On the gap between crop and land surface models: comparing irrigation and other land surface estimates from AquaCrop and Noah-MP over the Po Valley

Busschaert, Louise; Bechtold, Michel; Modanesi, Sara; Massari, Christian; Raes, Dirk; Kumar, Sujay V.; De Lannoy, Gabriëlle J. M.

doi:https://doi.org/10.5194/egusphere-2025-2550

Preprints

https://doi.org/10.5194/egusphere-2025-2550

Preprints

01 Jul 2025

| 01 Jul 2025

Status: this preprint is open for discussion and under review for Hydrology and Earth System Sciences (HESS).

On the gap between crop and land surface models: comparing irrigation and other land surface estimates from AquaCrop and Noah-MP over the Po Valley

Louise Busschaert, Michel Bechtold, Sara Modanesi, Christian Massari, Dirk Raes, Sujay V. Kumar, and Gabriëlle J. M. De Lannoy

Abstract. Land surface and crop models both simulate irrigation, but they differ in their approaches, primarily because they were originally developed for distinct purposes and scales. Through an example case study in a highly irrigated region, this research helps to better understand the gap between these models and the complexity of irrigation modeling. More specifically, irrigation was estimated over the Po Valley (Italy) at a 1-km² spatial resolution using (i) a crop model, AquaCrop, and (ii) a land surface model, Noah-MP. Both models were run with sprinkler irrigation using a similar setup within NASA's Land Information System. Irrigation estimates were evaluated at the pixel and basin scale, using in situ and satellite-based reference data. In addition, surface soil moisture (SSM), vegetation, and evapotranspiration (ET) estimates were compared with satellite retrievals.

Noah-MP has on average higher annual irrigation rates (434 mm yr^-1) than AquaCrop (268 mm yr^-1), mainly because Noah-MP simulates more irrigation water losses (not consumed by transpiration) via runoff, interception, and soil evaporative losses, whereas AquaCrop only accounts for soil evaporative losses. When adding representative application water losses to irrigation estimates from AquaCrop, and conveyance water losses to the estimates from both models, the irrigation estimates from both models fall within reported ranges of 500–600 mm yr^-1. For the field-based evaluation, Noah-MP presents large irrigation events (> 100 mm per event) and less interannual variability than AquaCrop. Two-week averaged SSM estimates from both models agree well with downscaled estimates from the Soil Moisture Active Passive (SMAP) mission, with spatially averaged unbiased root mean square differences of 0.05 and 0.04 m³ m^-3 for AquaCrop and Noah-MP, respectively. Both models show limitations in terms of vegetation and ET modeling, mainly due to simplistic vegetation modules and suboptimal parameterization in both models. The results highlight the complexity of irrigation modeling due to its anthropogenic nature, and also show the need for better observations to validate and guide model estimates: reference irrigation data are sparse and satellite retrievals under irrigated conditions are quite uncertain.

Received: 30 May 2025 – Discussion started: 01 Jul 2025

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.

Download & links

Louise Busschaert, Michel Bechtold, Sara Modanesi, Christian Massari, Dirk Raes, Sujay V. Kumar, and Gabriëlle J. M. De Lannoy

Status: open (until 21 Nov 2025)

Post a comment Subscribe to comment alert

CC1:
'Comment on egusphere-2025-2550', Nima Zafarmomen, 05 Jul 2025 reply
The preprint "On the gap between crop and land surface models: comparing irrigation and other land surface estimates from AquaCrop and Noah-MP over the Po Valley" by Busschaert et al. (2025) compares two distinct modeling approaches—crop model (AquaCrop) and land surface model (Noah-MP)—for estimating irrigation, soil moisture, vegetation, and evapotranspiration (ET) in the highly irrigated Po Valley, Italy. Using a 1-km² resolution within NASA’s Land Information System, the study evaluates model performance against in situ and satellite-based data, highlighting differences in irrigation estimates (Noah-MP: 434 mm yr⁻¹, AquaCrop: 268 mm yr⁻¹) due to varying treatments of water losses, vegetation dynamics, and soil parameters.
The study highlights that Noah-MP simulates higher irrigation rates (434 mm yr⁻¹) than AquaCrop (268 mm yr⁻¹) due to differences in accounting for losses (e.g., runoff, interception, and soil evaporation). However, the manuscript could better clarify how these losses are quantified and modeled in each system. For instance, the absence of canopy interception in AquaCrop is noted, but the implications for irrigation efficiency and model performance are not thoroughly discussed.

The study uses default soil hydraulic parameters (SHPs) for both models, which differ significantly (e.g., Noah-MP’s higher total available water, TAW). While Section 4.2 acknowledges this limitation, the manuscript lacks a sensitivity analysis to assess how variations in SHPs (e.g., θ_FC, θ_WP) impact irrigation estimates.

The comparison with satellite-based irrigation estimates from Dari et al. (2023) shows discrepancies, particularly in interannual variability (Section 3.1.2). The manuscript suggests potential inaccuracies in the satellite data but does not explore this further or provide evidence to support this claim.

Both models show significant overestimation of vegetation (DMP) compared to CGLS data, attributed to simplistic vegetation modules and suboptimal parameterization (Section 3.1.3). The manuscript could improve by discussing specific parameterization choices (e.g., maximum canopy cover in AquaCrop, vegetation module options in Noah-MP) and their impact on results.

While the authors briefly mention data assimilation (DA) as a potential approach for integrating observations (e.g., Abolafia-Rosenzweig et al., 2019; Busschaert et al., 2024; Igder et al., 2022; Maina et al., 2024; Modanesi et al., 2022; Nie et al., 2022), the discussion could benefit from the inclusion of additional relevant studies. For example, the work titled “Assimilation of Sentinel-Based Leaf Area Index for Modeling Surface-Ground Water Interactions in Irrigation Districts” presents a valuable application of DA in a coupled surface–subsurface hydrological context, specifically within agricultural settings. Including such studies would help strengthen the section by illustrating the diversity of DA techniques and their relevance to integrated hydrologic modeling in real-world systems.

The study finds that irrigation estimates correlate better with in situ data at longer temporal aggregations (e.g., monthly, two-monthly; Section 3.2). However, the manuscript does not adequately address how this aggregation might mask short-term inaccuracies in irrigation timing, which is critical for real-world applications.

The models rely on MERRA2 meteorological forcings, downscaled via bilinear interpolation (Section 2.2.1). The manuscript does not address potential uncertainties introduced by this downscaling or the coarse resolution of MERRA2 (0.5° × 0.625°). A brief analysis or reference to studies evaluating MERRA2’s accuracy in the Po Valley would enhance confidence in the input data.

The study excludes paddy rice areas and clay soils due to AquaCrop’s limitations in simulating sprinkler irrigation on low-conductivity soils (Section 2.1). This exclusion reduces the study’s scope, as rice paddies are significant in the Po Valley. The manuscript should discuss the implications of this exclusion for regional irrigation estimates and propose how future studies could incorporate these areas (e.g., by adapting AquaCrop for flood irrigation).

Reply
Citation: https://doi.org/10.5194/egusphere-2025-2550-CC1
- CC2: '1. Reply on CC1: 'Comment on egusphere-2025-2550', Nima Zafarmomen -- part 1', Gabriëlle De Lannoy, 11 Sep 2025 reply
  
  Thank you for the comments. For completeness, we first upload our earlier responses (to comments on our initial submission) here, and then provide new responses in part 2 of our reply.
  
  Reply
  
  Citation: https://doi.org/10.5194/egusphere-2025-2550-CC2
- CC3: '2. Reply on CC1: 'Comment on egusphere-2025-2550', Nima Zafarmomen -- part 2', Gabriëlle De Lannoy, 11 Sep 2025 reply
  
  We thank the reviewer for the comments, and we provide our responses and some questions to the reviewer in the pdf attached hereto.
  
  Reply
  
  Citation: https://doi.org/10.5194/egusphere-2025-2550-CC3
RC1:
'Comment on egusphere-2025-2550', Anonymous Referee #1, 20 Sep 2025 reply
General Assessment
The manuscript is well written, logically structured, and makes an important contribution by comparing AquaCrop (a crop model) and Noah-MP (a land surface model, LSM). This work effectively bridges the crop water and land surface modeling communities, providing valuable insights into irrigation, soil moisture, vegetation, and ET representation at both basin and field scales. The use of multiple observational and satellite products adds robustness. However, some methodological and interpretational aspects require clarification and further depth.
Major Comments
1. Section 4.4 – Uncertainties in satellite-based retrievals: The treatment of uncertainties is currently too brief. Methodological uncertainties should be discussed in more detail:
Soil moisture representation: The use of surface soil moisture (SSM) is problematic for irrigation studies, since what is required is modeling of root-zone soil moisture (RZSM) and the management allowable depletion (MA) in irrigation modules to trigger irrigation. This mismatch and the limitation of SSM in representing RZSM should be explicitly acknowledged.

Downscaling errors: Both SenET-derived ET and downscaled SMAP SSM rely on downscaling methods, which propagate errors into the reference datasets and thus into the model evaluation. It may be more consistent to use an ET reference dataset at a coarser resolution that avoids such downscaling errors.

Implications: These uncertainties influence not only irrigation assessment but also the reliability of the ET and SSM benchmarks used in the study. A deeper discussion would strengthen the credibility of the conclusions.

2. Temporal aggregation of evaluation metrics: The aggregation of daily values into 15-day means is understandable to reduce the impact of uncertain irrigation timing, but it introduces serious limitations:
Loss of dynamics: Short-term variability in ET and SSM, especially during crop water stress, is smoothed out. This may reduce the evaluation to potential ET comparisons.

Bias masking: Aggregation may artificially improve correlations while masking systematic model deficiencies in simulating fast drying, irrigation events, or recovery after stress.

The performance should be quantified at multiple aggregation scales (e.g., daily, weekly, bi-weekly), which would also be more coherent with irrigation comparison at different temporal scales. Discuss implications for agricultural applications, where short-term stress is critical.
3. Guidance for the irrigation modeling community: Given the comparative scope, the study should conclude with more explicit guidance:
How to reconcile crop models’ explicit treatment of management practices with generalized LSM schemes.

Implications of shared assumptions (e.g., optimal irrigation rules at ~50% TAW) for large-scale irrigation simulations. Moreover, in the case of sprinkler irrigation, the adoption of such low MA thresholds, essentially at the onset of water stress, appears unjustified, since sprinkler systems enable the frequent application of small water amounts, as evidenced in Fig. 8d.

Pathways to integrate crop model process detail into LSMs.

Minor comments.
Model inconsistency in vegetation vs ET (Fig. 6): The observation that AquaCrop overestimates vegetation (positive bias in Fig. 6k) while underestimating ET (Fig. 6l) is surprising, especially in quasi non-stressed conditions (as suggested by the MA threshold and Fig. 8c, where MA remains above 50% of TAW). Although possible causes are discussed in the article, this inconsistency deserves a more in-depth analysis.
L47-49: Right, but irrigation and other agricultural practices significantly alter water, energy, and carbon fluxes. LSMs need to better represent these processes.
L64: The statement ‘irrigation is not an addition to the model to better represent the water balance’ is unclear. In fact, in crop water models irrigation explicitly supplies water to the soil water storage, thereby improving the representation of water balance components (e.g., RZSM, ET) (Olivera-Guerra et al., 2018).”
L66: Beyond their use in estimating net irrigation requirements (i.e., crop water needs), crop models have also been applied to estimate actual irrigation water use under a range of regimes, from deficit to excess irrigation, making them highly relevant for large-scale irrigation assessments (Olivera-Guerra et al., 2023; Laluet et al., 2024).
L341-342: If runoff is not generated after irrigation in AquaCrop, percolation losses should be included explicitly as residuals in the water balance.
L356-359: Dari et al. (2023) account for evaporative losses in potential ET, similar to AquaCrop, which incorporates ET into the daily calculation of irrigation requirements.
L359-360: While assuming a 100% irrigated fraction for irrigated pixels could lead to overestimation of irrigation, this does not align with the reported values. Furthermore, it may be inconsistent with the approach described in L345–L347, where a conveyance efficiency factor is applied to adjust Noah-MP estimates to match management data.
L505-506: The conclusion that both models approximate average irrigation rates after accounting for losses seems overstated, given substantial biases. However, the results do not fully support this assertion. Significant errors arise not only from the accounted losses but also from assumptions inherent to each model, including the overestimation of irrigated areas noted above. Importantly, both models apply the same irrigation rule—filling the soil water storage once a critical soil moisture threshold (~50% of TAW) is reached, effectively simulating optimal irrigation, which constitutes a major source of bias.
L520: Add reference for SM_WP and SM_FC being tuned to obtain accurate ET fluxes.
L522: It is unclear how the use of ensembles can better represent irrigation. Moreover, it feels out of context.
L570-571: Note that irrigation has been evaluated at finer scales (weekly/bi-weekly) by Dari et al. (2022, 2023), Laluet et al. (2024), and Olivera-Guerra et al. (2023).
L575-576: Highlight Olivera-Guerra et al. (2023)’s approach for calibrating temporal dynamics of irrigation at the district scale, which is compatible with large-scale applications using either crop models or land surface models.
This manuscript presents a timely and well-structured comparison of AquaCrop and Noah-MP, making a valuable contribution by bridging crop and land surface modeling perspectives. The study has strong potential, but the conclusions are currently weakened by the use of heavily aggregated (15-day) ET and SSM comparisons and by limited discussion of uncertainties in satellite-based reference datasets. Expanding the uncertainty analysis, exploring multiple aggregation windows, and offering explicit recommendations to the irrigation community would substantially strengthen the paper.
References:
Dari et al. (2022) : https://doi.org/10.1016/j.agwat.2022.107537
Dari et al. (2023) : https://doi.org/10.5194/essd-15-1555-2023
Laluet et al. (2024) : https://doi.org/10.1016/j.agwat.2024.108704
Olivera-Guerra et al. (2018) : https://doi.org/10.1016/j.agwat.2018.06.014
Olivera-Guerra et al. (2023) : https://doi.org/10.1016/j.agwat.2022.108119

Reply
Citation: https://doi.org/10.5194/egusphere-2025-2550-RC1

Louise Busschaert, Michel Bechtold, Sara Modanesi, Christian Massari, Dirk Raes, Sujay V. Kumar, and Gabriëlle J. M. De Lannoy

Viewed

Total article views: 886 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
782	83	21	886	16	31

HTML: 782
PDF: 83
XML: 21
Total: 886
BibTeX: 16
EndNote: 31

Views and downloads (calculated since 01 Jul 2025)

Month	HTML	PDF	XML	Total
Jul 2025	136	35	11	182
Aug 2025	143	12	1	156
Sep 2025	463	21	7	491
Oct 2025	40	15	2	57

Cumulative views and downloads (calculated since 01 Jul 2025)

Month	HTML	PDF	XML	Total
Jul 2025	136	35	11	182
Aug 2025	143	12	1	156
Sep 2025	463	21	7	491
Oct 2025	40	15	2	57

Viewed (geographical distribution)

Total article views: 836 (including HTML, PDF, and XML) Thereof 836 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 23 Oct 2025

Short summary

This study estimates irrigation in the Po Valley using AquaCrop and Noah-MP models with sprinkler irrigation. Noah-MP shows higher annual rates than AquaCrop due to more water losses. After adjusting, both align with reported irrigation ranges (500–600 mm/yr). Soil moisture estimates from both models match satellite data, though both have limitations in vegetation and evapotranspiration modeling. The study emphasizes the need for observations to improve irrigation estimates.


Total:	0
HTML:	0
PDF:	0
XML:	0