By simple atmospheric COS dimensions community in this area, inversion fluxes for the a grid size try very not sure ( Quand Appendix, Fig. S9). And therefore, do not anticipate to be able to constrain fluxes on good spatial level that flux systems try sensitive and painful and you can manage not examine fluxes on unmarried-flux systems. Alternatively, i extracted and you may averaged monthly fluxes during the fifteen 1 o ? step 1 o grid structure where there’s a good GPP estimate reported out of flux towers regarding FLUXNET and you can AmeriFlux networks more than the fresh Us Cold and you can Boreal region. Our atmospherically derived GPP fundamentally agrees well (90% of time) with eddy covariance flux tower inferred mediocre GPP ( Quand Appendix, Fig. S10), subsequent supporting the legitimacy of our COS-established method.

The best guess out of yearly full GPP try 3. Here, the fresh new 36 outfit participants merely range from the of those projected out of a temporally varying LRU method (Methods). The reason being as soon as we think a temporally ongoing LRU approach (step one. Annual GPP derived using a constant LRU means is biased highest of the ten to 70% than whenever produced from temporally differing LRU values on account of large GPP in the early morning and later day throughout the late spring through june and all minutes throughout the fall owing to early spring ( Au moment ou Appendix, Fig. S11). If we take into account the dos ? mistake out of for each clothes affiliate, an entire uncertainty of our COS-dependent yearly GPP estimate will be dos.

The fresh uncertainty in our GPP estimate is mostly about 1 / 2 of brand new GPP range estimated from terrestrial activities more this region (step 1.