the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Relevance of feedbacks between water availability and crop systems using a coupled hydrology – crop growth model
Abstract. Individual hydrological and crop growth models often oversimplify underlying processes, reducing the accuracy of both simulated hydrology and crop growth dynamics. While crop models tend to generalize soil moisture processes, most hydrological models commonly use constant vegetation parameters and prescribed phenologies, neglecting the dynamic nature of crop growth. Despite some studies that have coupled hydrological and crop models, a limited understanding exists regarding the feedbacks between hydrology and crop growth. Our objective is to quantify the feedback between crop systems and hydrology on a fine-grained spatio-temporal level. To this end, the PCR-GLOBWB 2 hydrological model was coupled with the WOFOST crop growth model to quantify both the one-way and two-way interactions between hydrology and crop growth on a daily timestep and at 5 arc minutes (~10 km) resolution. Our study spans the Contiguous United States (CONUS) region and covers the period from 1979 to 2019, allowing a comprehensive evaluation of the feedback between hydrology and crop growth dynamics. We compare individual (stand-alone) as well as one-way and two-way coupled WOFOST and PCR-GLOBWB 2 model runs and evaluate the average crop yield and its interannual variability for rainfed and irrigated crops as well as simulated irrigation water withdrawal for maize, wheat and soybean. Our results reveal distinct patterns in the temporal and spatial variation of crop yield depending on the included interactions between hydrology and crop systems. Evaluating the model results against reported yield and water use data demonstrates the efficacy of the coupled framework in replicating observed irrigated and rainfed crop yields. Our results show that two-way coupling, with its dynamic feedback mechanisms, outperforms one-way coupling for rainfed crops. This improved performance stems from the feedback of WOFOST crop phenology to the crop parameters in the hydrological model. Our results suggest that when crop models are combined with hydrological models, a two-way coupling is needed to capture the impact of interannual climate variability on food production.
- Preprint
(2175 KB) - Metadata XML
-
Supplement
(5723 KB) - BibTeX
- EndNote
Status: open (until 07 Jul 2024)
-
RC1: 'Comment on egusphere-2024-465', Anonymous Referee #1, 11 Jun 2024
reply
Summary: This study couples a global hydrological model and a crop growth model in both one-way mode (hydrological model provides soil water storage to crop model) and two-way mode (additionally, crop model provides land surface fluxes, LAI, and rooting depth to hydrological model). This is a noteworthy technical achievement, and the developed code is openly shared. The chosen coupling approach is not discussed and justified much, and possible alternatives are not investigated and compared. Reported findings are largely dependent on modeling technicalities related to the chosen coupling approach. Moreover, it is unclear which specific phenomena / research questions require the use of a coupled hydrological / crop modeling approach. My recommendation would be to publish this as a technical note, either in HESS or a journal specialized on advances in environmental modeling and software after revisions.
Specific comments:
- How is reservoir release and inter-basin transfer handled in PCR-GLOBWB? From the text, it seems as if irrigation water availability is an exogenous boundary condition provided by PCR-GLOBWB. However, in reality, there is probably a feedback mechanism between crop state and reservoir management, i.e. system managers will be responsive to crop state and will adjust reservoir release and transfer decisions.
- Much discussion focuses on the importance of two-way coupling for the cropped areas, i.e. providing updated rooting depth, LAI and ET to PCR-GLOBWB, which are responsive to water stress, particularly on the rainfed areas. This leads to the question if such feedback mechanisms should not also be modelled on the non-agricultural portions of the landscape in PCR-GLOBWB?
- The chosen coupling approach through the variables of soil water storage and land-surface fluxes is postulated without much discussion of possible alternatives. Why not couple the two models through irrigation and percolation rates only, i.e. let WOFOST handle the soil moisture balance? Is this because phreatic ET could be relevant in some cropped system, i.e. you want to capture direct groundwater use of crops?
- Related to the previous comment: It would be valuable to explain on a more intuitive basis, what kind of feedback mechanisms should be investigated with this modelling system and why the chosen coupling approach is the most appropriate for these purposes. As far as I can see, crop yield responses to water and heat stress can be simulated in stand-alone mode with irrigation allocations provided by PCR-GLOBWB. Is it the influence of groundwater/phreatic ET that is in focus here?
- Figs 4/5, 6/7 and 9 – please add results from stand-alone simulations also for reference / comparison.
- Fig 9: I wonder why a logarithmic y-axis is used. It seems that differences between USGS data and models are a factor 10 or so… is that a good result?
- Figure 8: Please add 1:1 line for orientation.
- Modelling uncertainties: It would be very instructive if confidence bands could be added to some of the simulation results (e.g. Figures 2,3). Differences between simulations, especially between one-way coupling and two-way coupling, are often quite small and probably insignificant compared to overall modeling uncertainty.
Citation: https://doi.org/10.5194/egusphere-2024-465-RC1 -
RC2: 'Comment on egusphere-2024-465', Anonymous Referee #2, 16 Jun 2024
reply
This study presents a coupled hydrological and crop model that integrates PCR-GLOBWB 2 with WOFOST, effectively accounting for the interactions between hydrology and crop growth. The work focuses on how two-way interactions and feedback mechanisms between crop growth and hydrological systems would benefit the modeling. The authors show that this coupled framework can reproduce crop yields well, and highlight an improvement in performance when there is two-way coupling compared to one-way coupling. Overall, the study is clear and the model is open source. However, my main concern is that the paper focuses on presenting results from a specific coupling scheme without adequately discussing the broader implications or conducting a comprehensive sensitivity analysis of alternative schemes. While the paper is detailed as a technical report describing the operational aspects of the model, it falls short in exploring how different coupling schemes might affect our understanding of the feedbacks - a scientific point that the title suggests. Such a more thorough analysis would provide invaluable insights and greatly benefit future research on similar models.
Given the current scope of the paper, I recommend revising the title, abstract, and introduction to better reflect its focus on technical development (or specific reporting on a particular coupled model) rather than broader scientific (and methodological) discussions. Below are some specific suggestions for further improvement.
Specific comments:
1) WOFOST is a crop simulation data, I wonder how non-crop vegetation is considered in the model in the two-way coupling where phenologies are calculated by a crop simulation model. Please clarify.
2) L197: The meaning of "astro" here is not clear.
3) L390-406: It would be helpful to clarify why stand-alone WOFOST can produce a similar simulation with two-way coupling.
4) Section 3.2: If possible, the content can be merged with Section 3.1, and some metrics in Table 1 can be directly shown in Figure 2.
5) Figures 4 and 5, it's hard to see the difference between one-way and two-way from the color scales, and I don't think the difference is "notably" (line 471). I would suggest an additional visualization to better compare the simulated crop yields.Citation: https://doi.org/10.5194/egusphere-2024-465-RC2 -
RC3: 'Comment on egusphere-2024-465', Anonymous Referee #3, 23 Jun 2024
reply
This study analyzed the development of a coupled hydrology-crop model framework to investigate the intricate feedbacks between water availability and crop growth within the CONUS region focusing on maize, soybean, and wheat. The PCR-GLOBWB hydrological model was coupled with the WOFOST crop growth model to quantify both the one-way and two-way interactions. The result in the temporal and spatial variation of crop yield whether including interactions is interesting.
However, it is difficult to explain how the effect of one-way or two-way interaction is different as a result of discussing feedback, and for this reason, this is explained by the technical note. I recommend major revision after revising the title, abstract, and result to better introduce technical development of this study. This paper is a useful description in the hydrology and crop growth modeling technical area
Detailed comments are below.
- Overall, it is difficult to clearly identify the difference between one-way and two-way in result section. Can you discuss about the result from the difference between one-way and two-way method? It is difficult to discuss the feedback-based scientific point as to what feedback is due to the two-way interaction.
- How about showing additional difference figures between one-way coupled and two way coupled (Fig. 4- 7)?
- How about express Table 1 as a bar graphs like Fig.3? It would be easier to compare the differences if you provide the numbers with the figure
- For the convenience of the reader, it would be nice if you could refer to the figure subsection in the result section
- How about adding diagonal line to Fig. 8?
- In Figure 2, whether it is possible to compare the performance through interannual correlation with the reported field?
Citation: https://doi.org/10.5194/egusphere-2024-465-RC3 -
RC4: 'Comment on egusphere-2024-465', Anonymous Referee #4, 25 Jun 2024
reply
The work by Chevuru et al. developed a couple hydrology (PCR-GLOBWB 2) – crop growth (WOFOST) model and tested the model for the Contiguous United States (CONUS) region. The authors compared the one-way and two-way coupling schemes and found that the two-way coupling scheme outperformes one-way coupling scheme. In general, the idea (developing this model – coupling scheme) is relevant for the hydrological and crop modelling community, however, the results and description of the model need to be improved.
Main comments
The description of the two-way coupling scheme is not clear. A description of the study area, relevant information about the study area (especially for hydrological and crop modeling data), and model calibration/optimization technique are needed. The model results do not convince me, simulated crop yields (Figure 2) look very different with observed.
Detail comments:
Figure 2 could be improved for understanding the figure technically and the text L226-281 regrading the coupling schemes. I would be interested in seeing the figure showing the conceptual model PCR-GLOBWB 2 and WOFOST with their components, water fluxes, and exchange variables/fluxes between these two models. This would be very helpful for understanding the coupling schemes as well as for understanding the text.
L186: Where did the model get water for irrigation (from groundwater, river, or reservoir)?
L251: “…of the meteorological variables and…” change to “…of the meteorological variables at the current time step and…”?
L250-253: As I understand, this step is to calculate potential evapotranspiration (ETp) not actual evapotranspiration (ETa). So, where does actual Eta (L257) come from?
L267-268: “…is aggregated to the average value...” is it the summation or the average value? I think the we should pass the summation of soil moisture from the two layers of the PCR-GLOBWB 2 to the WOFOST model to ensure the amount of water is the same because the soil moisture is not reliable if the soil layer of two models have different depth and porosity.
L269:270: “WOFOST computes the actual transpiration…”. Was actual transpiration already calculated (including in the actual EVAPOtranspiration term in L257).
It is not clear to me which technique the authors used for model calibration/parameter optimization
Please show a figure and describe the study area somewhere before the Results section
L376:378: “our coupled PCR-GLOBWB 2 – WOFOST model framework simulated yields do not capture such trends, as the modelling approach intentionally omitted to incorporate trends in technology and management practices”: Please explain why? If the
L373-375: “This temporal evolution is primarily attributed to technological advancements, encompassing improved agricultural practices and the introduction of enhanced crop varieties over the study period” I did not see any trend from 1979-2007 (soybean and wheat yield in rainfed crops – Figure 2) but the model still cannot has a good match?
Citation: https://doi.org/10.5194/egusphere-2024-465-RC4
Viewed
HTML | XML | Total | Supplement | BibTeX | EndNote | |
---|---|---|---|---|---|---|
349 | 86 | 30 | 465 | 28 | 16 | 14 |
- HTML: 349
- PDF: 86
- XML: 30
- Total: 465
- Supplement: 28
- BibTeX: 16
- EndNote: 14
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1