Spatiotemporally consistent global dataset of the GIMMS Leaf Area Index (GIMMS LAI4g) from 1982 to 2020
Abstract. Leaf Area Index (LAI) with an explicit biophysical meaning is a critical variable to characterize terrestrial ecosystems. Long-term global datasets of LAI have served as fundamental data support for monitoring vegetation dynamics and exploring its interactions with other Earth components. However, current LAI products face several limitations associated with spatiotemporal consistency. In this study, we employed the Back Propagation Neural Network (BPNN) and a data consolidation method to generate a new version of the half-month 1/12° Global Inventory Modeling and Mapping Studies (GIMMS) LAI product, i.e., GIMMS LAI4g, for the period 1982−2020. The significance of the GIMMS LAI4g was the use of the latest PKU GIMMS NDVI product and 3.6 million high-quality global Landsat LAI samples to remove the effects of satellite orbital drift and sensor degradation and to develop spatiotemporally consistent BPNN models. The results showed that the GIMMS LAI4g exhibited higher accuracy than its predecessor (GIMMS LAI3g) and two mainstream LAI products (Global LAnd Surface Satellite [GLASS] LAI and Long-term Global Mapping [GLOBMAP] LAI), with an R2 of 0.95, mean absolute error of 0.18 m2/m2, and mean absolute percentage error of 15 % which meet the accuracy target proposed by the Global Climate Observation System. It outperformed other LAI products for most vegetation biomes in a majority area of the land. It efficiently eliminated the effects of satellite orbital drift and sensor degradation and presented a better temporal consistency before and after the year 2000 and a more reasonable global vegetation trend. The GIMMS LAI4g product could potentially facilitate mitigating the disagreements between studies of the long-term global vegetation changes and could also benefit the model development in Earth and environmental sciences.
Sen Cao et al.
Status: open (until 21 Apr 2023)
- RC1: 'Comment on essd-2023-68', Shangrong Lin, 21 Mar 2023 reply
Sen Cao et al.
Spatiotemporally consistent global dataset of the GIMMS Leaf Area Index (GIMMS LAI4g) from 1982 to 2020 https://doi.org/10.5281/zenodo.7649108
Sen Cao et al.
Viewed (geographical distribution)
Cao et al. generated a new global LAI dataset GIMMS LAI4g (hereafter, LAI4g), this dataset overcome the effect of satellite orbital drift and sensor degradation from NOAA series satellites. Its results may explain the LAI trend bias between pre-MODIS era (1982-1999) and MODIS era (2000-present) and it also explains whether the vegetation area is greening or browning over the world. This study fills the gap of how the AVHRR sensors affect the LAI trend (the significant improvement is showed at figure 6), which have not been solved in the previous studies.
However, three major points needed to be improved before publication.
#1 The authors used two datasets, i.e. PKU GIMMS NDVI (Li et al. under review), Landsat LAI samples (Zha et al. in preparation) as the principal data source for generating LAI4g product. However, these two datasets have not finished their peer-review process, so these two datasets are not fully convincing. At least, the authors should show the statistics on how robust these two datasets are. I also suggested the authors adding the ground measured LAI (could be found in Xu et al. 2018 RSE and Ma et al. 2022 RSE) as validation to verify LAI4g product.
Xu, B., Li, J., Park, T., Liu, Q., Zeng, Y., Yin, G., ... & Myneni, R. B. (2018). An integrated method for validating long-term leaf area index products using global networks of site-based measurements. Remote Sensing of Environment, 209, 134-151.
Ma, H., & Liang, S. (2022). Development of the GLASS 250-m leaf area index product (version 6) from MODIS data using the bidirectional LSTM deep learning model. Remote Sensing of Environment, 273, 112985.
#2 The saturation of LAI is common in LAI4g over biomes. For example, figure 3 a1 showed that the Landsat LAI ranged from 0 to 4 m2 m-2, however, the LAI4g is saturated at 3 m2 m-2. Similar conditions can be found at CRO, SHR, SVA, ENF, DNF. If there is a 1:1 line in each subplot, the underestimation of LAI4g in high range of Landsat LAI would be significant. I guess the reason the train data for LAI4g is NDVI, the NDVI may saturated but LAI not, so the model setting may lead to the LAI4g saturation.
#3 The analysis of LAI trend can be expanded. Zhu et al. 2021 showed that the CO2 fertilization effect (CFE) is diversity from satellite LAI products, and they showed different patterns before and after 2000. Since the LAI4g overcome the effect of satellite orbital drift and sensor degradation from NOAA series satellites, I think it could solve the problem raised in Zhu et al. 2021. So, the authors can add the analysis on how the LAI trend change before and after 2000 for LAI4g, which benefits the ecology research community on how the vegetation area is greening or browning over the world.
Zhu, Z., Zeng, H., Myneni, R. B., Chen, C., Zhao, Q., Zha, J., ... & MacLachlan, I. (2021). Comment on “Recent global decline of CO2 fertilization effects on vegetation photosynthesis”. Science, 373(6562), eabg5673.
Since the line number is not covered every line, so I just give a range of line number here.
~L45 ‘However, the accuracies of the current LAI products have been limited by uncertainties primarily in the remote sensing data and the LAI reference data (Fang et al., 2019).’ Jiang et al. 2017 GCB is also an important reference here.
~ L70 to 75 It is not common to cite two papers in under review process in the introduction section. So, the authors may consider citing other published papers.
Section 2.1 and 2.2, these two datasets are not finished peer reviewed especially the Landsat LAI sample dataset. To ensure the GIMMS LAI4g is reliable, the authors should give enough evidence to show these two data sources are robust.
Section 2.3 I found the authors have not used the ground measurement LAI as directly validate for LAI4g product. I recommended that the authors may use the ground measured LAI dataset in Xu et al. 2018 RSE and Ma et al. 2021 RSE to validate the LAI4g product.
Section 2.5 There are two version of GLASS-LAI v50, one is totally based on AVHRR data (http://www.glass.umd.edu/LAI/AVHRR/), this dataset didn’t use the MODIS data as data source. The other one is fully based on MODIS with 500m spatial resolution. From my experience, there are no such dataset used both AVHRR and MODIS data as input to generate GLASS LAI product, so the authors should ensure this point.
Section 2.6 There are two version of GLOBMAP product. The previous version showed that the LAI is decreasing after 2000 (see Jiang et al. 2017 GCB figure1). The current version is GLOBMAP V3, which showed an increasing trend after 2000. So which one is the reliable one, it would be an open question. The authors may consider not using this dataset to do the inter-comparison.
Section 3. A flow chart of how to generate LAI4g is useful.
Figure 1a I guess the SVA should be SAVanna? It would be more common to use the abbreviation SAV to represent savanna.
Figure 1c will the LAI distribution of different data source affect LAI4g? The Landsat LAI showed a peak around 0.5 m2 m-2, but other product didn’t show a similar pattern.
Figure 3 the saturation effect to LAI4g is significant (see my comment at major point #2).
Figure 4 There is data missing (in white color) at southwest China, Euro and southern part of USA.
Figure 6 this is a significant improvement for LAI product based on AVHRR data! Very nice result.
Figure 9 authors should add the statistics for each histogram.
Figure 10 authors may add the statistics for the global LAI trend before (pre-MODIS era) and after 2000 (MODIS era). This can provide the insight of whether the vegetated area is greening or browning.
L 435 ‘Two other LAI products, namely the GLASS LAI (1982−2018) and GLOBMAP LAI (1982−2020), also incorporated MODIS data (reflectance). ‘ This is not totally right, the GLASS-LAI product (1982−2018) is only based on AVHRR data (Xiao et al. 2017).
Xiao, Z., Liang, S., & Jiang, B. (2017). Evaluation of four long time-series global leaf area index products. Agricultural and Forest Meteorology, 246, 218-230.