Mar 20, 2026
Version 3
Modelling crop-growing stages in CLEWs V.3
- Camilla Lo Giudice1,
- Francesco Gardumi1
- 1KTH
- Camilla
Protocol Citation: Camilla Lo Giudice, Francesco Gardumi 2026. Modelling crop-growing stages in CLEWs. protocols.io https://dx.doi.org/10.17504/protocols.io.14egn1n6mv5d/v3Version created by camlg
License: This is an open access protocol distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited
Protocol status: Working
We use this protocol and it's working
Created: March 19, 2026
Last Modified: March 20, 2026
Protocol Integer ID: 313585
Keywords: higher water requirement from the crop, modelling crop, clews model for each land, seasonal crop, rainfall precipitation pattern, variability due to climate change impact, variable water demand, land water requirement, precipitation, growing crop, climate change impact, crop, water availability, irrigation, supply if water scarcity, water requirement, vegetation, runoff, clews model, water scarcity, amount of water, water need, using yearly average, higher water requirement, cropland, growth stage, stages in clew, growth stages in order, mismatch in water availability, growing stage, yearly average, variability, grassland, increased variability, harvested phase
Funders Acknowledgements:
European Union’s Horizon research program IAM COMPACT
Grant ID: 101056306
Abstract
In a CLEWs model for each land-use, the amount of water that goes in (precipitation and irrigation if needed) is equal to the output (evapotranspiration, runoff, and groundwater recharge). Previous CLEWs studies modelled land water requirements using yearly averages. However, despite this simplification is true for land covers where the vegetations regenerates naturally (such as forest, grassland and water bodies), water needs, and therefore outputs, can change greatly for cropland. Land used for growing crops has an intra-annual variability since crops are planted and harvested regularly within the year. Depending on the season, crops’ water requirements can be three times more if we are in the growing stage than the harvested phase. Rainfall precipitation patterns are forecasted to increase their variability due to climate change impact. The increased variability might lead to a mismatch in water availability and supply if water scarcity occurs when there is a higher water requirement from the crops. Because of that, we modelled seasonal crop-growth stages in order to represent variable water demand.
Image Attribution
Figure 1 Adding seasons in the CLEW reference system. The figure represents a schematic overview of the water balance represented in OSeMOSYS for a general crop (LNDCROP). On the left-side, we see the inputs to that technology: precipitation, land, season1 and season 2. On the right-side we see the output: evapotranspiration, groundwater, surface water, and crop demand. For each technology (represented as a square), there are two modes of operation associated (continued line or dotted line) which have to be consistent to the season represented (see section below). Figure 2 Harvested area in Season 1 (left) and Season 2 (right). The figure represents on the vertical axis the total harvested area in 1000 sq. Km, while on the horizontal axis, it represents its development through the modelled years. The horizontal axis is further divided into two sub-sections namely REF 1 and REF 2 which represents the behaviour of the two modes of operation: mode 1 and mode 2 (representing respectively growing and harvesting season).
Guidelines
For each modelled season, we increased the modes of operations. Since in this case they were two, we will show the model set-up with two modes of operations. Note that it is possible to increase the number of modelled seasons and modes, but this will lead also to increased model complexity and computational requirements. To each season-technology is associated a commodity, called S1 and S2. These commodities go into cropland uses, but the value of the input activity ratio varies according to the mode of operation. Each mode of operation (MoO) is associated with a season: in this case MoO 1 refers to Season 1 and MoO 2 to Season 2. This set-up has to be consistent for modelling all the water cycle. Another key factor of the modelling consists in associating a capacity factor to each season-technology. The capacity factor has to be 1 for the time slice that represents the season and 0 for the other one. This will force the land technology to switch between Season 1 and 2 changing water requirements and water outputs accordingly. One of the main challenges in modelling crop growing stages, is to ensure consistency in harvested area. The goal is to ensure that the same amount of land is allocated for every crop across the regions since the only variable must be the water cycle. The first step consists of keeping for both modes the same crop yield. OSeMOSYS is a demand driven model. Even if harvesting happens in the late stage (Season 2), OSeMOSYS uses linear optimization for minimizing the system’s costs and there is therefore a linear relationship between the technology’s outputs commodities. Moreover, unlike other CLEWs model, all the croplands will have 1 year as operational life and no capital costs. The reason is because it is important to ensure consistency between capacity and activity. For instance, if a land technology has 15 years of operational life, the installed capacity will last for 15 years. However, depending on the optimization process, the technology can either produce (activity = capacity), or it can be substituted by another technology that is replacing the former one (there is new capacity from another technology and therefore the former might face a decrease in the activity). In previous CLEWs model this set up was possible due to the annual average values used for characterizing cropland. In this case, we need to make sure that if there is “installed capacity” of a given crop, it is consistent across the seasons. Fully exploiting the capacity is therefore needed for avoiding having two optimized land-uses across the seasons. Because of the importance in keeping the consistency across the seasons, all the calibration was done using only the activity instead of the capacity. The reference year is characterized by a lower and upper annual activity limit (the upper limit and the max capacity can be 1% higher that the lower to avoid infeasibilities), and by a maximum capacity equal to the upper activity limit in order to make sure that the reference year respects the land-use calibration. After the first year, the model is calibrated using the parameters Technology Activity Decrease By Mode Limit and Technology Activity Increase By Mode Limit. Note that it is important to assign the same values for both modes of operations. Finally, each technology is characterized by Fixed and Variable costs. Note that also the Variable costs have to assigned for both modes. The capital costs are zero because we set 1 as Operational Life, assuming that the cost of buying land does not vary across crop types.
Materials
MUIO - OSeMOSYS User Interface
Troubleshooting
Prerequisites
Know how to design a simple CLEWs model. You can find more information on: https://www.open.edu/openlearncreate/course/view.php?id=16186
Before you start
Ensure you have calculated already (e.g on a separate Excel sheet) all the data you need for quantifying water inputs and outputs of each crop growing stage.
Ensure you have installed on your computer the MUIO v5 or higher
General methodological overviw
Figure 1 Key steps for modelling water seasonality and crop-growing stages in OSeMOSYS
Step 1 - Modelling seasons in OSeMOSYS (case with 2 seasons)
Increase the modes of operation
For each modelled season, increase the modes of operations. In this case they are two, so we will show the model set-up with two modes of operations. Having a mode of operation for each season allows modellers to characterize different water cycles.
MUIO set up:
Configure the model --> Model data --> Modes of operation
Update the model once you are done.
Note that it is possible to increase the number of modelled seasons and modes, but this will lead also to increased model complexity and computational requirements.
Create seasonal technologies and commodities
In this case the seasons will be called Season 1 and Season 2 corresponding to the growing and harvesting phase of crop-land.
MUIO set up:
Configure the model -->Technologies --> add:
Season1 = Growing season
Season2 = Harvesting season
Configure the model --> Commodities --> add:
S1
S2
Update the model once you are done.
Figure 2 Schematic reference system of the seasonal water balance represented in OSeMOSYS.
The figure represents a schematic overview of the water balance represented in OSeMOSYS for a general crop (LNDCROP). On the left-side, we see the inputs to that technology: precipitation, land, season1 and season 2. On the right-side we see the output: evapotranspiration, groundwater, surface water, and crop demand. For each technology (represented as a square), there are two modes of operation associated (continued line or dotted line) which have to be consistent to the season represented (see section below).
Set up seasonal output activity ratios
MUIO set up:
Configure the model --> Technologies --> Output Activity Ratio Only for season technologies Season1 and Season2
Figure 3 Seasonal output activity ratios - model set up
Data entry --> Output activity ratio -> Season 1, Season 2 equal to 1 only to the corresponding seasonal mode.
If for example the water balance of season 1 is represented by mode 1, the output activity ratio=1 is ONLY for mode one;
if season 2 is represented by mode 2, then the output activity ration = 1 ONLY for mode 2.
Figure 4 Seasonal output activity ratios - data entry
Set up seasonal input activity ratios for crop-land
S1 and S2 go into cropland, but the value of the input activity ratio varies according to the mode of operation. In this case MoO 1 refers to Season 1 and MoO 2 to Season 2. This set-up has to be consistent for modelling all the water cycle.
MUIO set up:
Configure the model --> Technologies -> Input Activity Ratio
Figure 5 Seasonal output input ratios - model set up
Data entry --> Input activity ratio --> input 1 for MoO 1 if the commodity is S1 and 1 for MoO 2 if the commodity is S2
Figure 6 Seasonal input activity ratios for S1 - data entry
Figure 7 Seasonal input activity ratios for S2 - data entry
Set a capacity factor to each season-technology.
Crop-growing stages and the seasonal variability need to be consistent with the model temporal set-up.
In OSeMOSYS we define the temporal resolution with the ”Time-Slices” which are the time split of each modelled year.The duration of each time-slice is defined by the parameter ”Year-Split”.
With the capacity factor, we impose the seasonal technologies to “work” only during the corresponding season. Here it is also crucial to maintain consistency across seasons and modeled time slices.The capacity factor has to be 1 for the time slice that represents the season and 0 for the other one. This will force the land technology to switch between Season 1 and 2 changing water requirements and water outputs accordingly.
Let's assume we have 6 time-slices: S11,S12,S13,S21,S22,S23.
While the second numbers (1,2,3) represent daily time, the first number (1,2) represents seasons. The capacity factors will be the following:
Figure 8 Capacity factors (CF) for seasonal technologies
Step 2 - Ensuring consistency across seasons
Consistency across seasons of crop harvested area
One of the main challenges in modelling crop growing stages, is to ensure consistency in harvested area across seasons. The goal is to ensure that the same amount of land is allocated for every crop across the regions since the only variable must be the water cycle.
Figure 9 Consistency across seasons
The crop yield
OSeMOSYS is a linear optimization, demand-driven tool. Although crops produce output only
during the harvesting season, setting the output activity ratio to zero in non-harvest seasons would prevent land from being allocated in those seasons, as no demand is being supplied.
MUIO set up:
Data entry --> Output Activity Ratio --> Paste the same values in MoO 1 to MoO 2 --> Save data
Figure 10 Crop yield - data entry
Crop demand
Current CLEWs models input crop demand as ”AccumulatedAnnualDemand”. This demand represents the total demand in one year and is not affected by the model time resolution. In order to ensure consistency across seasons, we need to exploit the model’s linear correlation between demand and supply. The demand will therefore have to be split across the model’s year split.
We use the parameter ”Specified Annual Demand” which is the total specified demand for the
year, linked to a specific ‘time of use’ during the year. The way of “linking” it to the time in OSeMOSYS is done with the “Specified Demand Profile “ parameter which is the annual fraction of commodity demand that is required in each time slice. For crops we will input the same value as the year-split.
Figure 11 The specified demand profile for crop demand
Because of the importance in keeping the consistency across the seasons, all the calibration is done using only the activity instead of the capacity. The reference year is characterized by a lower and upper annual activity limit (the upper limit and the max capacity can be 1% higher that the lower to avoid infeasibilities), and by a maximum capacity equal to the upper activity limit in order to make sure that the reference year respects the land-use calibration.
Moreover, unlike other CLEWs model, all the croplands will have 1 year as operational life and no capital costs. The reason is because it is important to ensure consistency between capacity and activity. For instance, if a land technology has 15 years of operational life, the installed capacity will last for 15 years. However, depending on the optimization process, the technology can either produce (activity = capacity), or it can be substituted by another technology that is replacing the former one (there is new capacity from another technology and therefore the former might face a decrease in the activity).
After the first year, the model is calibrated using the parameters Technology Activity Decrease By Mode Limit and Technology Activity Increase By Mode Limit. Note that it is important to assign the same values for both modes of operations.
Finally, each technology is characterized by Fixed and Variable costs. Note that also the Variable costs have to be assigned for both modes. The capital costs are zero because we set 1 as Operational Life, assuming that the cost of buying land does not vary across crop types.
Model set-up for techno-economic parameter.
| Parameter | Technology | Mode of operation | Value | |
| Total Annual Max Capacity | LNDCRP | - | Calibrated value only for the reference year | |
| Total Technology Annual Activity Lower Limit | LNDCRP | - | Calibrated value only for the reference year | |
| Total Technology Annual Activity Upper Limit | LNDCRP | - | Calibrated value only for the reference year | |
| Technology Activity Decrease By Mode Limit | LNDCRP | 1 | Same value as mode 2 | |
| Technology Activity Decrease By Mode Limit | LNDCRP | 2 | Same value as mode 1 | |
| Technology Activity Increase By Mode Limit | LNDCRP | 1 | Same value as mode 2 | |
| Technology Activity Increase By Mode Limit | LNDCRP | 2 | Same value as mode 1 | |
| Fixed Cost | LNDCRP | - | ||
| Variable Cost | LNDCRP | 1 | Same value as mode 2 | |
| Variable Cost | LNDCRP | 2 | Same value as mode 1 |
Figure 12 Adding techno-economic constraints.
Check that you have made no mistakes.
This is an example of how the harvested area should look like across the seasons (1 and 2). The figure has been created using the MUIO v.5.3 user interface, and displays the results of Rate Of Activity
filtered by croplands, and by modes of operation (Mo Id) in the
rows as shown in the figure below.
Figure 13 Consistency across seasons - result example done with MUIO v5.3
Step 3 - Modelling water seasonality
Figure 14 Schematic reference system of the seasonal water balance represented in OSeMOSYS.
Focus on the water balance
Enter the input activity ratio for crops (precipitation)
Also in this case, remember to keep the same consistency across modes and season. Mode 1 will have to represent the precipitation of season 1 and mode 2 of season 2.5
Figure 15 Input activity ratio for each crop (precipitation).
Enter the water output (evapotranspiration, groundwater, run-off water)
Figure 16 Output activity ratio for water.
Analyzing the results
The water seasonality
Figure 17 Evapotranspiration, Groundwater and run-off water seasonality in Bm3
The figure represents on the vertical axis the Bm3 of renewable groundwater and run-off water produced by the harvested area, while the horizontal axis represents its production through the modelled years. The horizontal axis is further divided into two sub-sections namely 1 and 2 which represents the behaviour of the two modes of operation: mode 1 and mode 2 (representing respectively growing and harvesting season).
Analysis
It is possible to observe differences in run-off and groundwater recharge across seasons. These differences depend on the overall seasonal water balance. For instance, run-off increases in Season 2 because this period corresponds to the crop-growth stage during which plants require less water. As a result, evapotranspiration decreases, leading to an increase in runoff.
On the other hand, groundwater recharge decreases in Season 2 because it corresponds to the dry season. The reduction in precipitation during this period negatively affects groundwater recharge.
Acknowledgements
The authors would like to acknowledge also the funding from UK Aid from the UK Government via the
Climate Compatible Growth programme. However, the views expressed herein do not necessarily reflect the UK government’s official policies.