the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 16 May 2025
Detecting irrigation signals from SMAP L3 and L4 soil moisture: A case study in California's Central Valley
Abstract. Recent advances in satellite-based soil moisture observations present a promising opportunity to monitor irrigation dynamics from space and support the refinement of hydrological and land surface model simulations. This study presents an approach for identifying irrigation signals using data from the Soil Moisture Active and Passive (SMAP) mission, with feasibility demonstrated in Central Valley, California. The approach leverages two SMAP products: the Level 3 Enhanced product, which provides satellite-based soil moisture observations that inherently capture irrigation effects, and the Level 4 data assimilation product, which incorporates only anomalies from SMAP Level 1 brightness temperature data, thereby excluding irrigation effects. The approach is based on the hypothesis that, after correcting for systematic differences not related to irrigation, the soil moisture difference between the Level 3 and 4 products during the cropping season is primarily attributable to irrigation. This hypothesis is first verified by evaluating soil moisture consistency (i.e., temporal variability and long-term mean values) between the two products. For grid cells that meet this criterion, the mean difference (MD) between the two soil moisture products is calculated, separately for the cropping and non-cropping seasons, and then the irrigation signal is identified as the difference in MD between the cropping and non-cropping seasons. Validation of the estimated irrigation signal is made by comparing with two benchmark irrigation maps. The results show reasonable spatial correlations between our estimate and the two benchmark maps, with Pearson's correlation coefficients of 0.66 and 0.50, respectively. Findings demonstrate the potential of using SMAP products to extract irrigation effects in regions that have limited precipitation during the cropping season. Compared to other satellite-based irrigation detection studies, the proposed method requires minimal additional data and avoids additional model tuning beyond the SMAP processing.
- Preprint
(7762 KB) - Metadata XML
-
Supplement
(19455 KB) - BibTeX
- EndNote
Status: open (until 27 Jun 2025)
-
RC1: 'Comment on egusphere-2025-2004', Anonymous Referee #1, 02 Jun 2025
reply
Here is my review for “Detecting irrigation signals from SMAP L3 and L4 soil moisture: A case study in California’s Central Valley” submitted to Hydrology and Earth System Sciences. The paper aims to detect irrigation signals by comparing SMAP Level 3 and Level 4 soil moisture products, which is demonstrated in California’s Central Valley. The results highlight the potential of satellite-based observations to identify irrigation effects. However, several aspects of the methodology, data usage, and interpretation require clarification and improvement to strengthen the study’s scientific rigor and clarity. My recommendation is major revision to address the concerns outlined below.
Below, I provide general comments on the manuscript’s overall contribution, followed by specific comments to guide the authors in revising their work.
General Comments:
The authors note that SMAP L4 assimilation is based on brightness temperature anomalies, thereby excluding persistent irrigation effects embedded in the climatological mean. However, in practice, irrigation is not always applied consistently during soil moisture deficits; its timing can be irregular and largely influences soil moisture anomalies rather than the climatology. Moreover, this assumption neglects potential non-stationary changes in irrigation practices driven by agricultural policies.
I recommend that the authors assess whether the identified irrigation dates are realistic and consider conducting a synthetic experiment to demonstrate that SMAP L4 indeed lacks any irrigation-related signal.
Specific Comments:
1.Line 145: Limitation of SMAP L4 Assumptions
The manuscript states that SMAP L4 excludes irrigation signals in croplands with continuous irrigation. This assumption significantly limits the method’s applicability, as it may not account for regions with variable or intermittent irrigation practices. The authors should clarify the conditions under which this assumption holds and discuss its implications for the method’s generalizability.
2.Section 2.1.3: Outdated Irrigated Area Map
The manuscript relies on a 2005 Global Map of Irrigated Areas (GMIA) dataset, despite using SMAP data from 2016–2022. Given potential changes in irrigation extent over this 15-year period due to agricultural expansion, urbanization, or policy shifts, the use of an outdated map introduces uncertainty. The authors should justify this choice or consider incorporating a time-series irrigated area dataset to align with the SMAP data period. A discussion of how changes in irrigation patterns might affect the results is also warranted.
3.Equation (2): Handling Precipitation Bias and Crop Rotation
Equation (2) defines the irrigation signal as the difference in mean soil moisture difference (MD) between cropping and non-cropping seasons, based on SMAP L2_E and L4 data in irrigated grid cells. However, line 270 notes that high MD values in the non-cropping season may result from precipitation, which could confound irrigation signals. The authors should address how precipitation biases and rain/no-rain errors are considered during the cropping season. Additionally, the method assumes distinct cropping and non-cropping seasons, which may not apply to regions with year-round crop rotation. The authors should discuss the method’s applicability to such regions and propose potential adaptations.
4.Line 255 and Table 1: Inconsistent Threshold for Non-Irrigated Grid Cells
Line 255 states that non-irrigated grid cells have an irrigated area fraction below 0.1%, but Table 1 reports a threshold of 0.11% (non-irrigated grid cell b).
5.Figures 2(d) and 3(c): Bias in Soil Moisture Differences
Figures 2(d) and 3(c) show a higher mean difference between SMAP L3 and L4 in non-irrigated grid cell E (0.0471 m³/m³) compared to irrigated grid cell C (0.0354 m³/m³). This bias may stem from using the outdated 2005 GMIA dataset, as irrigation patterns may have changed. The authors should investigate whether this discrepancy reflects changes in irrigation extent or other factors (e.g., soil properties, land cover) and discuss their findings.
6.Line 320: Error in Supplementary Figure Reference
The reference to “Supplementary Information Figure S5-3” appears incorrect. Please verify and correct the figure number.
7.Figures 5 and S-4: Interpretation of Rnon Performance
The manuscript reports that 72.7% of Rnon values fall below the consistency threshold, suggesting limitations in the method’s performance for SMAP L4. The authors attribute this to the absence of saturated soil moisture values in L4 compared to L3_E during the non-cropping season. However, SMAP L3_E’s sensitivity to surface wetting (e.g., post-rainfall or standing water) may lead to unrealistic retrievals. The authors should clarify whether these values reflect true soil saturation or retrieval error.
8.Figure 7: Interpretation of Irrigation Signal (IS)
The statement in line 245 that “a larger IS value indicates higher irrigation intensity” implies that any positive IS value signals irrigation activity. However, Figure 7 shows biases in some regions (e.g., region (i)), which may undermine this interpretation. The authors should clarify whether positive IS values consistently indicate irrigation or if biases (e.g., from precipitation or land cover changes) could lead to false positives. A sensitivity analysis of IS values would help address this concern.
9.Figure 8: The validation of the estimated irrigation signal, which relies on comparisons with the GMIA and irrigation water use datasets, is inadequate. Using GMIA as both an input to the method and a validation dataset introduces circularity, undermining the independence of the validation process. Additionally, the manuscript reports a spatial correlation of 0.5 with the ZL21 map, which is relatively low and raises questions about the method’s accuracy. The authors should clarify the significance of this correlation value and consider incorporating independent validation datasets (e.g., in-situ irrigation records) to strengthen the evaluation of the irrigation signal estimates.
Citation: https://doi.org/10.5194/egusphere-2025-2004-RC1 -
CC1: 'Comment on egusphere-2025-2004', Nima Zafarmomen, 12 Jun 2025
reply
The manuscript presents a novel and promising approach for detecting irrigation signals using SMAP (Soil Moisture Active and Passive) Level 3 Enhanced (L3_E) and Level 4 (L4) soil moisture products, with a focus on the Central Valley, California, and extended analyses in the Snake River Plain and Nebraska High Plains. The method leverages the difference between satellite-observed (L3_E) and model-assimilated (L4) soil moisture data to isolate irrigation effects, capitalizing on the fact that SMAP L4 excludes irrigation signals due to its assimilation of brightness temperature anomalies. The study is well-structured, clearly written, and makes a compelling case for its simplicity and minimal reliance on additional data or complex model tuning compared to previous approaches.
The introduction could better contextualize the novelty of the proposed method by briefly summarizing how it differs from prior soil moisture-based studies (e.g., Zaussinger et al., 2019; Lawston et al., 2017) beyond the mention of avoiding complex model tuning. A concise statement on how the use of SMAP L3_E and L4 together is unique would strengthen the rationale. Add a sentence or two explicitly stating how the proposed method advances beyond existing soil moisture-based approaches, particularly in terms of leveraging SMAP’s internal products to ensure climatological consistency.
The explanation of the SMAP L4 assimilation process is technically dense and may be challenging for readers unfamiliar with data assimilation. A simplified summary could improve accessibility.
The choice of the ±0.04 m³/m³ threshold for MD consistency is based on the unbiased RMSE accuracy requirement of SMAP products, but its suitability for irrigation detection could be further justified, as irrigation signals may vary in magnitude across regions.
The discussion of discrepancies between the IS map, GMIA, and ZL21 map is thorough but could better address why the IS map fails to detect irrigation in areas with subsurface irrigation (e.g., region iv). A brief explanation of how SMAP’s 5 cm penetration depth limits detection of deeper irrigation could clarify this.
The conclusions could better highlight the study’s contribution to the field, such as its advancement over previous satellite-based methods or its potential to improve hydrological modeling. I strogly recommend autors to cite papers such as: Assimilation of Sentinel-Based Leaf Area Index for Modeling Surface-Ground Water Interactions in Irrigation District
Citation: https://doi.org/10.5194/egusphere-2025-2004-CC1
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 180 | 56 | 10 | 246 | 13 | 6 | 9 |
- HTML: 180
- PDF: 56
- XML: 10
- Total: 246
- Supplement: 13
- BibTeX: 6
- EndNote: 9
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1