the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 14 Jul 2023
Validation of a new global irrigation scheme in the land surface model ORCHIDEE v2.2
Abstract. Irrigation activities are important for sustaining food production, and account for 70 % of total global water withdrawals. In addition, due to increased evapotranspiration (ET) and changes on leaf area index (LAI), these activities have an impact on hydrology and climate. In this paper we present a new irrigation scheme within the land surface model ORCHIDEE. It restrains actual irrigation according to available freshwater by including a simple environmental limit and using allocation rules depending on local infrastructure. We perform a simple sensitivity analysis and parameter tuning to set the parameter values and match the observed irrigation amounts against reported values, assuming uniform parameter values over land. Our scheme matches irrigation withdrawals amounts at global scale, but we identify some areas in India, China and the US (some of the most intensively irrigated regions worldwide) where irrigation is underestimated. In all irrigated areas, the scheme reduces the negative bias of ET. It also exacerbates the positive bias of the leaf area index (LAI) except for the very intensively irrigated areas, where irrigation reduces a negative LAI bias. The increase of ET decreases river discharge values, in some cases significantly, although this does not necessarily lead to a better representation of discharge dynamics. Irrigation, however, does not have a large impact on the simulated total water storage anomalies (TWSA) and its trends. This may be partly explained by the absence of non-renewable groundwater use, and its inclusion could increase irrigation estimates in arid and semiarid regions by increasing the supply. Correlation of irrigation biases with landscape descriptors suggests that inclusion of irrigated rice and dam management could improve the irrigation estimates as well. Regardless of this complexity, our results show that the new irrigation scheme helps simulating acceptable land surface conditions and fluxes in irrigated areas, which is important to explore the joint evolution of climate, water resources and irrigation activities.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(23395 KB)
-
Supplement
(3654 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(23395 KB) - Metadata XML
-
Supplement
(3654 KB) - BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-1323', Anonymous Referee #1, 24 Aug 2023
Review comments on ‘Validation of a new global irrigation scheme in the land surface model ORCHIDEE v2.2” by Arboleda-Obando et al.
The authors present a new global irrigation scheme inside the ORCHIDEE land surface model. The irrigation model calculates the irrigation water demand based on soil moisture deficit against their target soil moisture after irrigation, and the irrigation rate is constrained by the available water supply from three major reservoirs (stream, overland, and groundwater). Irrigation model parameter beta that controls the irrigation target soil moisture was tuned to match some existing global irrigation estimates. Global-scale irrigation estimate from the model is comparable with other existing estimates, e.g., FAO’s AQUASTAT and Sacks et al. (2009), but its regional estimates show noticeable differences, particularly, underestimated irrigation in some irrigation hotspots in China, India and the US is notable. However, with the irrigation on, negative biases in ET over irrigated areas improved.
The new irrigation scheme shares common features with other some existing irrigation schemes that adopt similar concepts of adding irrigation water to soil up to a (tuneable) target value during the prescribed cropping seasons. But this work convincingly shows the importance of including irrigation scheme in global land surface (or hydrological) modelling to correctly reproduce evapotranspiration, which has important implications to relevant land surface processes and land-atmosphere interaction. Moreover, thanks to the explicit representations of irrigation water source, the authors argue the possible role of irrigation sourced from non-renewable groundwater storage in explaining the gap between modelled TWS and GRACE-derived TWS. The manuscript is well written and the topic is within the interests of EGUsphere’s readers. I recommend that the manuscript is considered for publication in EGUsphere once some technical concerns listed in the following section are addressed.
According to the description of irrigation scheme (Section 2.2), the soil moisture deific D is set to zero when crops and grasses are below a certain threshold value, LAI_lim. Although this might be a practical choice for the latter part of a crop growth cycle (maturity stage to harvest), irrigations from sowing the emergence stages, would be missed. Given that most crops require sufficient irrigation in the early stage of their growth cycle, this would lead to an underestimated irrigation overall.
In addition, it is stated in Section 2.2 that “we do not separate the irrigated area into a separate soil column, i.e. the soil column includes crops (both irrigated and rainfed) and grasses…The effective irrigation (I, see below) is uniformly applied over the crops
and grasses soil column.” This implies that if a fraction (<1) of crop/grass column is irrigated to meet the target soil moisture content (beta x field capacity), the added water is spread over the whole column, leading to the soil moisture content still under the targe soil moisture. This will in turn make the model add more water for irrigation until the entire column that contains fractional crops/grasses receives water up to the target soil moisture. Is this the case? The authors mention overestimated evapotranspiration as a possible result of the simplified water addition scheme, but in combination with the additional irrigation caused by the uniform spread of water to the entire crop/grass column, over-irrigation effect can be fairly significant, particularly when a crop/grass column represent sparse cropping area with a small crop fraction.
The new irrigation scheme add irrigation water to close to soil moisture deficit, the difference between the actual soil moisture and ‘beta x field capacity’, but the manuscript does not provide the soil moisture value that triggers irrigation (this is a different trigger than the LAI_min). Does this mean that irrigation is triggered whenever soil moisture drops below ‘beta x field capacity’? This would result in continuous irrigation over the whole duration of the cropping season (when LAI > LAI_min), leading to unrealistic emulation of irrigation (and likely overestimation of irrigation).
Specific Comments
Line 43: “…potential evapotranspiration PET, ET0…” -> (PET)?
Line 44-45: “This reference PET…parameter.” This sentence needs to be rewritten for clarity.
Line 125: “Tree” -> Three
Line 138: Correct “crop- grass soil column”
Line 156-158: For the reason described in the general comment section, I think ‘the fraction irrigated is greater than the crop/grass soil column’ would be less of a concern for correction irrigation simulation.
Line 166: What is the definition of ‘renewable-groundwater reservoirs’ in this work?
Line 193: “100x 100 km” -> “100 x 100 km”
Line 214-217: 3.0 x 10^6, 2.5 x 10^6. Does HID include LUHv2 or separate? It appears to be assumed in two different ways?
Line 221: “a-priori” -> a priroi
Line 320-321: This justifies close exam of threshold SM triggering irrigation and possible flaws in irrigation application to the entire crop/grass column with a fractional coverage, particularly when the fraction is small.
Line 341-342: Irr simulation result, 2452.5 k^3/y appears to be closer to the higher end of 3755-2465? Also results in Figure 4 indicate that the difference in global irrigation may not reflect that large continental/country scale difference.
Figure 3 caption: The caption describes what authors did with the beta vs. irr, but it does not properly describe what the figures are about.
Line 349-350: This is not a convincing explanation because Figure 4c shows noticeable contrasts between over- and under-irrigation within a country, for example in the US and India.
Line 358-359: Again, global annual irrigation may not be the best way to quantify errors in irr when they show strong biases with different signs between continents/countries.
Line 436-437: I am not sure what the model would do with a beta > 1. Since soil moisture would be max at the (effective) porosity, if beta x field capacity > porosity, would the irrgation scheme keep adding water every time step?
Line 445: “basedon” -> based on
Figure 9 inset text: Reduce the font size to make the whole text visible.
Citation: https://doi.org/10.5194/egusphere-2023-1323-RC1 - AC1: 'Reply on RC1', Pedro Arboleda, 30 Oct 2023
-
RC2: 'Comment on egusphere-2023-1323', Anonymous Referee #2, 02 Sep 2023
Review of “Validation of a new global irrigation scheme in the land surface model ORCHIDEE v2.2” by Arboleda-Obando et al. for GMD
The authors improved an irrigation scheme in LSM ORCHIDEE and evaluated the improvement. Using reported irrigation statistics for the year 2000, they globally-uniformly tuned key parameters of the irrigation scheme used in ORCHIDEE to achieve a better balance in estimating the global total irrigation volume and spatially minimizing irrigation bias. It is also investigated how each of these parameters can change the irrigation estimate. In addition to modifying their irrigation scheme, this study shows how much irrigation can affect simulations on hydrological processes in terms of several hydrological variables: evapotranspiration, leaf area index, river discharge and total water storage. In addition, their factor analyses indicate potential research directions to further improve the ORCHIDEE irrigation model, such as the explicit inclusion of paddy rice.
Such model improvement is essential for a better understanding of land surface processes in Earth system science. Considering the fact that human activities have influenced the Earth system, irrigation should also be a critical component to be further investigated. I understand that this is an important step for ORCHIDEE. However, I have some major concerns that the authors need to address.
< Major comments >
(1) Why do the authors insist on globally uniform parameters (tuning)? While better estimation of total global irrigation withdrawals is an important challenge, better estimation of irrigation in heavily irrigated regions should also be a priority in a global study. The results show that this irrigation water estimate is relatively small compared to other irrigation estimates (Section5.1), and this should be related to the underestimation of irrigation volume in heavily irrigated countries (Fig4b-c). On the other hand, the globally uniform parameter tuning reduced irrigation volume to exacerbate the underestimation in these regions (Fig5c-1, Fig9). The authors also state that this is a drawback (Line.331). Therefore, I wonder if this tuning is sufficient to improve the simulation skill of ORCHIDEE. Since the authors already present spatially varying beta value, which is a key tuned parameter, to minimize the irrigation bias in this study (Fig9), I wonder why the authors did not apply this spatially varying parameter to estimate the main irrigation estimate. I assume that there are reasons (perhaps, related to modeling philosophy) for this decision to apply globally uniform parameters and their tuning. If so, I expect the authors to clarify their thoughts in an earlier part of this manuscript.
(2) Another concern related to the parameter tuning is the reference year, 2000. Given the spatiotemporal uncertainty in the meteorological forcing data (even with reanalysis-based forcing data), I wonder if the single reference year allows the authors to robustly tune the parameter. I require an explanation or methodological modification in this regard.
(3) I would like the authors to revisit irrigation efficiency and describe in more detail how they account for this factor in their irrigation estimate. I may be wrong, but as far as I have read Section 2.2, evaporative, infiltration, and seepage losses during conveyance, distribution, and application processes, are not considered in the calculation of irrigation water withdrawal from irrigation requirement. Other models that use soil moisture target methods generally consider irrigation efficiency (such as CLM5, LPJmL, H08, and HiGWMAT etc.). Although irrigation efficiency has a large uncertainty, if irrigation efficiency is not considered, irrigation withdrawal cannot be properly estimated from irrigation requirement. Thus, the relatively smaller global total irrigation volume estimated in this study may be related to this point. (Note that this is the different irrigation efficiency defined in line316.)
(4) Could you add a description about the croup calendar? It should be explained how the authors defined the irrigation period. The authors mention that they did not consider double cropping, but there does not seem to be any explanation about the crop calendar in the current manuscript.
(5) How does ORCHIDEE define “renewable”-groundwater resource?
(6) All figures are blurred. It seems that dpi needs to be higher.
(7) Regarding Fig4b-c, could you provide (supplementary) figures in % compared to Sackes et al. 2009? It is difficult to see how the irrigation bias is critical compared to the reference values.
< Minor comments >
Line 43: PET needs to be spelled out here.
Line 44: If I remember correctly, H08 applies the soil moisture target method.
Line 60: In this context, GHM should also be included.
Table1: Probably, Ai should be ai.
Line 271: It would be better to explain the original spatial resolution of the observed data in Section 3.2.
Fig5a: Could it be possible to add a reference plot(s) (AQUASTAT or other models’ estimate)?
Line 407, “we observe higher peaks and low values In Huang He when irrigation is activated”: I can not understand which “low values” is about. Could you rephrase this?
Line 418-420: I could not understand the point of this sentence in my first reading. Could you exemplify basins in Fig8 in this sentence?
Fig 6: Add x-axis label.
Line 452: Hanasaki et al. 2008b seems to be a wrong reference here because H08 uses groundwater resource when surface water availability is not sufficient to meet irrigation demand.
Line 459, “… like topography and environmental flow”: Refer Hanasaki et al. 2018 here.
Line 461: The following models also include detailed irrigation schemes: doi:10.5194/hess-19-3073-2015, doi:10.1029/2022MS003074.
Citation: https://doi.org/10.5194/egusphere-2023-1323-RC2 - AC2: 'Reply on RC2', Pedro Arboleda, 30 Oct 2023
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-1323', Anonymous Referee #1, 24 Aug 2023
Review comments on ‘Validation of a new global irrigation scheme in the land surface model ORCHIDEE v2.2” by Arboleda-Obando et al.
The authors present a new global irrigation scheme inside the ORCHIDEE land surface model. The irrigation model calculates the irrigation water demand based on soil moisture deficit against their target soil moisture after irrigation, and the irrigation rate is constrained by the available water supply from three major reservoirs (stream, overland, and groundwater). Irrigation model parameter beta that controls the irrigation target soil moisture was tuned to match some existing global irrigation estimates. Global-scale irrigation estimate from the model is comparable with other existing estimates, e.g., FAO’s AQUASTAT and Sacks et al. (2009), but its regional estimates show noticeable differences, particularly, underestimated irrigation in some irrigation hotspots in China, India and the US is notable. However, with the irrigation on, negative biases in ET over irrigated areas improved.
The new irrigation scheme shares common features with other some existing irrigation schemes that adopt similar concepts of adding irrigation water to soil up to a (tuneable) target value during the prescribed cropping seasons. But this work convincingly shows the importance of including irrigation scheme in global land surface (or hydrological) modelling to correctly reproduce evapotranspiration, which has important implications to relevant land surface processes and land-atmosphere interaction. Moreover, thanks to the explicit representations of irrigation water source, the authors argue the possible role of irrigation sourced from non-renewable groundwater storage in explaining the gap between modelled TWS and GRACE-derived TWS. The manuscript is well written and the topic is within the interests of EGUsphere’s readers. I recommend that the manuscript is considered for publication in EGUsphere once some technical concerns listed in the following section are addressed.
According to the description of irrigation scheme (Section 2.2), the soil moisture deific D is set to zero when crops and grasses are below a certain threshold value, LAI_lim. Although this might be a practical choice for the latter part of a crop growth cycle (maturity stage to harvest), irrigations from sowing the emergence stages, would be missed. Given that most crops require sufficient irrigation in the early stage of their growth cycle, this would lead to an underestimated irrigation overall.
In addition, it is stated in Section 2.2 that “we do not separate the irrigated area into a separate soil column, i.e. the soil column includes crops (both irrigated and rainfed) and grasses…The effective irrigation (I, see below) is uniformly applied over the crops
and grasses soil column.” This implies that if a fraction (<1) of crop/grass column is irrigated to meet the target soil moisture content (beta x field capacity), the added water is spread over the whole column, leading to the soil moisture content still under the targe soil moisture. This will in turn make the model add more water for irrigation until the entire column that contains fractional crops/grasses receives water up to the target soil moisture. Is this the case? The authors mention overestimated evapotranspiration as a possible result of the simplified water addition scheme, but in combination with the additional irrigation caused by the uniform spread of water to the entire crop/grass column, over-irrigation effect can be fairly significant, particularly when a crop/grass column represent sparse cropping area with a small crop fraction.
The new irrigation scheme add irrigation water to close to soil moisture deficit, the difference between the actual soil moisture and ‘beta x field capacity’, but the manuscript does not provide the soil moisture value that triggers irrigation (this is a different trigger than the LAI_min). Does this mean that irrigation is triggered whenever soil moisture drops below ‘beta x field capacity’? This would result in continuous irrigation over the whole duration of the cropping season (when LAI > LAI_min), leading to unrealistic emulation of irrigation (and likely overestimation of irrigation).
Specific Comments
Line 43: “…potential evapotranspiration PET, ET0…” -> (PET)?
Line 44-45: “This reference PET…parameter.” This sentence needs to be rewritten for clarity.
Line 125: “Tree” -> Three
Line 138: Correct “crop- grass soil column”
Line 156-158: For the reason described in the general comment section, I think ‘the fraction irrigated is greater than the crop/grass soil column’ would be less of a concern for correction irrigation simulation.
Line 166: What is the definition of ‘renewable-groundwater reservoirs’ in this work?
Line 193: “100x 100 km” -> “100 x 100 km”
Line 214-217: 3.0 x 10^6, 2.5 x 10^6. Does HID include LUHv2 or separate? It appears to be assumed in two different ways?
Line 221: “a-priori” -> a priroi
Line 320-321: This justifies close exam of threshold SM triggering irrigation and possible flaws in irrigation application to the entire crop/grass column with a fractional coverage, particularly when the fraction is small.
Line 341-342: Irr simulation result, 2452.5 k^3/y appears to be closer to the higher end of 3755-2465? Also results in Figure 4 indicate that the difference in global irrigation may not reflect that large continental/country scale difference.
Figure 3 caption: The caption describes what authors did with the beta vs. irr, but it does not properly describe what the figures are about.
Line 349-350: This is not a convincing explanation because Figure 4c shows noticeable contrasts between over- and under-irrigation within a country, for example in the US and India.
Line 358-359: Again, global annual irrigation may not be the best way to quantify errors in irr when they show strong biases with different signs between continents/countries.
Line 436-437: I am not sure what the model would do with a beta > 1. Since soil moisture would be max at the (effective) porosity, if beta x field capacity > porosity, would the irrgation scheme keep adding water every time step?
Line 445: “basedon” -> based on
Figure 9 inset text: Reduce the font size to make the whole text visible.
Citation: https://doi.org/10.5194/egusphere-2023-1323-RC1 - AC1: 'Reply on RC1', Pedro Arboleda, 30 Oct 2023
-
RC2: 'Comment on egusphere-2023-1323', Anonymous Referee #2, 02 Sep 2023
Review of “Validation of a new global irrigation scheme in the land surface model ORCHIDEE v2.2” by Arboleda-Obando et al. for GMD
The authors improved an irrigation scheme in LSM ORCHIDEE and evaluated the improvement. Using reported irrigation statistics for the year 2000, they globally-uniformly tuned key parameters of the irrigation scheme used in ORCHIDEE to achieve a better balance in estimating the global total irrigation volume and spatially minimizing irrigation bias. It is also investigated how each of these parameters can change the irrigation estimate. In addition to modifying their irrigation scheme, this study shows how much irrigation can affect simulations on hydrological processes in terms of several hydrological variables: evapotranspiration, leaf area index, river discharge and total water storage. In addition, their factor analyses indicate potential research directions to further improve the ORCHIDEE irrigation model, such as the explicit inclusion of paddy rice.
Such model improvement is essential for a better understanding of land surface processes in Earth system science. Considering the fact that human activities have influenced the Earth system, irrigation should also be a critical component to be further investigated. I understand that this is an important step for ORCHIDEE. However, I have some major concerns that the authors need to address.
< Major comments >
(1) Why do the authors insist on globally uniform parameters (tuning)? While better estimation of total global irrigation withdrawals is an important challenge, better estimation of irrigation in heavily irrigated regions should also be a priority in a global study. The results show that this irrigation water estimate is relatively small compared to other irrigation estimates (Section5.1), and this should be related to the underestimation of irrigation volume in heavily irrigated countries (Fig4b-c). On the other hand, the globally uniform parameter tuning reduced irrigation volume to exacerbate the underestimation in these regions (Fig5c-1, Fig9). The authors also state that this is a drawback (Line.331). Therefore, I wonder if this tuning is sufficient to improve the simulation skill of ORCHIDEE. Since the authors already present spatially varying beta value, which is a key tuned parameter, to minimize the irrigation bias in this study (Fig9), I wonder why the authors did not apply this spatially varying parameter to estimate the main irrigation estimate. I assume that there are reasons (perhaps, related to modeling philosophy) for this decision to apply globally uniform parameters and their tuning. If so, I expect the authors to clarify their thoughts in an earlier part of this manuscript.
(2) Another concern related to the parameter tuning is the reference year, 2000. Given the spatiotemporal uncertainty in the meteorological forcing data (even with reanalysis-based forcing data), I wonder if the single reference year allows the authors to robustly tune the parameter. I require an explanation or methodological modification in this regard.
(3) I would like the authors to revisit irrigation efficiency and describe in more detail how they account for this factor in their irrigation estimate. I may be wrong, but as far as I have read Section 2.2, evaporative, infiltration, and seepage losses during conveyance, distribution, and application processes, are not considered in the calculation of irrigation water withdrawal from irrigation requirement. Other models that use soil moisture target methods generally consider irrigation efficiency (such as CLM5, LPJmL, H08, and HiGWMAT etc.). Although irrigation efficiency has a large uncertainty, if irrigation efficiency is not considered, irrigation withdrawal cannot be properly estimated from irrigation requirement. Thus, the relatively smaller global total irrigation volume estimated in this study may be related to this point. (Note that this is the different irrigation efficiency defined in line316.)
(4) Could you add a description about the croup calendar? It should be explained how the authors defined the irrigation period. The authors mention that they did not consider double cropping, but there does not seem to be any explanation about the crop calendar in the current manuscript.
(5) How does ORCHIDEE define “renewable”-groundwater resource?
(6) All figures are blurred. It seems that dpi needs to be higher.
(7) Regarding Fig4b-c, could you provide (supplementary) figures in % compared to Sackes et al. 2009? It is difficult to see how the irrigation bias is critical compared to the reference values.
< Minor comments >
Line 43: PET needs to be spelled out here.
Line 44: If I remember correctly, H08 applies the soil moisture target method.
Line 60: In this context, GHM should also be included.
Table1: Probably, Ai should be ai.
Line 271: It would be better to explain the original spatial resolution of the observed data in Section 3.2.
Fig5a: Could it be possible to add a reference plot(s) (AQUASTAT or other models’ estimate)?
Line 407, “we observe higher peaks and low values In Huang He when irrigation is activated”: I can not understand which “low values” is about. Could you rephrase this?
Line 418-420: I could not understand the point of this sentence in my first reading. Could you exemplify basins in Fig8 in this sentence?
Fig 6: Add x-axis label.
Line 452: Hanasaki et al. 2008b seems to be a wrong reference here because H08 uses groundwater resource when surface water availability is not sufficient to meet irrigation demand.
Line 459, “… like topography and environmental flow”: Refer Hanasaki et al. 2018 here.
Line 461: The following models also include detailed irrigation schemes: doi:10.5194/hess-19-3073-2015, doi:10.1029/2022MS003074.
Citation: https://doi.org/10.5194/egusphere-2023-1323-RC2 - AC2: 'Reply on RC2', Pedro Arboleda, 30 Oct 2023
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 390 | 174 | 19 | 583 | 40 | 15 | 16 |
- HTML: 390
- PDF: 174
- XML: 19
- Total: 583
- Supplement: 40
- BibTeX: 15
- EndNote: 16
Viewed (geographical distribution)
| Country | # | Views | % |
|---|---|---|---|
| United States of America | 1 | 153 | 26 |
| France | 2 | 118 | 20 |
| Germany | 3 | 57 | 10 |
| China | 4 | 43 | 7 |
| Australia | 5 | 19 | 3 |
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
- 153
Pedro Felipe Arboleda-Obando
Agnès Ducharne
Philippe Ciais
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(23395 KB) - Metadata XML
-
Supplement
(3654 KB) - BibTeX
- EndNote
- Final revised paper