Simulation of crop yield using the global hydrological model H08 (crp.v1)
Abstract. Food and water are essential for life. A better understanding of the food–water nexus requires the development of an integrated model that can simultaneously simulate food production and the requirements and availability of water resources. H08 is a global hydrological model that considers human water use and management (e.g., reservoir operation and crop irrigation). Although a crop growth sub-model has been included in H08 to estimate the global crop-specific calendar, its performance as a yield simulator is poor, mainly because a globally uniform parameter set was used for each crop type. Here, through country-wise parameter calibration and algorithm improvement, we enhanced H08 to simulate the yields of four major staple crops: maize, wheat, rice, and soybean. The simulated crop yield was compared with the Food and Agriculture Organization (FAO) national yield statistics and the global data set of historical yield for major crops (GDHY) gridded yield estimates with respect to mean bias (across nations) and time series correlation (for individual nations). The improved simulations showed good consistency with FAO national yield. The mean biases of the major producer countries were considerably reduced to −4 %, 3 %, −1 %, and 1 % for maize, wheat, rice, and soybean, respectively. The corresponding coefficients of determination (R2) of the simulated and FAO statistical yield increased from 0.01 to 0.98, 0.21 to 0.99, 0.06 to 0.99, and 0.14 to 0.97 for maize, wheat, rice, and soybean, respectively; the corresponding root mean square error (RMSE) decreased from 7.1 to 1.1, 2.2 to 0.6, 2.7 to 0.5, 2.3 to 0.3 t/ha. Comparison with the reported performances of other mainstream global crop models revealed that our improved simulations have comparable ability to capture the temporal yield variability. The grid-level analysis showed that the improved simulations had similar capacity to GDHY yield, in terms of reproducing the temporal variation over a wide area, although substantial differences were observed in other places. Using the improved model, we confirmed that an earlier study on quantifying the contributions of irrigation on global food production can be reasonably reproduced. Overall, our improvements enabled H08 to estimate crop production and hydrology in a single framework, which will be beneficial for global food–water–land–energy nexus studies.
Zhipin Ai and Naota Hanasaki
Zhipin Ai and Naota Hanasaki
Model code and software
H08 (crp.v1) https://zenodo.org/record/7344809#.Y4APSbJBzjB
Zhipin Ai and Naota Hanasaki
Viewed (geographical distribution)
in their submitted paper 'Simulation of crop yield using the global hydrological model H08 (crp.v1)', the authors enhance the H08 crop sub-model with parameter calibration and algorithm improvement.
Thereby, the CO2 fertilization effect and the effect of vapor pressure deficit change has been included to the model. Additionally, a model calibration has been applied.
In order to evaluate the model results, simulated yields are compared with statistical yields and other global crop models for the major 4 crops (maize, wheat, rice, soybean) at country and grid-level.
The paper is well written and understandable. Nevertheless, the paper has a main weakness:
If you calibrate your model towards yields that you also use to validate/evaluate your model, it's not a surprise that R2 is >0.99.
But does that mean that your model improved? I wouldn't say so. It just says that the calibration was successful.
First, I'd suggest to say 'calibrated simulations' and 'default simulations' instead of 'improved' and 'default' simulations throughout the manuscript.
Second, given the fact that you added the effects of CO2 and vapor pressure deficit to the H08 model, it would be interesting in this study to quantify the difference between considering these effects and not.
As in Deryng et al. (2016) I would encourage you to quantify the difference of CO2 effect on crop water productivity for C3 and C4 crops.
A calibration can be done in a next step but after the validation. The quantification of water flows would require to validate crop evapotranspiration, which is not done in this study.
Given the contextual and structural deficits in this study, I'd suggest major revisions.
Abstract ln 1: 'Food and water are essential for life'. To me, this is trivial and a bit pathethic.
Abstract ln 19: What means 'reasonably'? Can you quantify that with a statistical value?
Line 35: You could add the PROMET model to this list of models, because it is also a hydrological model with an enhanced crop growth module included that has been applied at global (Zabel et al. 2019) and regional scale (Degife et al. 2021).
Line 76: Is it still a 'process-based model', after calibrating parameters to match statistical yields that are subsequently used for model evaluation?
Line 125: Nitrogen and phosphorous stress are implicitly considered in your calibration procedure! It possibly one of the main factors that influences your calibration.
Line 224: Two full stops.
Line 222-231: See above comment on calibration and validation.
Line 233-239: Interannual yield variabilities are much higher in both, calibrated and default simulations than in observations. Can you explain why?
Line 243: According to the GGCMI phase 3 protocol, none of the models in Jägermeyr et al. (2021) are calibrated to yields.
Line 295: Same problem than with constant irrigation occurs for the crop calendar, I guess?
Line 305: What means 'remain good references'? Don't you contradict yourself with what is said before? Of course, we need better data and this is a strong limitation.
Line 307: 'factors' or better say 'processes' here.
Line 317: Therefore, it would be required to validate simulated crop evapotranspiration. If you can reproduce yields, it does not mean that evapotranspiration is simulated correctly.
Line 331: You just mentioned that irrigated areas are kept constant over time, which impacts the results. Another question is the amount of irrigation that is applied. Do you consider only full irrigation, or do you also consider deficit irrigation?
Line 336: As I understood, H08 had the capacity to simulate yields before. You added two processes and applied a calibration.
Line 343: Please avoid qualitative statements like 'a good tool'. Nobody can say what is a good tool.
Figure 1: Font size too small. Country names, etc. are very hard to read, even after zooming in.
Figure 3: Interesting to see R and RMSE values as a bar plot, which is not intuitive. I'd suggest to show a Taylor diagram or a scatterplot.
Deryng, D., Elliott, J., Folberth, C. et al. Regional disparities in the beneficial effects of rising CO2 concentrations on crop water productivity. Nature Clim Change 6, 786–790 (2016). https://doi.org/10.1038/nclimate2995
Degife, A. W., Zabel, F., Mauser, W. (2021): Climate change impacts on potential maize yields in Gambella region, Ethiopia. Regional Environmental Change, 21:60. doi: 10.1007/s10113-021-01773-3.
Zabel, F., Delzeit, R., Schneider, J. M., Seppelt, R., Mauser, M., Václacík, T. (2019): Global impacts of future cropland expansion and intensification on agricultural markets and biodiversity. Nature Communications, 10:2844, 11.