the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Development of a HAMSTER: Hyperspectral Albedo Maps dataset with high Spatial and TEmporal Resolution
Abstract. Surface albedo is an important parameter in radiative transfer simulations of the Earth's system, as it is fundamental to correctly calculate the energy budget of the planet. The Moderate Resolution Imaging Spectroradiometer (MODIS) instruments on NASA's Terra and Aqua satellites continuously monitor daily and yearly changes in reflection at the planetary surface. The MODIS Surface Reflectance dataset (MCD43C3, Version 6.1) gives detailed albedo maps in seven different spectral bands in the visible and near-infrared range. These albedo maps allow us to classify different Lambertian surface types and their seasonal and yearly variability and change, albeit only in seven spectral bands. However, a complete set of albedo maps covering the entire wavelength range is required to simulate radiance spectra, and to correctly retrieve atmospheric and cloud properties from Earth's remote sensing. We use a Principal Component Analysis (PCA) regression algorithm to generate hyperspectral albedo maps of Earth. Combining different datasets of hyperspectral reflectance laboratory measurements for various dry soils, vegetation surfaces, and mixtures of both, we reconstruct the albedo maps in the entire wavelength range from 400 to 2500 nm. The PCA method is trained with a 10-years average of MODIS data for different times of the year. We obtain hyperspectral albedo maps with a spatial resolution of 0.05° in latitude and longitude, a spectral resolution of 10 nm, and a temporal resolution of 8 days. Using the hyperspectral albedo maps, we estimate the spectral profiles of different land surfaces, such as forests, deserts, cities and icy surfaces, and study their seasonal variability. These albedo maps shall enable to refine calculations of Earth's energy budget, its seasonal variability, and improve climate simulations.
- Preprint
(27123 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (extended)
-
RC1: 'Comment on egusphere-2024-167', Anonymous Referee #1, 09 Apr 2024
reply
Review of Roccetti et al., “Development of a HAMSTER: Hyperspectral Albedo Maps dataset with high Spatial and Temporal Resolution”
Here, the authors detail an effort towards generating hyperspectral global albedo maps based on MODIS albedo estimates combined with laboratory-measured spectra for various surface types. The need for this type of study is sound, and the dataset could well be of interest to ESM modelers in particular.
Regarding the execution of the study, the manuscript is of good writing quality and mostly of logical structure. However, there are still numerous choices made in the analysis which require further clarification and/or justification until publishable quality may be reached. Therefore, at this stage a major revision of the manuscript is required to address the issues listed below:
Major issues
Choice of MODIS data: The MCD product user manuals quite clearly state that “We recommend that users use the MCD43D 30 arcsecond products (or the 30 arcsecond MCD43GF gapfilled products) instead of the MCD43C products as much as possible, since the MCD43C products merely represent averages of the direct BRDF retrievals obtained with the MCD43D processing. The averaged MCD43C products thus necessarily include various QA flags within each pixel, and are therefore less rigorously high quality than the MCD43D products.” Yet, the authors have chosen to base their analysis on MCD43C3. Why?
Reference spectra: One is tempted to ask about the geographical distribution of the soil and vegetation spectra used here. For example, boreal forests are and behave quite differently from African rainforests – how well are the various regions of the Earth represented in the data? Do you have any means of assessing the uncertainty that is potentially related to misrepresented vegetation in particular, as the soil-vegetation references appeared to be heavily weighed towards soil sampling? What is the overall uncertainty of the HAMSTER data layers?
Treatment of snow within the HAMSTER maps: The reference spectra did not seem to contain snow and there was no mention of simulating snow spectra with snow models. How then did you achieve the PCA deconstruction over snow? What about mixed pixels with fractional snow cover and fractional vegetation cover, how are those handled?
Structure: Inclusion of HAMSTER validation in section on data and methods is confusing, as these are clearly results and not source data or its processing. Recommend revising the structure to start section 3 with the validation, and then proceeding towards analysis of features in HAMSTER data.
Gap filling on multiple layers: There are 5 different layers of gap filling needed – what are the respective percentages of global MODIS grid that are filled out at each of these steps? Do I understand correctly that in step 2, albedo values spanning to +/-40 (5x8) days of target date may be selected to fill the gap? 40 days during snowmelt or vegetation greenup could result in dramatically different gap-filled albedos relative to what was actually on the ground, what is the justification for this wide search window?
Minor issues (line# or element)
130 – I was under the impression that MCD products do not contain sea ice albedo or sea ice cover? Certainly the result figures show no Arctic sea ice, so why is sea ice referenced here and again in section 3? This also refers to the major issue on snow treatment – sea ice albedo is even more complex than snow, and I haven’t seen anything in the manuscript about its treatment?
Fig 7 – TROPOMI coverage extends elsewhere from Africa, why only focus the comparison there?
Section 3 – the PCA-interpolated spectral albedos appear reasonable for most land surfaces, certainly. But I question the validity of the urban areas’ spectra, there did not seem to be any reference spectra on man-made structures in the study? And how would PCA handle the extremely nonlinear shifts in land cover and surface material that are common to urban areas? Since the overall quality of MODIS albedos over cities has not been quantitatively assessed and the validity of the RossThick-LiSparse retrieval is uncertain over them, I would be very careful of highlighting those areas in particular unless the authors can prove that their spectra are valid.
Citation: https://doi.org/10.5194/egusphere-2024-167-RC1
Data sets
HAMSTER: Hyperspectral Albedo Maps dataset with high Spatial and TEmporal Resolution Giulia Roccetti et al. https://zenodo.org/records/10494404
Video supplement
HAMSTER dataset Giulia Roccetti https://av.tib.eu/media/66248
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
276 | 76 | 10 | 362 | 6 | 5 |
- HTML: 276
- PDF: 76
- XML: 10
- Total: 362
- BibTeX: 6
- EndNote: 5
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1