DSSAT for agrivoltaics: complete workflow

DSSAT can simulate crop growth, soil water, nitrogen, phenology and yield, but it does not calculate PV shading geometry or electricity production. A complete agrivoltaic study therefore couples three components:

PV/shadow model → modified crop-level weather → DSSAT crop model → combined food–energy analysis

A recent implementation coupled Honeybee–Radiance for irradiance, DSSAT-CERES-Rice for crop yield and PVlib for electrical output. It divided the field into irradiance zones, ran DSSAT separately for each zone and area-weighted the resulting yields. (MDPI)

As of July 2026, the current official DSSAT release is DSSAT 4.8.5, released December 1, 2024. (DSSAT.net)

---

1. Understand what DSSAT needs

The minimum inputs are:

Input group	Required information
Weather	Daily solar radiation, maximum temperature, minimum temperature and rainfall
Weather-station metadata	Latitude, longitude and elevation
Soil	Soil layers, texture, bulk density, organic carbon, pH, drained upper limit, lower limit and saturation
Crop management	Crop, cultivar, planting date, density, row spacing, sowing depth
Water management	Irrigation dates, amounts and method
Nutrient management	Fertilizer dates, forms, placement and amounts
Initial conditions	Initial soil water, nitrate, ammonium, residue and previous crop
Calibration observations	Flowering, maturity, LAI, biomass, yield and yield components

Dew-point temperature, humidity and wind speed are optional but improve evapotranspiration calculations. (DSSAT.net)

For agrivoltaics, also collect:

• Solar radiation or PAR at crop-canopy height.
• Air temperature and humidity below and outside the panels.
• Soil temperature and soil moisture.
• Rainfall, throughfall and concentrated panel-edge runoff.
• Wind speed beneath and outside the structure.
• PV front and rear irradiance.
• Module temperature and AC/DC electricity production.
• Spatial position relative to the PV rows.

---

2. First learn DSSAT using an existing experiment

Do not begin by creating an agrivoltaic experiment from nothing.

1. Install DSSAT 4.8.5 on Windows.
2. Open the DSSAT Shell.
3. Select a crop, such as maize, rice, wheat or soybean.
4. Select one of the included experiments.
5. Click Run.
6. Open the Analysis tab.
7. Examine:
   • Summary.OUT
   • PlantGro.OUT
   • SoilWat.OUT
   • SoilNi.OUT
   • Evaluation.OUT
8. Plot simulated and observed yield, biomass and phenology.

PlantGro.OUT contains daily crop variables such as leaf, stem, grain and root biomass, LAI, root density and stress indices. DSSAT can plot simulated versus observed measurements and calculate evaluation statistics. (DSSAT.net)

Repeat this process until you understand how treatments, weather stations, soils and cultivars are connected.

---

3. DSSAT tools you will use

DSSAT tool	Purpose
WeatherMan	Import, inspect, quality-control and export daily weather
SBuild	Create and edit soil profiles
XBuild	Build the experimental FileX, treatments and management
ATCreate	Enter measured crop and soil observations
GenSelect / GENCALC	Estimate cultivar coefficients
GLUESelect	Estimate cultivar coefficients and uncertainty
DSSATSens	Change input values systematically for sensitivity analysis
GBuild / EasyGrapher	Plot results and calculate model statistics
Seasonal analysis / Varan	Run management scenarios across many weather years
Sequence analysis / Sustain	Run rotations and long-term soil simulations

XBuild creates the main experiment file. ATCreate creates summary observations in FileA and time-series measurements in FileT. DSSATSens can vary cultivar parameters, planting details, soil properties and weather inputs. (DSSAT.net)

---

4. Design the agrivoltaic field experiment

Recommended experimental structure

PV structures are difficult to randomize, so a split-plot or blocked design is normally more practical than a completely randomized design.

Main PV treatments

For example:

1. Open-field control.
2. Fixed-tilt PV.
3. Single-axis tracking PV.
4. Shade-aware or anti-tracking PV.

Spatial subplots

Within each PV treatment, measure distinct positions:

• Directly below the panel.
• Panel drip edge.
• Near the PV support column.
• Middle of the crop alley.
• East and west sides of the row.
• An open-field reference position.

Additional treatment factors

Depending on your research question:

• Rainfed versus irrigated.
• Multiple cultivars.
• Nitrogen rates.
• Planting dates.
• Crop row orientation.
• PV row spacing.
• Panel height.
• Ground coverage ratio, or GCR.

Use at least several genuine spatial replicates when possible and perform a statistical power calculation before fixing the final sample size.

Example treatment arrangement

Main treatment	Position	Water treatment
No PV	Reference plot	Rainfed
No PV	Reference plot	Irrigated
Tracking PV	Below panel	Rainfed
Tracking PV	Drip edge	Rainfed
Tracking PV	Alley center	Rainfed
Tracking PV	Below panel	Irrigated
Tracking PV	Drip edge	Irrigated
Tracking PV	Alley center	Irrigated

Keep crop cultivar, planting density, fertilizer and planting date identical between the PV and open-field treatments unless these are explicit experimental factors.

---

5. Collect the field data correctly

Weather and microclimate

Record at least:

• Global horizontal solar radiation above the PV system.
• Crop-level radiation or PAR in every representative position.
• Maximum and minimum air temperature.
• Relative humidity or dew-point temperature.
• Rainfall outside the system.
• Throughfall below panels.
• Runoff at panel drip lines.
• Wind speed.
• Soil temperature.

Use sub-hourly measurements—typically 5-, 10-, 15- or 30-minute data—during the experiment. Aggregate them to DSSAT daily values later.

Crop observations

Collect:

• Emergence date.
• Flowering or anthesis date.
• Grain-filling stages.
• Physiological maturity.
• Plant population after establishment.
• Plant height.
• LAI throughout the season.
• Aboveground biomass at several growth stages.
• Leaf, stem and reproductive biomass when possible.
• Final marketable or grain yield.
• Yield components.
• Crop nitrogen concentration.
• Root observations when feasible.

Soil observations

Measure by depth:

• Sand, silt and clay.
• Bulk density.
• Organic carbon.
• pH.
• Saturated water content.
• Drained upper limit or field capacity.
• Lower limit or wilting point.
• Initial soil water.
• Initial nitrate and ammonium.
• Soil moisture through the season.

The measurements required should correspond to the processes being evaluated. For example, a water-management study requires reliable soil-water measurements, while a nitrogen study requires soil and crop nitrogen observations. (DSSAT.net)

---

6. Prepare the standard open-field DSSAT model

Build and validate the ordinary crop model before adding PV effects.

Step 6.1: Weather file

Use WeatherMan to create the open-field weather station.

Required daily variables:

text
DATE   SRAD   TMAX   TMIN   RAIN

Where:

• SRAD = MJ m⁻² day⁻¹
• TMAX and TMIN = °C
• RAIN = mm day⁻¹

Check:

• Missing dates.
• Impossible temperatures.
• Negative radiation.
• Unit errors.
• Rain gauges with accumulated multi-day rainfall.
• Radiation sensors affected by PV shadows.

Step 6.2: Soil file

Use SBuild to create the soil profile. Avoid relying only on texture-based estimates when measured water limits are available.

For each layer, enter:

• Layer depth.
• Lower limit.
• Drained upper limit.
• Saturation.
• Bulk density.
• Organic carbon.
• pH.
• Root-growth factor.
• Initial water and nitrogen conditions.

Step 6.3: Experiment and management

Use XBuild to create the experiment:

1. Select crop model.
2. Select weather station.
3. Select soil profile.
4. Select cultivar.
5. Enter planting date.
6. Enter planting population and row spacing.
7. Enter irrigation.
8. Enter fertilizer.
9. Enter tillage and residue information.
10. Enter simulation controls.
11. Create one treatment for each field treatment.

Step 6.4: Observations

Use ATCreate to enter:

• FileA: flowering, maturity, final yield, yield components and final biomass.
• FileT: sequential LAI, biomass, soil water and nitrogen observations.

---

7. Calibrate DSSAT

Calibration means estimating uncertain model parameters, especially cultivar coefficients. It does not mean changing parameters until every observation matches perfectly.

Correct calibration order

1. Soil water balance

First check:

• Runoff.
• Drainage.
• Evaporation.
• Root-zone water.
• Soil water by depth.

Use measured soil moisture data when available.

2. Phenology

Calibrate cultivar coefficients controlling:

• Emergence-to-flowering duration.
• Photoperiod sensitivity.
• Grain-filling duration.
• Maturity.

Match flowering and maturity before calibrating yield.

3. Growth and yield

Then calibrate:

• Leaf area development.
• Biomass accumulation.
• Grain number.
• Grain weight.
• Harvest index.

4. Nitrogen response

Finally check:

• Crop N uptake.
• Soil nitrate and ammonium.
• Biomass response to fertilizer.
• Grain nitrogen.

Use well-watered, adequately fertilized, disease-free treatments to estimate cultivar coefficients. DSSAT’s calibration tools are designed around preferably non-stressed treatments. (DSSAT.net)

Critical agrivoltaic rule

Do not calibrate cultivar coefficients using the PV-shaded treatment alone.

Calibrate the cultivar using:

• Open-field observations, or
• Independent non-stressed experiments.

Then keep the same cultivar coefficients for the PV treatments. Otherwise, panel effects may be incorrectly hidden inside the genetic coefficients.

---

8. Validate independently

Use observations that were not used for calibration.

A good division is:

• Some years or treatments for calibration.
• Different years, locations or treatments for validation.
• PV treatments reserved primarily for testing the agrivoltaic representation.

Evaluate:

• Flowering date error.
• Maturity date error.
• Yield RMSE.
• Biomass RMSE.
• LAI RMSE.
• Soil-water RMSE.
• Mean bias.
• Willmott agreement index.
• Model efficiency.

DSSAT specifically supports simulated-versus-observed comparison and reports statistics such as RMSE and the agreement index. (DSSAT.net)

Do not rely only on $R^2$. A model can have a high correlation while consistently overpredicting or underpredicting yield.

---

9. Represent PV shading in DSSAT

There are three practical levels.

Level 1: Simple shade experiment

Use this for an initial sensitivity study.

Create treatments such as:

• 0% radiation reduction.
• 10% reduction.
• 20% reduction.
• 30% reduction.
• 40% reduction.

DSSAT includes environmental modification capability that can alter solar radiation to represent solar shades or other artificial environmental changes. (DSSAT.net)

This approach answers:

How sensitive is this crop to a given seasonal radiation reduction?

It does not represent real panel geometry, moving shadows or spatial variation.

---

Level 2: Zone-based agrivoltaic simulation

This is the recommended practical approach.

Step 9.1: Calculate the shadows

Use one of the following:

• Radiance or Honeybee-Radiance.
• Ladybug Tools.
• PVsyst ground irradiance outputs.
• pvfactors.
• bifacial_radiance.
• A validated geometric shadow model.
• Measured crop-level irradiance sensors.

Input the PV geometry:

• Latitude and longitude.
• Panel height.
• Panel length and width.
• Row pitch.
• Ground coverage ratio.
• Tilt.
• Azimuth.
• Tracker rotation schedule.
• Backtracking or anti-tracking settings.
• Bifaciality.
• Ground albedo.
• Crop height.

Calculate hourly or sub-hourly irradiance at crop-canopy height across a spatial grid.

Step 9.2: Define radiation zones

Group locations with similar radiation into a manageable number of zones, for example:

1. Heavy-shade zone.
2. Moderate-shade zone.
3. Light-shade zone.
4. Open-alley zone.

For each zone and day, calculate:

$$F_{z,d} = \frac{\text{Daily irradiance at crop level in PV zone }z} {\text{Daily open-field irradiance}}$$

Then:

$$SRAD_{z,d}=SRAD_{open,d}\times F_{z,d}$$

Example

Suppose open-field radiation is:

$$SRAD_{open}=20\;MJ\,m^{-2}\,day^{-1}$$

Zone factors are:

Zone	Radiation factor	DSSAT SRAD
Below panel	0.55	11 MJ m⁻² day⁻¹
Drip edge	0.75	15 MJ m⁻² day⁻¹
Alley center	0.90	18 MJ m⁻² day⁻¹

Create one modified weather series for each zone.

Step 9.3: Include microclimate changes

For each zone, modify or measure:

• SRAD
• TMAX
• TMIN
• RAIN
• Dew point or humidity, when available
• Wind speed, when available

Using only reduced solar radiation assumes that the panels do not affect temperature, humidity, rain distribution or wind. That may be acceptable for preliminary screening, but not for final design or scientific validation.

Step 9.4: Run DSSAT independently for each zone

Each zone becomes a separate DSSAT environment or run:

text
Open field
Heavy shade
Moderate shade
Light shade
Open alley
Drip edge

The soil and management can initially remain identical, while the weather differs.

If panel runoff causes different water inputs, assign zone-specific rainfall or irrigation. For example:

• Below panel: reduced effective rainfall.
• Drip edge: increased effective water input.
• Alley: normal or intermediate rainfall.

Step 9.5: Area-weight the yield

For $n$ zones:

$$Y_{APV}=\sum_{z=1}^{n}w_zY_z$$

where:

• $Y_z$ = simulated yield in zone $z$
• $w_z$ = proportion of cultivated area in zone $z$
• $\sum w_z=1$

Example:

Zone	Area fraction	DSSAT yield
Below panel	0.25	6.0 t/ha
Drip edge	0.20	7.2 t/ha
Alley	0.55	8.0 t/ha

The whole-field agrivoltaic yield is the area-weighted sum, not the simple average.

The published DSSAT agrivoltaic rice framework used irradiation bins, calculated yield for each bin and area-weighted those results to obtain field-scale yield. (MDPI)

---

Level 3: Dynamic or source-code coupling

DSSAT normally receives daily weather and internally generates hourly patterns. Consequently, two days with the same daily radiation but very different morning and afternoon shade patterns may appear similar to the standard model. (DSSAT.net)

This can matter for tracking PV because crop response may depend on:

• Time of shading.
• Shadow duration.
• Which rows are shaded.
• Shade during flowering.
• Shade during midday heat.
• Alternation between sun and shade.

For higher fidelity:

1. Use many narrow spatial zones.
2. Recalculate zones as crop height changes.
3. Use measured daily microclimate for every zone.
4. Modify DSSAT source code or create an external coupling interface for finer time steps.
5. Compare the daily-zone approximation with detailed field observations.

The DSSAT source code is available and has a modular crop–soil–weather structure, but modifying it requires Fortran compilation and model-development experience. (DSSAT.net)

---

10. Calculate PV electricity separately

Use PVlib, PVsyst, SAM or another validated PV model.

Required PV inputs include:

• Module specifications.
• Inverter specifications.
• DC capacity.
• AC capacity.
• Tilt and azimuth.
• Tracker limits.
• Pitch and GCR.
• Bifaciality.
• Albedo.
• Module temperature.
• Soiling.
• Wiring and mismatch losses.
• Availability and curtailment.

Calculate:

• Annual DC energy.
• Annual AC energy.
• Specific yield, kWh/kWp.
• Performance ratio.
• Bifacial gain.
• Clipping.
• Shading and mismatch losses.

DSSAT should supply the crop outcome; the PV model supplies the electricity outcome.

---

11. Combine crop and energy results

Crop yield ratio

$$CYRATIO=\frac{Y_{APV}}{Y_{open}}$$

Crop yield reduction

$$CYR=1-\frac{Y_{APV}}{Y_{open}}$$

Multiply by 100 for percentage reduction.

Land equivalent ratio

A common food–energy form is:

$$LER= \frac{Y_{APV}}{Y_{open}} + \frac{E_{APV}}{E_{PV-only}}$$

Where the reference systems must use consistent land area, weather and system boundaries.

Interpretation:

• $LER>1$: combined land use produces more than separate reference uses.
• $LER=1$: no combined land-use advantage.
• $LER<1$: separate uses outperform the combined system.

Also report:

• Water-use efficiency.
• Irrigation-use efficiency.
• Crop gross margin.
• Electricity revenue.
• Installation and operating costs.
• Net present value.
• Crop-yield variability.
• Probability of yield falling below an acceptable threshold.

The 2025 DSSAT–PVlib rice study evaluated crop-yield reduction, annual energy, LER and economics together rather than selecting a configuration based only on electricity. (MDPI)

---

12. Run design scenarios

Once the model is calibrated and validated, vary:

PV variables

• GCR.
• Panel height.
• Row spacing.
• Tilt.
• Azimuth.
• Fixed versus tracking.
• Tracker rotation limits.
• Backtracking.
• Shade-aware tracking.
• Transparent-module fraction.
• Bifacial versus monofacial modules.

Crop variables

• Crop species.
• Cultivar.
• Planting date.
• Row orientation.
• Plant density.
• Irrigation strategy.
• Nitrogen rate.
• Residue management.

Climate variables

• Historical weather years.
• Dry and wet years.
• Hot years.
• Future climate scenarios.
• Elevated CO₂, where relevant.

For management-risk analysis, DSSAT recommends long historical weather records, commonly approximately 30 years, so that yield distributions can be compared instead of relying on one season. (DSSAT.net)

Use DSSATSens for simple parameter sweeps or scripts for thousands of combinations.

---

13. Suggested complete workflow

text
1. Define crop, location and research question
                ↓
2. Install DSSAT and run included examples
                ↓
3. Design open-field and PV field treatments
                ↓
4. Measure weather, shade, soil, crop and PV electricity
                ↓
5. Create open-field DSSAT weather, soil and management files
                ↓
6. Calibrate cultivar with open-field/non-stressed observations
                ↓
7. Validate with independent open-field experiments
                ↓
8. Build or validate the PV shadow model
                ↓
9. Divide the agrivoltaic field into irradiance/microclimate zones
                ↓
10. Create modified daily weather for every zone
                ↓
11. Run DSSAT for every zone
                ↓
12. Compare simulated and measured PV-zone crop observations
                ↓
13. Area-weight crop yield and water use
                ↓
14. Simulate electricity using PVlib/PVsyst/SAM
                ↓
15. Calculate yield ratio, LER, water use and economics
                ↓
16. Test alternative PV and crop-management configurations
                ↓
17. Quantify uncertainty and report limitations

---

14. Common errors to avoid

1. Reducing SRAD by one constant percentage for the whole season.
   Real PV shading changes by hour, season, position, crop height and tracker angle.

2. Calibrating cultivar coefficients separately under panels.
   This can hide errors in the shading or microclimate model.

3. Using the weather station above the panels as crop-level radiation.
   The crop experiences a different irradiance environment.

4. Ignoring panel runoff.
   Below-panel soil may receive less rain while drip edges receive concentrated water.

5. Using one DSSAT run for the complete PV field.
   A single run cannot represent strong spatial gradients.

6. Validating only final yield.
   Check phenology, LAI, biomass, soil water and yield.

7. Using only one weather year.
   PV–crop performance can differ greatly between dry, cool, hot and cloudy years.

8. Treating DSSAT output as exact.
   Report parameter, weather, soil and PV-shadow uncertainty.

9. Ignoring pests, weeds or diseases.
   Unless specifically represented, DSSAT generally assumes these are adequately controlled.

10. Comparing inappropriate reference systems.
    Open-field crop and PV-only references must use consistent land-area and weather assumptions.

For a tailored setup, the essential details are the crop, location, PV type, panel geometry, available field observations and whether the system is rainfed or irrigated.