test values is one candidate.
  3. Run: every candidate is simulated for every selected cell and season (DSSAT batches split over many cloud workers; on the CMCC cluster the same batches run as job arrays).
  4. Score each candidate against observed yields, per unit (each observed region, each climate zone, or the whole grid), on the calibration years.
  5. Check the best candidate on the validation years it was not fitted to.
  6. Refine (round 2): narrower ranges around the best values.
  7. Apply the best values to the grid in 01 Generate (cells outside the observed units take the nearest unit in a similar climate), then run the simulation.

Outcome maps

Any DSSAT season outcome of a run, as a map, with statistics and downloads.

Distribution over grid cells

Economic gain or loss from simulated productivity: the yield difference between a scenario run and a baseline run, times the crop's harvested area (SPAM 2020) and the farm-gate price. Optional input costs give the net change, so adaptation measures can be judged by profit as well as yield.

1 · Runs to compare

Both runs must cover the same cells; seasons are compared year by year where both have them, otherwise as averages. With no baseline, the value of production of the scenario run is shown.

2 · Prices and costs

Input costs (for the net change)

Defaults are indicative (N ≈ EUR 1–1.5/kg; 1 mm on 1 ha = 10 m³ of water, so EUR 1/mm/ha = EUR 0.10/m³ including pumping). Set your own.

Registrations

People who registered on the portal. Approve to let them run simulations; Delete removes the request and its details.

RegisteredNameEmailRoleOrganisationCountrySimulation regionStatus