the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 10 Apr 2026
Incorporating irrigation effects into high-resolution daily land evaporation estimates over the Iberian Peninsula
Abstract. Land evaporation (E) links the water, energy, and carbon cycles and plays a central role in agriculture, water management, and land–climate interactions. However, estimating E at high spatial and temporal resolution remains challenging, especially in irrigated regions. This study presents a novel framework to generate daily 1 km E estimates for 2018–2022 over the Iberian Peninsula by explicitly representing irrigation in the recently released Global Land Evaporation Amsterdam Model version 4 (GLEAM4). To this end, high-resolution (1 km) meteorological forcing is combined with Sentinel-1 soil moisture and ancillary information on irrigated extent and satellite-based crop phenology. Our method constrains E below potential evaporation (Ep) even in irrigated land, leveraging observational data of vapour pressure deficit, air temperature, vegetation optical depth, leaf area index, wind speed, and shortwave radiation, which allows irrigated crops to respond realistically to diverse sources of vegetation stress, rather than assuming Ep rates. Results reveal increases in E over irrigated areas of up to 450 mm yr-1 when irrigation is explicitly considered, with spatial patterns consistent with independent irrigation estimates. Evaluation against eddy-covariance measurements demonstrates marked improvements at two irrigated sites in the Iberian Peninsula, with increases in daily Kling-Gupta Efficiency (KGE) compared to simulations without irrigation of 0.40 and 0.70, respectively. The approach is also relatively robust to false positives in the irrigation mask owing to the fractional vegetation structure of GLEAM4. Overall, the resulting high-resolution E dataset provides a realistic representation of irrigation practices and supports applications in both agricultural management and regional water-resource assessments. The approach will be extended to global scales through integration into future GLEAM releases.
Competing interests: At least one of the (co-)authors is a member of the editorial board of Hydrology and Earth System Sciences.
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.- Preprint
(17881 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-1856', Stefanie Fischer, 09 Jun 2026
-
RC2: 'Comment on egusphere-2026-1856', Anonymous Referee #2, 20 Aug 2026
The paper entitled “Incorporating irrigation effects into high-resolution daily land evaporation estimates over the Iberian Peninsula” shows how the inclusion of irrigation in the GLEAM framework can improve the estimates of evaporation (E). The manuscript is well written and well structured, and I think it is a useful contribution. However, some aspects lack clarity, especially related to the workflow to include irrigation. I asked for minor revisions because I believe most comments can be answered via in-text clarifications. They come from my own background in irrigation and data assimilation, so they are somewhat focused on that part of the workflow. I hope that this helps to improve and strengthen the manuscript.
Major:
- The connection between the “irrigation scheme”, i.e. bringing the SM to FC, and the subsequent S1-SSM DA is not entirely clear. The anomalies are rescaled to the climatology of the open loop, which already contains the irrigation (SM to FC) (L233-235), so does the SM end up hovering around FC during the growing season? If that is the case, the DA can likely only adjust the timing and not the irrigation amounts. Related to this, if there is consistent irrigation (S1-SSM consistently higher), the anomalies would lose the irrigation signal. Could the authors clarify what the DA actually adds here? Did the authors perform any experiments of irrigation (back to FC) without DA to better clarify the impact of the DA?
- The distinction between which soil depth is meant is not always clear. Is the surface soil moisture set to FC, or the entire profile, or the root zone? The manuscript never mentions rooting depth, so it is not clear whether a constant root zone is assumed or something dynamic. This matters more than usual here, because L141–143 states that GLEAM4 runs a multi-layer water balance that also accounts for plant access to groundwater (Hulsman et al., 2023), but this component is not mentioned again anywhere in the paper. Could the authors clarify which layers are reset to FC, whether FC is layer-specific, and how the groundwater access module behaves in the irrigated run? In general, the authors should clarify which soil depth they are talking about throughout the manuscript.
- GLEAM4 accounts for plant access to groundwater (L141–143), but this is not discussed further. Since plants may already be sustaining E through groundwater in the benchmark run, the difference in E between the two runs may not be fully representative of irrigation. Could the authors discuss this?
- The propagation of the DA updates from the surface to the deeper layers is said to be done via process-based modelling (L237). Could the authors clarify which processes (Richards equation, other)? It might be defined in the description of GLEAM v4 but it would be useful to have it here. Also, is the added water conserved, i.e. could it drain?
- If I read L219-220 correctly, SM is increased to FC only within the short vegetation fraction. But ES-Cnd is described in Table 1 as a woody savanna with irrigated olives, and L115-116 lists olives, grapes and fruit trees among the irrigated crops of the Ebro, is that defined as short? If not, how does irrigation then reach the tall vegetation fraction, which is actually irrigated?
- At La Cendrosa the irrigated run still underestimates E quite strongly (Fig 5a: observations up to about 8 mm/day against about 4 mm/day simulated). Even with SM at FC the model does not reach the observed fluxes, which points more towards the stress module than towards water availability. Were irrigated croplands part of the training data of the S model? On a related note, γ going from 1.63 to 0.52 is described as moving closer to one, but it flips from over- to underestimation of the variability, so I would phrase this more carefully.
- Areas with flood irrigation have (1) a different evaporation and (2) high uncertainty in the retrievals, and La Cendrosa, which shows the largest improvement, is exactly a flood or gravity irrigated site (Table 1). Could the authors acknowledge this?
- In general, the S1-SSM uncertainties and limitations could be better acknowledged. The manuscript mentions subsurface scattering over dry soils (L300, L368), but not the other end, i.e. saturation and canopy attenuation over denser crops such as maize.
Minor:
- The KGE numbers in the abstract do not match the results. The abstract reports increases of 0.40 and 0.70, while L305-310 give -0.48 to 0.19 at La Cendrosa (an increase of 0.67) and 0.21 to 0.40 at ES-Cnd (an increase of 0.19). The 0.40 looks like the final KGE of ES-Cnd rather than the increase.
- Fig 3 c): does the delta truly stop at 0? Are there any negative values? Since run (a) also lacks the DA there could be some, so please explain.
- The claim that the approach is relatively robust to false positives (abstract and L407) rests on three sites, of which two (ES-LM1 and ES-LM2) are almost co-located, and the KGE decreases at both. Only FR-Aur improves slightly. I would soften this. False negatives, i.e. irrigated areas missed by the mask, are not assessed at all.
- Fig 4: the text boxes on the map are difficult to read. Would it be possible to increase the fontsize? Alternatively, a table with KGE, r, β and γ for all sites and both experiments could also help.
Text:
- L401: missing space after the comma in “transpiration,thereby”
- L216: “grid-cells” → “Grid-cells”
- L407: “constrained by fractional-vegetation approach” → “constrained by the fractional-vegetation approach”
Citation: https://doi.org/10.5194/egusphere-2026-1856-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 670 | 227 | 34 | 931 | 28 | 36 |
- HTML: 670
- PDF: 227
- XML: 34
- Total: 931
- BibTeX: 28
- EndNote: 36
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The study under review (“Incorporating irrigation effects into high-resolution daily land evaporation estimates over the Iberian Peninsula”) evaluates the GLEAM4 version at a 1 km spatial resolution over the Iberian Peninsula, comparing two approaches: (1) the default GLEAM4 framework, which implicitly accounts for irrigation through soil moisture and vegetation dynamics, and (2) an enhanced version that explicitly incorporates irrigation by integrating spatial information on irrigation extent and crop phenology (used to infer agricultural management timing) and by increasing soil water content to field capacity in the irrigated areas.
The model first computes potential evaporation (Ep) using the Penman equation, which is then reduced by a stress factor (S, ranging from 0 to 1). This stress factor is derived via a machine learning approach that accounts for environmental controls on evaporation (i.e. soil moisture, vapor pressure deficit, air temperature, global radiation, wind speed), particularly their influence on stomatal conductance.
The analysis covers the period 2018–2022, selected primarily because the underlying irrigation extent data (Meier et al., 2018) are static and based on 2018. Similarly, the crop phenology data (FAO, 2018) are also from 2018 and used to derive a daily irrigation mask. While the authors likely assume that irrigation practices remained relatively stable over this 5-year window, the rationale for choosing this specific time period is not explicitly stated in the manuscript. A key innovation lies in the dynamic application of this static irrigation mask: irrigation is only applied when the irrigation area intersects with short vegetation (as captured by the fraction of absorbed photosynthetically active radiation, fAPAR), thereby introducing a dynamic, biologically informed constraint on irrigation timing. Additionally, the model incorporates surface soil moisture data assimilation to improve soil moisture estimates.
Model performance is evaluated against daily evaporation observations from eight eddy covariance stations. Time series of modeled and observed evaporation are presented for two representative irrigated sites, with particular emphasis on the physical behavior of the stress factor and soil moisture dynamics under both irrigated and non-irrigated conditions. This allows for a meaningful assessment of the model’s ability to represent the physical mechanisms governing evaporation.
Accurate estimation of evaporation is crucial for understanding the energy balance and hydrological cycle, as it directly influences other components such as runoff, soil moisture, and groundwater storage. In agricultural regions, especially in water-scarce environments and under climate change, capturing the impact of irrigation is essential for effective water resource management. The study addresses this need by developing a method to incorporate irrigation into the GLEAM4 framework, enabling the derivation of temporally continuous, high-resolution evaporation estimates at regional scales (with potential for extension to global applications).
The main limitation of the study is the use of static information on irrigation extent and timing, which is critically addressed. The method of soil moisture assimilation using the climatology from the irrigated GLEAM4 version is not entirely clear, nor is the rationale for running the non-irrigated GLEAM4 version without soil moisture data assimilation. This needs to be explained and justified more precisely in the methods chapter. In the results chapter, certain regions of the study area are mentioned without a description of their location, which may hinder understanding for readers unfamiliar with the Iberian Peninsula or with limited background in European geography. Further details are provided in the comments attached.
Overall, the study is well designed, well structured, and clearly written. The results are presented in a clear manner and are supported by an appropriate number of figures that are well integrated into the text. I recommend the manuscript for publication after minor revisions.
Specific comments:
L112: These rivers should be indicated in the map of the study area (e.g. in Figure 1), which helps the reader to keep track of the described areas (particularly those being not familiar with the region).
L150: Some variables are abbreviated, others are not... please first introduce the variable and give abbreviation in brackets: e.g. vegetation optical depth (VOD), air temperature (Ta) ...
L193: Irrigated areas are represented on static information based on the year 2018 assuming these are still representative for 2019 - 2022... this needs to be considered in the discussion.
L216: Grid-cells
L234: This is not clear to me. What is the baseline climatology for the anomalies?
... GLEAM4 soil moisture estimates including irrigation are based on static information of 2018 (despite the modulation through short vegetation fraction)... are these used to calculate the baseline climatology? Please justify
Caption Figure 2: “(a) the model run without irrigation and soil moisture data assimilation”. This is not clear to me... without irrigation and without soil moisture data assimilation?
The original GLEAM4 Model without irrigiation (Miralles et al. 2025) includes soil water assimilation, why omitting data assimilation here?
L320: The argumentation for the limited effect of false positives on soil moisture dynamics is plausibly explained but requires further analysis, as only two stations are used as a reference (discussion).
L337: I understand that this paragraph aims to first introduce total E, before evaluating its components, but it’s a bit hard to navigate, as it introduces Figure 6 but actually describes Figure 3b.
L340: Readers from outside the Iberian Peninsula or those without a strong background in European geography might struggle to visualize or locate the mentioned regions (e.g., Aragon, Andalusia). Please use a more descriptive language (in northern and southern Spain, respectively).
L343: The placement of the reference Miralles et al. (2025) is confusing, as it may be interpreted as a dataset, whereas Miralles et al. (2025) conducted a comparative analysis of GLEAM4 against other state-of-the-art evaporation datasets.
L347: Please also add the range of Ei.
Caption Figure 5: The location of ES-Cnd (analogous to La Cendrosa) should be specified for clarity.
L357: In addition, is the fraction of bare soil increasing in winter (due to dormancy or agricultural practices), area corresponds to the cropland extent?