2.26. Crops and Irrigation

2.26.1. Summary of CLM5.0 updates relative to the CLM4.5

We describe here the complete crop and irrigation parameterizations that appear in CLM5.0. Corresponding information for CLM4.5 appeared in the CLM4.5 Technical Note (Oleson et al. 2013).

CLM5.0 includes the following new updates to the CROP option, where CROP refers to the interactive crop management model and is included as an option with the BGC configuration:

• New crop functional types

• All crop areas are actively managed

• Fertilization rates updated based on crop type and geographic region

• New Irrigation triggers

• Phenological triggers vary by latitude for some crop types

• Ability to simulate transient crop management

• Adjustments to allocation and phenological parameters

• Crops reaching their maximum LAI triggers the grain fill phase

• Grain C and N pools are included in a 1-year product pool

• C for annual crop seeding comes from the grain C pool

• Initial seed C for planting is increased from 1 to 3 g C/m^2

These updates appear in detail in the sections below. Many also appear in Levis et al. (2016).

2.26.1.1. Available new features since the CLM5 release

• Addition of bioenergy crops

• Ability to customize crop calendars (sowing windows/dates, maturity requirements) using stream files

• Cropland soil tillage

• Crop residue removal

2.26.2. The crop model: cash and bioenergy crops

2.26.2.1. Introduction

Groups developing Earth System Models generally account for the human footprint on the landscape in simulations of historical and future climates. Traditionally we have represented this footprint with natural vegetation types and particularly grasses because they resemble many common crops. Most modeling efforts have not incorporated more explicit representations of land management such as crop type, planting, harvesting, tillage, fertilization, and irrigation, because global scale datasets of these factors have lagged behind vegetation mapping. As this begins to change, we increasingly find models that will simulate the biogeophysical and biogeochemical effects not only of natural but also human-managed land cover.

AgroIBIS is a state-of-the-art land surface model with options to simulate dynamic vegetation (Kucharik et al. 2000) and interactive crop management (Kucharik and Brye 2003). The interactive crop management parameterizations from AgroIBIS (March 2003 version) were coupled as a proof-of-concept to the Community Land Model version 3 [CLM3.0, Oleson et al. (2004) ] (not published), then coupled to the CLM3.5 (Levis et al. 2009) and later released to the community with CLM4CN (Levis et al. 2012), and CLM4.5BGC. Additional updates after the release of CLM4.5 were available by request (Levis et al. 2016), and those are now incorporated into CLM5.

With interactive crop management and, therefore, a more accurate representation of agricultural landscapes, we hope to improve the CLM’s simulated biogeophysics and biogeochemistry. These advances may improve fully coupled simulations with the Community Earth System Model (CESM), while helping human societies answer questions about changing food, energy, and water resources in response to climate, environmental, land use, and land management change (e.g., Kucharik and Brye 2003; Lobell et al. 2006). As implemented here, the crop model uses the same physiology as the natural vegetation but with uses different crop-specific parameter values, phenology, and allocation, as well as fertilizer and irrigation management.

2.26.2.2. Crop plant functional types

To allow crops to coexist with natural vegetation in a grid cell, the vegetated land unit is separated into a naturally vegetated land unit and a managed crop land unit. Unlike the plant functional types (PFTs) in the naturally vegetated land unit, the managed crop PFTs in the managed crop land unit do not share soil columns and thus permit for differences in the land management between crops. Each crop type has a rainfed and an irrigated PFT that are on independent soil columns. Crop grid cell coverage is assigned from satellite data (similar to all natural PFTs), and the managed crop type proportions within the crop area is based on the dataset created by Portmann et al. (2010) for present day. New in CLM5, crop area is extrapolated through time using the dataset provided by Land Use Model Intercomparison Project (LUMIP), which is part of CMIP6 Land use timeseries (Lawrence et al. 2016). For more details about how crop distributions are determined, see Chapter 2.27.

CLM5 includes ten actively managed crop types (temperate soybean, tropical soybean, temperate corn, tropical corn, spring wheat, cotton, rice, sugarcane, miscanthus, and switchgrass) that are chosen based on the availability of corresponding algorithms in AgroIBIS and as developed by Badger and Dirmeyer (2015) and described by Levis et al. (2016), or from available observations as described by Cheng et al. (2019). The representations of sugarcane, rice, cotton, tropical corn, and tropical soy were new in CLM5; miscanthus and switchgrass were added after the CLM5 release. Sugarcane and tropical corn are both C4 plants and are therefore represented using the temperate corn functional form. Tropical soybean uses the temperate soybean functional form, while rice and cotton use the wheat functional form. In tropical regions, parameter values were developed for the Amazon Basin, and planting date window is shifted by six months relative to the Northern Hemisphere. Plantation areas of bioenergy crops are projected to expand throughout the 21st century as a major energy source to replace fossil fuels and mitigate climate change. Miscanthus and switchgrass are perennial bioenergy crops and have quite different physiological traits and land management practices than annual crops, such as longer growing seasons, higher productivity, and lower demands for nutrients and water. About 70% of biofuel aboveground biomass (leaf & livestem) is removed at harvest. Parameter values were developed by using observation data collected at the University of Illinois Energy Farm located in Central Midwestern United States (Cheng et al., 2019).

In addition, CLM’s default list of plant functional types (PFTs) includes an irrigated and unirrigated unmanaged C3 crop (Table 2.26.1) treated as a second C3 grass. The unmanaged C3 crop is only used when the crop model is not active and has grid cell coverage assigned from satellite data, and the unmanaged C3 irrigated crop type is currently not used since irrigation requires the crop model to be active. The default list of PFTs also includes twenty-one inactive crop PFTs that do not yet have associated parameters required for active management. Each of the inactive crop types is simulated using the parameters of the spatially closest associated crop type that is most similar to the functional type (e.g., C3 or C4), which is required to maintain similar phenological parameters based on temperature thresholds. Information detailing which parameters are used for each crop type is included in Table 2.26.1. It should be noted that PFT-level history output merges all crop types into the actively managed crop type, so analysis of crop-specific output will require use of the land surface dataset to remap the yields of each actively and inactively managed crop type. Otherwise, the actively managed crop type will include yields for that crop type and all inactively managed crop types that are using the same parameter set.

Table 2.26.1 Crop plant functional types (PFTs) included in CLM5BGCCROP.

IVT	Plant function types (PFTs)	Management Class	Crop Parameters Used
15	c3 unmanaged rainfed crop	none	not applicable
16	c3 unmanaged irrigated crop	none	not applicable
17	rainfed temperate corn	active	rainfed temperate corn
18	irrigated temperate corn	active	irrigated temperate corn
19	rainfed spring wheat	active	rainfed spring wheat
20	irrigated spring wheat	active	irrigated spring wheat
21	rainfed winter wheat	inactive	rainfed spring wheat
22	irrigated winter wheat	inactive	irrigated spring wheat
23	rainfed temperate soybean	active	rainfed temperate soybean
24	irrigated temperate soybean	active	irrigated temperate soybean
25	rainfed barley	inactive	rainfed spring wheat
26	irrigated barley	inactive	irrigated spring wheat
27	rainfed winter barley	inactive	rainfed spring wheat
28	irrigated winter barley	inactive	irrigated spring wheat
29	rainfed rye	inactive	rainfed spring wheat
30	irrigated rye	inactive	irrigated spring wheat
31	rainfed winter rye	inactive	rainfed spring wheat
32	irrigated winter rye	inactive	irrigated spring wheat
33	rainfed cassava	inactive	rainfed rice
34	irrigated cassava	inactive	irrigated rice
35	rainfed citrus	inactive	rainfed spring wheat
36	irrigated citrus	inactive	irrigated spring wheat
37	rainfed cocoa	inactive	rainfed rice
38	irrigated cocoa	inactive	irrigated rice
39	rainfed coffee	inactive	rainfed rice
40	irrigated coffee	inactive	irrigated rice
41	rainfed cotton	active	rainfed cotton
42	irrigated cotton	active	irrigated cotton
43	rainfed datepalm	inactive	rainfed cotton
44	irrigated datepalm	inactive	irrigated cotton
45	rainfed foddergrass	inactive	rainfed spring wheat
46	irrigated foddergrass	inactive	irrigated spring wheat
47	rainfed grapes	inactive	rainfed spring wheat
48	irrigated grapes	inactive	irrigated spring wheat
49	rainfed groundnuts	inactive	rainfed rice
50	irrigated groundnuts	inactive	irrigated rice
51	rainfed millet	inactive	rainfed tropical corn
52	irrigated millet	inactive	irrigated tropical corn
53	rainfed oilpalm	inactive	rainfed rice
54	irrigated oilpalm	inactive	irrigated rice
55	rainfed potatoes	inactive	rainfed spring wheat
56	irrigated potatoes	inactive	irrigated spring wheat
57	rainfed pulses	inactive	rainfed spring wheat
58	irrigated pulses	inactive	irrigated spring wheat
59	rainfed rapeseed	inactive	rainfed spring wheat
60	irrigated rapeseed	inactive	irrigated spring wheat
61	rainfed rice	active	rainfed rice
62	irrigated rice	active	irrigated rice
63	rainfed sorghum	inactive	rainfed tropical corn
64	irrigated sorghum	inactive	irrigated tropical corn
65	rainfed sugarbeet	inactive	rainfed spring wheat
66	irrigated sugarbeet	inactive	irrigated spring wheat
67	rainfed sugarcane	active	rainfed sugarcane
68	irrigated sugarcane	active	irrigated sugarcane
69	rainfed sunflower	inactive	rainfed spring wheat
70	irrigated sunflower	inactive	irrigated spring wheat
71	rainfed miscanthus	active	rainfed miscanthus
72	irrigated miscanthus	active	irrigated miscanthus
73	rainfed switchgrass	active	rainfed switchgrass
74	irrigated switchgrass	active	irrigated switchgrass
75	rainfed tropical corn	active	rainfed tropical corn
76	irrigated tropical corn	active	irrigated tropical corn
77	rainfed tropical soybean	active	rainfed tropical soybean
78	irrigated tropical soybean	active	irrigated tropical soybean

2.26.2.3. Phenology

CLM5-BGC includes evergreen, seasonally deciduous (responding to changes in day length), and stress deciduous (responding to changes in temperature and/or soil moisture) phenology algorithms (Chapter 2.20). CLM5-BGC-crop uses the AgroIBIS crop phenology algorithm, consisting of three distinct phases.

Phase 1 starts at planting and ends with leaf emergence, phase 2 continues from leaf emergence to the beginning of grain fill, and phase 3 starts from the beginning of grain fill and ends with physiological maturity and harvest.

2.26.2.3.1. Planting

All crops must meet the following requirements between the minimum planting date and the maximum planting date (for the northern hemisphere) in Table 2.26.2:

(2.26.1)\[\begin{split}\begin{array}{c} {T_{10d} >T_{p} } \\ {T_{10d}^{\min } >T_{p}^{\min } } \\ {GDD_{8} \ge GDD_{\min } } \end{array}\end{split}\]

where \({T}_{10d}\) is the 10-day running mean of \({T}_{2m}\), (the simulated 2-m air temperature during each model time step) and \(T_{10d}^{\min}\) is the 10-day running mean of \(T_{2m}^{\min }\) (the daily minimum of \({T}_{2m}\)). \({T}_{p}\) and \(T_{p}^{\min }\) are crop-specific coldest planting temperatures (Table 2.26.2), \({GDD}_{8}\) is the 20-year running mean growing degree-days (units are °C day) tracked from April through September (NH) above 8°C with maximum daily increments of 30 degree-days (see equation (2.26.3)), and \({GDD}_{min }\)is the minimum growing degree day requirement (Table 2.26.2). \({GDD}_{8}\) does not change as quickly as \({T}_{10d}\) and \(T_{10d}^{\min }\), so it determines whether it is warm enough for the crop to be planted in a grid cell, while the 2-m air temperature variables determine the day when the crop may be planted if the \({GDD}_{8}\) threshold is met. If the requirements in equation (2.26.1) are not met by the maximum planting date, crops are still planted on the maximum planting date as long as \({GDD}_{8} > 0\). In the southern hemisphere (SH) the NH requirements apply 6 months later.

At planting, each crop seed pool is assigned 3 gC m^-2 from its grain product pool. The seed carbon is transferred to the leaves upon leaf emergence. An equivalent amount of seed leaf N is assigned given the PFT’s C to N ratio for leaves (\({CN}_{leaf}\) in Table 2.26.3; this differs from AgroIBIS, which uses a seed leaf area index instead of seed C). The model updates the average growing degree-days necessary for the crop to reach vegetative and physiological maturity, \({GDD}_{mat}\), according to the following AgroIBIS rules:

(2.26.2)\[\begin{split}\begin{array}{lll} GDD_{{\rm mat}}^{{\rm corn,sugarcane}} =0.85 GDD_{{\rm 8}} & {\rm \; \; \; and\; \; \; }& 950 <GDD_{{\rm mat}}^{{\rm corn,sugarcane}} <1850{}^\circ {\rm days} \\ GDD_{{\rm mat}}^{{\rm spring\ wheat,cotton}} =GDD_{{\rm 0}} & {\rm \; \; \; and\; \; \; } & GDD_{{\rm mat}}^{{\rm spring\ wheat,cotton}} <1700{}^\circ {\rm days} \\ GDD_{{\rm mat}}^{{\rm temp.soy}} =GDD_{{\rm 10}} & {\rm \; \; \; and\; \; \; } & GDD_{{\rm mat}}^{{\rm temp.soy}} <1900{}^\circ {\rm days} \\ GDD_{{\rm mat}}^{{\rm rice}} =GDD_{{\rm 0}} & {\rm \; \; \; and\; \; \; } & GDD_{{\rm mat}}^{{\rm rice}} <2100{}^\circ {\rm days} \\ GDD_{{\rm mat}}^{{\rm trop.soy}} =GDD_{{\rm 10}} & {\rm \; \; \; and\; \; \; } & GDD_{{\rm mat}}^{{\rm trop.soy}} <2100{}^\circ {\rm days} \end{array}\end{split}\]

where \({GDD}_{0}\), \({GDD}_{8}\), and \({GDD}_{10}\) are the 20-year running mean growing degree-days tracked from April through September (NH) over 0°C, 8°C, and 10°C, respectively, with maximum daily increments of 26 degree-days (for \({GDD}_{0}\)) or 30 degree-days (for \({GDD}_{8}\) and \({GDD}_{10}\)). Equation (2.26.3) shows how we calculate \({GDD}_{0}\), \({GDD}_{8}\), and \({GDD}_{10}\) for each model timestep:

(2.26.3)\[\begin{split}\begin{array}{lll} GDD_{{\rm 0}} =GDD_{0} +T_{2{\rm m}} -T_{f} & \quad {\rm \; \; \; where\; \; \; } & 0 \le T_{2{\rm m}} -T_{f} \le 26{}^\circ {\rm days} \\ GDD_{{\rm 8}} =GDD_{8} +T_{2{\rm m}} -T_{f} -8 & \quad {\rm \; \; \; where\; \; \; } & 0 \le T_{2{\rm m}} -T_{f} -8\le 30{}^\circ {\rm days} \\ GDD_{{\rm 10}} =GDD_{10} +T_{2{\rm m}} -T_{f} -10 & \quad {\rm \; \; \; where\; \; \; } & 0 \le T_{2{\rm m}} -T_{f} -10\le 30{}^\circ {\rm days} \end{array}\end{split}\]

where, if \({T}_{2m}\) - \({T}_{f}\) takes on values outside the above ranges within a day, then it equals the minimum or maximum value in the range for that day. \({T}_{f}\) is the freezing temperature of water and equals 273.15 K, \({T}_{2m}\) is the 2-m air temperature in units of K, and GDD is in units of degree-days.

2.26.2.3.2. Leaf emergence

According to AgroIBIS, leaves may emerge when the growing degree-days of soil temperature to 0.05 m depth (\(GDD_{T_{soi} }\) ), which is tracked since planting, reaches 1 to 5% of \({GDD}_{mat}\) (see Phase 2 % \({GDD}_{mat}\) in Table 2.26.2). The base temperature threshold values for \(GDD_{T_{soi} }\) are listed in Table 2.26.2 (the same base temperature threshold values are also used for \(GDD_{T_{{\rm 2m}} }\) in section 2.26.2.3.3), and leaf emergence (crop phenology phase 2) starts when this threshold is met. Leaf onset occurs in the first time step of phase 2, at which moment all seed C is transferred to leaf C. Subsequently, the leaf area index generally increases throughout phase 2 until it reaches a predetermined maximum value. Stem and root C also increase throughout phase 2 based on the carbon allocation algorithm in section 2.26.2.4.1.

2.26.2.3.3. Grain fill

The grain fill phase (phase 3) begins in one of two ways. The first potential trigger is based on temperature, similar to phase 2. A variable tracked since planting, similar to \(GDD_{T_{soi} }\) but for 2-m air temperature, \(GDD_{T_{{\rm 2m}} }\), must reach a heat unit threshold, h, of of 40 to 65% of \({GDD}_{mat}\) (see Phase 3 % \({GDD}_{mat}\) in Table 2.26.2). For crops with the C4 photosynthetic pathway (temperate and tropical corn, sugarcane), the \({GDD}_{mat}\) is based on an empirical function and ranges between 950 and 1850. The second potential trigger for phase 3 is based on leaf area index. When the maximum value of leaf area index is reached in phase 2 (Table 2.26.3), phase 3 begins. In phase 3, the leaf area index begins to decline in response to a background litterfall rate calculated as the inverse of leaf longevity for the PFT as done in the BGC part of the model.

2.26.2.3.4. Harvest

Harvest is assumed to occur as soon as the crop reaches maturity. When \(GDD_{T_{{\rm 2m}} }\) reaches 100% of \({GDD}_{mat}\) or the number of days past planting reaches a crop-specific maximum (Table 2.26.2), then the crop is harvested. Harvest occurs in one time step using the BGC leaf offset algorithm.

Table 2.26.2 Crop phenology and morphology parameters for the active crop plant functional types (PFTs) in CLM5BGCCROP. Numbers in the first row correspond to the list of PFTs in Table 2.26.1.

	temperate corn	spring wheat	temperate soybean	cotton	rice	sugarcane	tropical corn	tropical soybean	miscanthus	switchgrass
IVT	17, 18	19, 20	23, 24	41, 42	61, 62	67, 68	75, 76	77, 78	71, 72	73, 74
\(Date_{planting}^{min}\)	April 1	April 1	May 1	April 1	Janurary 1	Janurary 1	March 20	April 15	April 1	April 1
\(Date_{planting}^{max}\)	June 15	June 15	June 15	May 31	Feburary 28	March 31	April 15	June 31	June 15	June 15
\(T_{p}\)(K)	283.15	280.15	286.15	294.15	294.15	294.15	294.15	294.15	283.15	283.15
\(T_{p}^{ min }\)(K)	279.15	272.15	279.15	283.15	283.15	283.15	283.15	283.15	279.15	279.15
\({GDD}_{min}\) (degree-days)	50	50	50	50	50	50	50	50	50	50
base temperature for GDD (°C)	8	0	10	10	10	10	10	10	8	8
\({GDD}_{mat}\) (degree-days)	950-1850	≤ 1700	≤ 1900	≤ 1700	≤ 2100	950-1850	950-1850	≤ 2100	950-1850	950-1850
Phase 2 % \({GDD}_{mat}\)	3%	5%	3%	3%	1%	3%	3%	3%	3%	3%
Phase 3 % \({GDD}_{mat}\)	65%	60%	50%	50%	40%	65%	50%	50%	40%	40%
Max. growing season length (\(mxmat\))	165	150	150	160	150	300	160	150	210	210
\(z_{top}^{\max }\) (m)	2.5	1.2	0.75	1.5	1.8	4	2.5	1	2.5	2.5
SLA (m 2 leaf g -1 C)	0.05	0.035	0.035	0.035	0.035	0.05	0.05	0.035	0.057	0.049
\(\chi _{L}\) index	-0.5	-0.5	-0.5	-0.5	-0.5	-0.5	-0.5	-0.5	-0.5	-0.5
grperc	0.11	0.11	0.11	0.11	0.11	0.11	0.11	0.11	0.11	0.11
flnr	0.293	0.41	0.41	0.41	0.41	0.293	0.293	0.41	0.293	0.293
fcur	1	1	1	1	1	1	1	1	1	1

Notes:

• \(Date_{planting}^{min}\) and \(Date_{planting}^{max}\) are the minimum and maximum planting dates (defining the “sowing window”) in the Northern Hemisphere; the corresponding dates in the Southern Hemisphere are shifted by 6 months. (See Sect. 2.26.2.3.1.) These parameters can also be set with more geographic variation via input map stream files stream_fldFileName_swindow_start and stream_fldFileName_swindow_end.

• \(T_{p}\) and \(T_{p}^{ min }\) are crop-specific average and coldest planting temperatures, respectively. (See Sect. 2.26.2.3.1.)

• \(GDD_{min}\) is a threshold describing the coolest historical climate a patch can have had in order for a crop to be sown there; see Sect. 2.26.2.3.1 for details.

• \(GDD_{mat}\) is the heat unit index, in units of accumulated growing degree-days, a crop needs to reach maturity.

• \(mxmat\) is the maximum growing season length (days past planting), at which harvest occurs even if heat unit index has not reached \(GDD_{mat}\).

• \(z_{top}^{\max }\) is the maximum top-of-canopy height of a crop (see Sect. 2.2.1.3).

• SLA is specific leaf area (see Chapter 2.10).

• \(\chi _{L}\) is the leaf orientation index, equals -1 for vertical, 0 for random, and 1 for horizontal leaf orientation. (See Sect. 2.3.1.)

• grperc is the growth respiration factor (see Sect. 2.17.1.2).

• flnr is the fraction of leaf N in the Rubisco enzyme (a.k.a. \(N_{cb}\) in Sect. 2.10.2.1).

• fcur is the fraction of allocation that goes to currently displayed growth (i.e., that is not sent to storage). See Sect. 2.19.4.

2.26.2.4. Allocation

Allocation changes based on the crop phenology phases phenology (section 2.26.2.3). Simulated C assimilation begins every year upon leaf emergence in phase 2 and ends with harvest at the end of phase 3; therefore, so does the allocation of such C to the crop’s leaf, live stem, fine root, and reproductive pools.

Typically, C:N ratios in plant tissue vary throughout the growing season and tend to be lower during early growth stages and higher in later growth stages. In order to account for this seasonal change, two sets of C:N ratios are established in CLM for the leaf, stem, and fine root of crops: one during the leaf emergence phase (phenology phase 2), and a second during grain fill phase (phenology phase 3). This modified C:N ratio approach accounts for the nitrogen retranslocation that occurs during the grain fill phase (phase 3) of crop growth. Leaf, stem, and root C:N ratios for phase 2 are calculated using the new CLM5 carbon and nitrogen allocation scheme (Chapter 2.19), which provides a target C:N value (Table 2.26.3) and allows C:N to vary through time. During grain fill (phase 3) of the crop growth cycle, a portion of the nitrogen in the plant tissues is moved to a storage pool to fulfill nitrogen demands of organ (reproductive pool) development, such that the resulting C:N ratio of the plant tissue is reflective of measurements at harvest. All C:N ratios were determined by calibration process, through comparisons of model output versus observations of plant carbon throughout the growing season.

The BGC part of the model keeps track of a term representing excess maintenance respiration, which supplies the carbon required for maintenance respiration during periods of low photosynthesis (Chapter 2.17). Carbon supply for excess maintenance respiration cannot continue to happen after harvest for annual crops, so at harvest the excess respiration pool is turned into a flux that extracts CO₂ directly from the atmosphere. This way any excess maintenance respiration remaining at harvest is eliminated as if such respiration had not taken place.

2.26.2.4.1. Leaf emergence

During phase 2, the allocation coefficients (fraction of available C) to each C pool are defined as:

(2.26.4)\[\begin{split}\begin{array}{l} {a_{repr} =0} \\ {a_{froot} =a_{froot}^{i} -(a_{froot}^{i} -a_{froot}^{f} )\frac{GDD_{T_{{\rm 2m}} } }{GDD_{{\rm mat}} } {\rm \; \; \; where\; \; \; }\frac{GDD_{T_{{\rm 2m}} } }{GDD_{{\rm mat}} } \le 1} \\ {a_{leaf} =(1-a_{froot} )\cdot \frac{a_{leaf}^{i} (e^{-b} -e^{-b\frac{GDD_{T_{{\rm 2m}} } }{h} } )}{e^{-b} -1} {\rm \; \; \; where\; \; \; }b=0.1} \\ {a_{livestem} =1-a_{repr} -a_{froot} -a_{leaf} } \end{array}\end{split}\]

where \(a_{leaf}^{i}\), \(a_{froot}^{i}\), and \(a_{froot}^{f}\) are initial and final values of these coefficients (Table 2.26.3), and h is a heat unit threshold defined in section 2.26.2.3.3. At a crop-specific maximum leaf area index, \({L}_{max}\) (Table 2.26.3), carbon allocation is directed exclusively to the fine roots.

2.26.2.4.2. Grain fill

The calculation of \(a_{froot}\) remains the same from phase 2 to phase 3. During grain fill (phase 3), other allocation coefficients change to:

(2.26.5)\[\begin{split}\begin{array}{ll} a_{leaf} =a_{leaf}^{i,3} & {\rm when} \quad a_{leaf}^{i,3} \le a_{leaf}^{f} \quad {\rm else} \\ a_{leaf} =a_{leaf} \left(1-\frac{GDD_{T_{{\rm 2m}} } -h}{GDD_{{\rm mat}} d_{L} -h} \right)^{d_{alloc}^{leaf} } \ge a_{leaf}^{f} & {\rm where} \quad \frac{GDD_{T_{{\rm 2m}} } -h}{GDD_{{\rm mat}} d_{L} -h} \le 1 \\ \\ a_{livestem} =a_{livestem}^{i,3} & {\rm when} \quad a_{livestem}^{i,3} \le a_{livestem}^{f} \quad {\rm else} \\ a_{livestem} =a_{livestem} \left(1-\frac{GDD_{T_{{\rm 2m}} } -h}{GDD_{{\rm mat}} d_{L} -h} \right)^{d_{alloc}^{stem} } \ge a_{livestem}^{f} & {\rm where} \quad \frac{GDD_{T_{{\rm 2m}} } -h}{GDD_{{\rm mat}} d_{L} -h} \le 1 \\ \\ a_{repr} =1-a_{froot} -a_{livestem} -a_{leaf} \end{array}\end{split}\]

where \(a_{leaf}^{i,3}\) and \(a_{livestem}^{i,3}\) (initial values) equal the last \(a_{leaf}\) and \(a_{livestem}\) calculated in phase 2, \(d_{L}\), \(d_{alloc}^{leaf}\) and \(d_{alloc}^{stem}\) are leaf area index and leaf and stem allocation decline factors, and \(a_{leaf}^{f}\) and \(a_{livestem}^{f}\) are final values of these allocation coefficients (Table 2.26.3).

2.26.2.4.3. Nitrogen retranslocation for crops

Nitrogen retranslocation in crops occurs when nitrogen that was used for tissue growth of leaves, stems, and fine roots during the early growth season is remobilized and used for grain development (Pollmer et al. 1979, Crawford et al. 1982, Simpson et al. 1983, Ta and Weiland 1992, Barbottin et al. 2005, Gallais et al. 2006, Gallais et al. 2007). Nitrogen allocation for crops follows that of natural vegetation, is supplied in CLM by the soil mineral nitrogen pool, and depends on C:N ratios for leaves, stems, roots, and organs. Nitrogen demand during organ development is fulfilled through retranslocation from leaves, stems, and roots. Nitrogen retranslocation is initiated at the beginning of the grain fill stage for all crops except soybean, for which retranslocation is after LAI decline. Nitrogen stored in the leaf and stem is moved into a storage retranslocation pool for all crops, and for wheat and rice, nitrogen in roots is also released into the retranslocation storage pool. The quantity of nitrogen mobilized depends on the C:N ratio of the plant tissue and is calculated as

(2.26.6)\[leaf\_ to\_ retransn=N_{leaf} -\frac{C_{leaf} }{CN_{leaf}^{f} }\]

(2.26.7)\[stemn\_ to\_ retransn=N_{stem} -\frac{C_{stem} }{CN_{stem}^{f} }\]

(2.26.8)\[frootn\_ to\_ retransn=N_{froot} -\frac{C_{froot} }{CN_{froot}^{f} }\]

where \({C}_{leaf}\), \({C}_{stem}\), and \({C}_{froot}\) is the carbon in the plant leaf, stem, and fine root, respectively, \({N}_{leaf}\), \({N}_{stem}\), and \({N}_{froot}\) is the nitrogen in the plant leaf, stem, and fine root, respectively, and \(CN^f_{leaf}\), \(CN^f_{stem}\), and \(CN^f_{froot}\) is the post-grain fill C:N ratio of the leaf, stem, and fine root respectively (Table 2.26.3). Since C:N measurements are often taken from mature crops, pre-grain development C:N ratios for leaves, stems, and roots in the model are optimized to allow maximum nitrogen accumulation for later use during organ development, and post-grain fill C:N ratios are assigned the same as crop residue. After nitrogen is moved into the retranslocated pool, the nitrogen in this pool is used to meet plant nitrogen demand by assigning the available nitrogen from the retranslocated pool equal to the plant nitrogen demand for each organ (\({CN_{[organ]}^{f} }\) in Table 2.26.3). Once the retranslocation pool is depleted, soil mineral nitrogen pool is used to fulfill plant nitrogen demands.

2.26.2.4.4. Harvest

Whereas live crop C and N in grain was formerly transferred to the litter pool upon harvest, CLM5 splits this between “food” and “seed” pools. In the former—more generally a “crop product” pool—C and N decay to the atmosphere over one year, similar to how the wood product pools work. The latter is used in the subsequent year to account for the C and N required for crop seeding.

Live leaf and stem biomass at harvest is transferred to biofuel, removed residue, and/or litter pools.

For the biofuel crops Miscanthus and switchgrass, 70% of live leaf and stem biomass at harvest is transferred to the crop product pool as described for “food” harvest above. This value can be changed for these crops—or set to something other than the default zero for any other crop—with the parameter \(biofuel\_harvfrac\) (0-1).

50% of any remaining live leaf and stem biomass at harvest (after biofuel removal, if any) is removed to the crop product pool to represent off-field uses such as use for animal feed and bedding. This value can be changed with the parameter \(crop\_residue\_removal\_frac\) (0–1). The default 50% is derived from Smerald et al. 2023, who found a global average of 50% of residues left on the field. This includes residues burned in the field, meaning that our implementation implictly assumes the CLM crop burning representation will handle those residues appropriately.

The following equations illustrate how this works. Subscript \(p\) refers to either the leaf or live stem biomass pool.

(2.26.9)\[ CF_{p,biofuel} = \left({CS_{p} \mathord{\left/ {\vphantom {CS_{p} \Delta t}} \right.} \Delta t} \right) * biofuel\_harvfrac\]

(2.26.10)\[ CF_{p,removed\_residue} = \left({CS_{p} \mathord{\left/ {\vphantom {CS_{p} \Delta t}} \right.} \Delta t} \right) * (1 - biofuel\_harvfrac) * crop\_residue\_removal\_frac\]

(2.26.11)\[ CF_{p,litter} = \left({CS_{p} \mathord{\left/ {\vphantom {CS_{p} \Delta t}} \right.} \Delta t} \right) * \left( 1-biofuel\_harvfrac \right) * \left( 1-crop\_residue\_removal\_frac \right) +CF_{p,alloc}\]

with corresponding nitrogen fluxes:

(2.26.12)\[ NF_{p,biofuel} = \left({NS_{p} \mathord{\left/ {\vphantom {NS_{p} \Delta t}} \right.} \Delta t} \right) * biofuel\_harvfrac\]

(2.26.13)\[ NF_{p,removed\_residue} = \left({NS_{p} \mathord{\left/ {\vphantom {NS_{p} \Delta t}} \right.} \Delta t} \right) * \left( 1 - biofuel\_harvfrac \right) * crop\_residue\_removal\_frac\]

(2.26.14)\[ NF_{p,litter} = \left({NS_{p} \mathord{\left/ {\vphantom {NS_{p} \Delta t}} \right.} \Delta t} \right) * \left( 1-biofuel\_harvfrac \right) * \left( 1-crop\_residue\_removal\_frac \right)\]

where CF is the carbon flux, CS is stored carbon, NF is the nitrogen flux, NS is stored nitrogen, and \(biofuel\_harvfrac\) is the harvested fraction of leaf/livestem for biofuel feedstocks.

Annual food crop yields (g dry matter m^-2) can be calculated by saving the GRAINC_TO_FOOD_ANN variable once per year, then postprocessing with Equation (2.26.15). This calculation assumes that grain C is 45% of the total dry weight. Additionally, harvest is not typically 100% efficient, so analysis needs to assume that harvest efficiency is less—we use 85%.

(2.26.15)\[ \text{Grain yield} = \frac{GRAINC\_TO\_FOOD\_ANN)*0.85}{0.45}\]

Table 2.26.3 Crop allocation parameters for the active crop plant functional types (PFTs) in CLM5BGCCROP. Numbers in the first row correspond to the list of PFTs in Table 2.26.1.

	temperate corn	spring wheat	temperate soybean	cotton	rice	sugarcane	tropical corn	tropical soybean	miscanthus	switchgrass
IVT	17, 18	19, 20	23, 24	41, 42	61, 62	67, 68	75, 76	77, 78	71, 72	73, 74
\(a_{leaf}^{i}\)	0.6	0.9	0.85	0.85	0.75	0.6	0.6	0.85	0.9	0.7
\({L}_{max}\) (m 2 m -2)	5	7	6	6	7	5	5	6	10	6.5
\(a_{froot}^{i}\)	0.1	0.05	0.2	0.2	0.1	0.1	0.1	0.2	0.11	0.14
\(a_{froot}^{f}\)	0.05	0	0.2	0.2	0	0.05	0.05	0.2	0.09	0.09
\(a_{leaf}^{f}\)	0	0	0	0	0	0	0	0	0	0
\(a_{livestem}^{f}\)	0	0.05	0.3	0.3	0.05	0	0	0.3	0	0
\(d_{L}\)	1.05	1.05	1.05	1.05	1.05	1.05	1.05	1.05	1.05	1.05
\(d_{alloc}^{stem}\)	2	1	5	5	1	2	2	5	2	2
\(d_{alloc}^{leaf}\)	5	3	2	2	3	5	5	2	5	5
\({CN}_{leaf}\)	25	20	20	20	20	25	25	20	25	25
\({CN}_{stem}\)	50	50	50	50	50	50	50	50	50	50
\({CN}_{froot}\)	42	42	42	42	42	42	42	42	42	42
\(CN^f_{leaf}\)	65	65	65	65	65	65	65	65	65	65
\(CN^f_{stem}\)	120	100	130	130	100	120	120	130	120	120
\(CN^f_{froot}\)	0	40	0	0	40	0	0	0	0	0
\({CN}_{grain}\)	50	50	50	50	50	50	50	50	50	50

Notes: Crop growth phases and corresponding variables are described throughout the text. \({CN}_{leaf}\), \({CN}_{stem}\), and \({CN}_{froot}\) are the target C:N ratios used during the leaf emergence phase (phase 2).

2.26.2.5. Other Features

2.26.2.5.1. Physical Crop Characteristics

Leaf area index (L) is calculated as a function of specific leaf area (SLA, Table 2.26.2) and leaf C. Stem area index (S) is equal to 0.1L for temperate and tropical corn, sugarcane, switchgrass, and miscanthus and 0.2L for other crops, as in AgroIBIS. All live C and N pools go to 0 after crop harvest, but the S is kept at 0.25 to simulate a post-harvest “stubble” on the ground.

Crop heights at the top and bottom of the canopy, \({z}_{top}\) and \({z}_{bot}\) (m), come from the AgroIBIS formulation:

(2.26.16)\[\begin{split}\begin{array}{l} {z_{top} =z_{top}^{\max } \left(\frac{L}{L_{\max } -1} \right)^{2} \ge 0.05{\rm \; where\; }\frac{L}{L_{\max } -1} \le 1} \\ {z_{bot} =0.02{\rm m}} \end{array}\end{split}\]

where \(z_{top}^{\max }\) is the maximum top-of-canopy height of the crop (Table 2.26.2) and \(L_{\max }\) is the maximum leaf area index (Table 2.26.3).

2.26.2.5.2. Interactive Fertilization

CLM simulates fertilization by adding nitrogen directly to the soil mineral nitrogen pool to meet crop nitrogen demands using both industrial fertilizer and manure application. CLM’s separate crop land unit ensures that natural vegetation will not access the fertilizer applied to crops. Fertilizer in CLM5BGCCROP is prescribed by crop functional types and varies spatially for each year based on the LUMIP land use and land cover change time series (LUH2 for historical and SSPs for future) (Lawrence et al. 2016). One of two fields is used to prescribe industrial fertilizer based on the type of simulation. For non-transient simulations, annual fertilizer application in g N/m²/yr is specified on the land surface data set by the field CONST_FERTNITRO_CFT. In transient simulations, annual fertilizer application is specified on the land use time series file by the field FERTNITRO_CFT, which is also in g N/m²/yr. The values for both of these fields come from the LUMIP time series for each year. In addition to the industrial fertilizer, background manure fertilizer is specified on the parameter file by the field manunitro. For perennial bioenergy crops, little fertilizer (56kg/ha/yr) is applied to switchgrass and no fertilizer is applied to Miscanthus. Note these rates are only based on local land management practices at the University of Illinois Energy Farm located in Central Midwestern United States (Cheng et al., 2019) rather than the LUMIP timeseries. For the current CLM5BGCCROP, manure N is applied at a rate of 0.002 kg N/m²/yr. Because previous versions of CLM (e.g., CLM4) had rapid denitrification rates, fertilizer is applied slowly to minimize N loss (primarily through denitrification) and maximize plant uptake. The current implementation of CLM5 inherits this legacy, although denitrification rates are slower in the current version of the model (Koven et al. 2013). As such, fertilizer application begins during the leaf emergence phase of crop development (phase 2) and continues for 20 days, which helps reduce large losses of nitrogen from leaching and denitrification during the early stage of crop development. The 20-day period is chosen as an optimization to limit fertilizer application to the emergence stage. A fertilizer counter in seconds, f, is set as soon as the leaf emergence phase for crops initiates:

(2.26.17)\[ f = n \times 86400\]

where n is set to 20 fertilizer application days and 86400 is the number of seconds per day. When the crop enters phase 2 (leaf emergence) of its growth cycle, fertilizer application begins by initializing fertilizer amount to the total fertilizer at each column within the grid cell divided by the initialized f. Fertilizer is applied and f is decremented each time step until a zero balance on the counter is reached.

2.26.2.5.3. Biological nitrogen fixation for soybeans

Biological N fixation for soybeans is calculated by the fixation and uptake of nitrogen module (Chapter 2.18) and is the same as N fixation in natural vegetation. Unlike natural vegetation, where a fraction of each PFT are N fixers, all soybeans are treated as N fixers.

2.26.2.5.4. Latitudinal variation in base growth tempereature

For most crops, \(GDD_{T_{{\rm 2m}} }\) (growing degree days since planting) is the same in all locations. However, for both rainfed and irrigated spring wheat and sugarcane, the calculation of \(GDD_{T_{{\rm 2m}} }\) allows for latitudinal variation:

(2.26.18)\[\begin{split}latitudinal\ variation\ in\ base\ T = \left\{ \begin{array}{lr} baset +12 - 0.4 \times latitude &\qquad 0 \le latitude \le 30 \\ baset +12 + 0.4 \times latitude &\qquad -30 \le latitude \le 0 \end{array} \right\}\end{split}\]

where \(baset\) is the base temperature for GDD (7^th row) in Table 2.26.2. Such latitudinal variation in base temperature could slow \(GDD_{T_{{\rm 2m}} }\) accumulation extend the growing season for regions within 30°S to 30°N for spring wheat and sugarcane.

2.26.2.5.5. Separate reproductive pool

One notable difference between natural vegetation and crops is the presence of reproductive carbon and nitrogen pools. Accounting for the reproductive pools helps determine whether crops are performing reasonably through yield calculations. The reproductive pool is maintained similarly to the leaf, stem, and fine root pools, but allocation of carbon and nitrogen does not begin until the grain fill stage of crop development. Equation (2.26.5) describes the carbon and nitrogen allocation coefficients to the reproductive pool. In CLM5BGCCROP, as allocation declines in stem, leaf, and root pools (see section 2.26.2.4.2) during the grain fill stage of growth, increasing amounts of carbon and nitrogen are available for grain development.

2.26.2.5.6. Tillage

Tillage is represented as an enhancement of the decomposition rate coefficient; see section 2.21.4.

2.26.3. The irrigation model

The CLM includes the option to irrigate cropland areas that are equipped for irrigation. The application of irrigation responds dynamically to the soil moisture conditions simulated by the CLM. This irrigation algorithm is based loosely on the implementation of Ozdogan et al. (2010).

When irrigation is enabled, the crop areas of each grid cell are divided into irrigated and rainfed fractions according to a dataset of areas equipped for irrigation (Portmann et al. 2010). Irrigated and rainfed crops are placed on separate soil columns, so that irrigation is only applied to the soil beneath irrigated crops.

In irrigated croplands, a check is made once per day to determine whether irrigation is required on that day. This check is made in the first time step after 6 AM local time. Irrigation is required if crop leaf area \(>\) 0, and the available soil water is below a specified threshold.

The soil moisture deficit \(D_{irrig}\) is

(2.26.19)\[\begin{split}D_{irrig} = \left\{ \begin{array}{lr} w_{target} - w_{avail} &\qquad w_{thresh} > w_{avail} \\ 0 &\qquad w_{thresh} \le w_{avail} \end{array} \right\}\end{split}\]

where \(w_{target}\) is the irrigation target soil moisture (mm)

(2.26.20)\[w_{target} = \sum_{j=1}^{N_{irr}} \theta_{target} \Delta z_{j} \ .\]

The irrigation moisture threshold (mm) is

(2.26.21)\[w_{thresh} = f_{thresh} \left(w_{target} - w_{wilt}\right) + w_{wilt}\]

where \(w_{wilt}\) is the wilting point soil moisture (mm)

(2.26.22)\[w_{wilt} = \sum_{j=1}^{N_{irr}} \theta_{wilt} \Delta z_{j} \ ,\]

and \(f_{thresh}\) is a tuning parameter. The available moisture in the soil (mm) is

(2.26.23)\[w_{avail} = \sum_{j=1}^{N_{irr}} \theta_{j} \Delta z_{j} \ ,\]

Note that \(w_{target}\) is truly supposed to give the target soil moisture value that we’re shooting for whenever irrigation happens; then the soil moisture deficit \(D_{irrig}\) gives the difference between this target value and the current soil moisture. The irrigation moisture threshold \(w_{thresh}\), on the other hand, gives a threshold at which we decide to do any irrigation at all. The way this is written allows for the possibility that one may not want to irrigate every time there becomes even a tiny soil moisture deficit. Instead, one may want to wait until the deficit is larger before initiating irrigation; at that point, one doesn’t want to just irrigate up to the “threshold” but instead up to the higher “target”. The target should always be greater than or equal to the threshold.

\(N_{irr}\) is the index of the soil layer corresponding to a specified depth \(z_{irrig}\) (Table 2.26.4) and \(\Delta z_{j}\) is the thickness of the soil layer in layer \(j\) (section 2.2.2). \(\theta_{j}\) is the volumetric soil moisture in layer \(j\) (section 2.7.3). \(\theta_{target}\) and \(\theta_{wilt}\) are the target and wilting point volumetric soil moisture values, respectively, and are determined by inverting (2.7.53) using soil matric potential parameters \(\Psi_{target}\) and \(\Psi_{wilt}\) (Table 2.26.4). After the soil moisture deficit \(D_{irrig}\) is calculated, irrigation in an amount equal to \(\frac{D_{irrig}}{T_{irrig}}\) (mm/s) is applied uniformly over the irrigation period \(T_{irrig}\) (s). Irrigation water is applied directly to the ground surface, bypassing canopy interception (i.e., added to \({q}_{grnd,liq}\): section 2.7.1).

To conserve mass, irrigation is removed from river water storage (Chapter 2.14). When river water storage is inadequate to meet irrigation demand, there are two options: 1) the additional water can be removed from the ocean model, or 2) the irrigation demand can be reduced such that river water storage is maintained above a specified threshold.

Table 2.26.4 Irrigation parameters

Parameter
\(f_{thresh}\)	1.0
\(z_{irrig}\) (m)	0.6
\(\Psi_{target}\) (mm)	-3400
\(\Psi_{wilt}\) (mm)	-150000