the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Satellite-based Near-Real-Time Global Daily Terrestrial Evapotranspiration Estimates
Abstract. Accurate and timely information on global terrestrial actual evapotranspiration (ET) is crucial in agriculture, water resource management and drought forecasting in a changing climate. While numerous satellite-based ET products have been developed in recent decades, few provide near-real-time global terrestrial ET estimates. The MOD16 ET dataset, currently updating at the fastest rate, still experiences a delay of over two weeks. This is because most satellite-based ET algorithms rely on meteorological data from land surface models or in situ measurements, which cannot be obtained in near-real-time, resulting in delays of more than two weeks. To expedite global ET data access, we developed the Moderate Resolution Imaging Spectroradiometer (MODIS) based Variation of Standard Evapotranspiration Algorithm (VISEA) to provide global daily ET data within a week of the actual measurements at a spatial resolution of 0.05°. The VISEA model incorporates several key components: (1) A vegetation index (VI)-temperature (Ts) triangle method to simulate air temperature (Ta), serves as a basis for calculating other meteorological parameters (e.g., water vapor deficit and wind speed); (2) A daily evaporation fraction (EF) method based on the decoupling parameter, converts satellite-based instantaneous observations into daily ET estimates; (3) A net radiation calculation program takes into account cloud coverage in the atmosphere's downward longwave radiation. The VISEA model is driven by shortwave radiation from the European Centre for Medium-range Weather Forecasts (ERA5-Land) and MODIS land products, e.g., surface reflectance, land surface temperature/emissivity, land cover products), vegetation indices, and albedo as inputs. To assess its accuracy, we compared VISEA-with measurements from 149 flux towers, five other satellite-based global ET products, and precipitation data from the Global Precipitation Climatology Centre (GPCC). The evaluations show that the near-real-time ET using VISEA performs with similar accuracy to other existing data products and offers a significantly shorter time frame for daily data availability. Over 12 landcover types, the mean R is about 0.6 with an RMSE of 1.4 mm day-1 at a daily scale. Furthermore, the consistent spatial patterns of multi-year average VISEA align closely with GPCC precipitation data, reaffirming the dataset's ability to accurately represent global terrestrial ET distribution. To emphasize the capabilities of the VISEA for drought monitoring, we analyzed the spatial and temporal variations of ET during a drought event and subsequent recovery with precipitation in the Yangtze River basin from August 28th to September 1st, 2022. The VISEA distinctly illustrated low ET levels (<0.2 mm day-1) across most areas of the Yangtze River Basin on August 28th, indicating the severity of the drought. Conversely, a noticeable increase in ET (>0.9 mm day-1) is observed on August 29th, signifying the retreat of the drought due to precipitation. The near-real-time global daily terrestrial ET estimates could be valuable for meteorology and hydrology applications requiring real-time data, particularly in coordinating relief efforts during droughts. The VISEA code and dataset are available at https://doi.org/10.11888/Terre.tpdc.300782 (Huang et al., 2023a).
- Preprint
(4162 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on essd-2023-495', Mingliang Liu, 01 Feb 2024
This paper claimed that the advantage of this new global ET data product is that it is near-real-time which has about one week's delay, but with a little bit better or comparable accuracies with other data products. The authors should discuss how much contribution of this one-week earlier (than other models/data products) to the community and how possible accuracies could be improved if more observations such as climate data are added and the estimation be delayed to two-weeks or even further. The authors also need make the near-real-time data public available or accessible (i.e. the data products will be updated at real time if being needed); or the source code or tool could be used to calculate global ET through cloud computing platforms, such as GEE. The bias on estimated air temperature and net radiation (might also need add some comparisons with other regional and global data sets on net radiation estimations) should be addressed before publishing this data products since the estimated air temperature and the application of shortwave radiation as input is the bases of this near-real-time ET products. What are the uncertainties coming from the estimated EF by using calculated vegetation and soil resistance, rather than directly from remote sensed information, such as surface temperature and NDVI (through Ts-VI triangle method)?What is the differences of this data product with other global ET on the long-term trend, inter-annual variation, and under extreme climate events?Line 228: correct the label for surface temperature.L118-120: this description seems not correct since VISEA also use the thermal information such as surface temperature.L209-216: how Ta is estimated for each 0.05 degree pixel?L238-246: how is air temperature of each pixel estimated?L297-301: rephrase.L390-399: confusing on the differences between Ts and Ta.Figs 3-6: explain what frequency mean.Figure 8: suggest adding a plot to show the global annual ET during the study period from these various data products.L515-529: need more precised description. Is there no regions with energy limited ET? Should be better to describe the differences in regions with moisture and energy limited ET.L530-546: can other global data product reveal the same pattern? Also need adding the conditions before 8/27/2022 to support the point that estimated ET could represent the soil moisture condition. Need clearly highlight the points for introducing this case analysis.L579-582: It is confusing.L584: are you talking about the estimated air temperature?L586-587: should be more specific. What does "VISEA relies solely on vegetation coverage as an indirect constraint" mean?L604-605: where this conclusion come from?L607-608: the claim that "VISEA aligns with GPCC a... in most areas worldwide" is confusing. Does VISEA generate precipitation?L668-670: add more literature to support this claim. Why are the effects of VPD and leaf water potential on canopy surface resistance so minor that it could be removed totally in the equation?Citation: https://doi.org/
10.5194/essd-2023-495-RC1 - AC1: 'Reply on RC1', Lei Huang, 20 Mar 2024
-
RC2: 'Comment on essd-2023-495', Seungcheol Oh, 05 Feb 2024
Summary:
The study introduces a novel approach leveraging the Moderate Resolution Imaging Spectroradiometer (MODIS) to deliver global daily actual evapotranspiration (ET) estimates with a spatial resolution of 0.05°, available within a week of satellite measurements. VISEA employs a combination of a vegetation index-temperature triangle method, a daily evaporation fraction method, and a net radiation calculation that incorporates cloud coverage, utilizing inputs from both ERA5-Land and MODIS land products. The algorithm's efficacy is validated through comparisons with data from 149 flux towers, other satellite-based ET products, and GPCC precipitation data, demonstrating VISEA's comparable performance.
General Comments:
The manuscript is promising and contributes valuable insights to the field of Earth System Science Data. However, it requires minor revisions before it can be considered for publication. My suggestions mainly pertain to enhancements in figures and tables, as well as a need for a more in-depth discussion.
Abstract: Line 39: Delete ‘)’.
Introduction:
Line 69-87: Are there any other satellite-based daily ET products not covered in this part? If not, it's recommended to more clearly highlight why these specific ET products were chosen for discussion. Clarify that each represents unique algorithmic approaches and are widely recognized within the scientific community for their contributions to global ET estimation.
Table 2:
Please double-check the time periods listed for MOD16 and GLEAM in Table 2. It's possible that more recent data are available for both datasets, extending beyond the years currently noted in your table. Additionally, there seems to be a discrepancy between the time period for MOD16 mentioned in Figure 8, which is listed as 2001-2014, and what's noted in Table 2 as 2001-2013.
Figure 8:
I suggest adding a comparison of the annual variation of ET with latitude for different remote sensing products to Figure 8 for an improved analysis. For reference, please see Figure 3 in the paper by Chen et al., 2021. https://doi.org/10.1029/2020JD032873
Discussion:
- The discussion could benefit from a more detailed analysis of the methodological uncertainties inherent in VISEA. Consider exploring not only the input data challenges but also the underlying assumptions and limitations of the model itself.
- The discussion mentions several critical points regarding the sources of bias and inaccuracies but seems to lack sufficient citation from existing literature to contextualize these findings within the broader field.
- Conclude the discussion with specific suggestions for future research directions that could address the identified gaps and uncertainties. This may include the development of alternative methods for estimating air temperature and net radiation, the incorporation of additional variables such as soil moisture and water availability into the model, or the potential for integrating machine learning techniques to improve estimation accuracy.
Citation: https://doi.org/10.5194/essd-2023-495-RC2 - AC2: 'Reply on RC2', Lei Huang, 20 Mar 2024
-
RC3: 'Comment on essd-2023-495', Ren Wang, 11 Feb 2024
Comments to Huang et al. Satellite-based Near-Real-Time Global Daily Terrestrial Evapotranspiration Estimates
This manuscript aims to present a near-real-time global terrestrial evapotranspiration (ET) estimate by utilizing their previous developed VISEA algorithm, satellite observation, and ERA5-land reanalysis data. The development of a near-real-time ET product holds great significance for drought monitoring, water resources management, and climate change study. I am pleased to witness this progress, and I think the manuscript may have potential to be published in ESSD. However, there are several major concerns that the authors should further revise or clarify.
Major comments:
- The authors essentially use an energy balance based approach to estimate ET,but they do not discuss the limitations of this methodology. In my experience, the EF/energy balance method may not perform well in the local winter season or at high latitudes (e.g., 60-90N) due to energy Although the authors only utilize ERA5-Land downward shortwave radiation values greater than 0, they do not discuss the difficulties and limitations of the methods. However, since the focus of the study in on daily-scale ET estimation, this issue remains relevant and requires further attention. If I am right, the resulting ET estimates in this study is a near-global ET product. In the northern hemisphere (60-90 N), the ET estimates may be unavailable or miss data for nearly half of the year.
- The validation of the new product appears to be insufficient. I recommend adding regional average curves to compare their changes over time. Additionally, while Figure 8 presentsthe spatial distribution characteristics of the multi-year average, what about the extreme values (e.g., 5th and 95th percentiles)? Considering these products are averaged over different time periods, does it affects the comparison results? Clarifying these aspects will enhance the robustness of the validation.
- The objectiveof this study is to provide a near-real-time global ET product, yet the methodology duplicates information from the authors’ previously published paper (Huang et al., 2021, Earth and Space Science) (e.g., Figure1, the description of the VISEA model, and the decoupling parameter for daily EF). Therefore, it is very crucial to carefully address and further clarify the distinctions between these repeated details.
Minor comments:
- The near-real-time ET proposed in this studyprimarily relies on MODIS Land product at 05 degrees and ERA5 data at 0.1 degrees. Shouldn't the ET products should be limited to a relatively coarse resolution of 0.1 degree?
- Regarding Figure 3, it would be beneficial to include the level of significance of the correlation analysis.
- Table 3: Why does the VISEA exhibit greater bias than other ET products in several vegetation types, such as CRO, DNF, ENF, GRA? These vegetation types are the main types on the land surface. Further clarification on the reasons for this discrepancy would enhance the interpretation of the results.
- Figure 8: Why VISEA presents an obvious higher ET values compared to other products in most tropical areas of South America and Africa?
- Section 2.1 should avoid repetition if the different modules and steps of these methods have been clearly described in previous published papers, ensuring that the current manuscript offers new and valuable information.
- Line 64: “these models often have limited spatial resolutions, making them less effective...”, I do not think this is the truth. The ERA5 reanalysis, which combines climate model simulation and observational data, also produced latent heat flux (ET in energy units) with a delay of six days.
- Line 65-68: It would be valuable to highlight the advantages of satellite remote sensing-based ET estimates compared to climate model simulation.
- Did the authors perform energy closure correction or validation for the FLUXNET observational data?
- Line 197: Is this the truth? It may be necessary to provide additional context or references to support the claim that the daily-scale G is approximately 0 and can be ignored.
- Considering the availability of more recent products, such as Jung’s FLUXCOM (2019), and the potential impact of sensor developments, the choice of using older data sources, such as AVHRR (2001-2006), should be justified. Comparisons with older data may be influenced by advancements in sensor and computer technologies.
- Line 409: Please briefly explain the method used to estimate Rn in VISEA. I suspect it is fundamentally different from ERA5_Rd. Please clarify the purpose and significance of comparing these two variables.
- Line 548: “actual ET measurement”?
- Line 605: Please why the VISEA approach performs better than other products at DNF? I am not sure, but the bias and RMSE values of VISEA at DNF are larger than other products (see Table 3)?
Citation: https://doi.org/10.5194/essd-2023-495-RC3 - AC3: 'Reply on RC3', Lei Huang, 20 Mar 2024
Data sets
Satellite-based Near-Real-Time Global Daily Terrestrial Evapotranspiration Estimates Lei Huang https://doi.org/10.11888/Terre.tpdc.300782
Model code and software
Satellite-based Near-Real-Time Global Daily Terrestrial Evapotranspiration Estimates Lei Huang https://doi.org/10.6084/m9.figshare.24647721.v1
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
429 | 84 | 32 | 545 | 15 | 19 |
- HTML: 429
- PDF: 84
- XML: 32
- Total: 545
- BibTeX: 15
- EndNote: 19
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1