Dissident and Pstarr
with reference to the Zhu et al, 20166. Greening of the Earth and its drivers paper it is worth reading the whole thing to understand what they did. My read is that it addresses the issues Dissident points out.
Long-term changes in vegetation greenness are driven by multiple interacting biogeochemical drivers and land-use effects9. Biogeochemical drivers include the fertilization effects of elevated atmospheric CO2 concentration (eCO2), regional climate change (temperature, precipitation and radiation), and varying rates of nitrogen deposition. Land-use-related drivers involve changes in land cover and in land management intensity, including fertilization, irrigation, forestry and grazing. None of these driving factors can be considered in isolation, given their strong interactions with one another. Previously, a few studies had investigated the drivers of global greenness trends, with a limited number of models and satellite observations, which prevented an appropriate quantification of uncertainties.
Here, we investigate trends of leaf area index (LAI) and their drivers for the period 1982 to 2009 using three remotely sensed data sets (GIMMS3g, GLASS and GLOMAP) and outputs from ten ecosystem models run at global extent (see Supplementary Information). We use the growing season integrated leaf area index (hereafter, LAI; Methods) as the variable of our study. We first analyse global and regional LAI trends for the study period and differences between the three data sets. Using modelling results, we then quantify the contributions of CO2 fertilization, climatic factors, nitrogen deposition and LCC to the observed trends.
Trends from the three long-term satellite LAI data sets consistently show positive values over a large proportion of the global vegetated area since 1982
The regions with the largest greening trends, consistent across the three data sets, are in southeast North America, the northern Amazon, Europe, Central Africa and Southeast Asia. The GLASS LAI data shows the most extensive statistically significant greening (Mann–Kendall test, p < 0.05 ) over 50% of vegetated lands, followed by GLOBMAP LAI (43%) and GIMMS LAI3g (25%). All three LAI data sets also consistently show a decreasing LAI trend (browning) over less than 4% of global vegetated land—these are observed in northwest North America and central South America.
and they also "sense check" the observations against 10 global ecosystem models
We compare satellite-based LAI anomalies with LAI anomalies simulated by ten global ecosystem models driven by eCO2 (+46ppm over the study period), climate, nitrogen deposition and LCC (Supplementary Section 7). Multi-Model Ensemble Mean (MMEM) LAI anomalies, with all these drivers considered, generally agree with averaged satellite observations at the global scale (r = 0.85, p < 0.01; Fig. 2a). The trend in MMEM LAI anomalies (0.062 m2 m−2 yr−1 ) is within the range of estimates from the three satellite data sets.
and they try to derive the contribution of CO2 to the results from the models
We used an optimal fingerprint detection method13 to assess the ability of the models to simulate response patterns of LAI to eCO2 , climate change, nitrogen deposition and LCC.
Globally, the model factorial simulations suggest that CO2 fertilization explains the largest contribution to the satellite-observed LAI trend (70.1 ± 29.4%, 0.048 ± 0.020 m2 m−2 yr−1 ), followed by nitrogen deposition (8.8 ± 11.8%, 0.006 ± 0.008 m2 m−2 yr−1 ), climate change (8.1 ± 20.6%, 0.006 ± 0.014 m2 m−2 yr−1 ) and LCC (3.7 ± 14.7%, 0.003 ± 0.010m2 m−2 yr−1)
so to sum:
-the study uses 3 recent remote sensing datasets and the trends in all 3 agree with one another over all areas
-they incorporate land cover change into their analysis (LCC) along with greening and browning analysis (LAI)
-they test their results against a climate model that shows general agreement with observations
-the rationale as to what contribution CO2 plays in the observed greening comes from analysis of the models