On August 26 this paper, written by me with Rosie Fisher and Peter Lawrence from NCAR, was published in Biogeosciences! The 'in a nutshell' story is that I spent much of my time as a postdoc at NCAR trying to understand how the Community Land Model works with respect to drought deciduous phenology, and how it works in comparison with satellite data (using LAI3g from Zhu et al 2013). After lots of false starts, we found that the CLM lets plants in dry areas put on leaves even when it's super dry out because (massive simplification here) CLM assumes that there is always groundwater available everywhere, and this groundwater can move up to where plants can access it. IRL this is likely true in some places, not true in others, but without better soil moisture data (hello, SMAP satellite!), it's hard to say. What's important is that if dryland leaf phenology is weird, that has implications for the fire cycle, the carbon cycle, and, ultimately, the climate, so we should probably try to get a better handle on this remote corner of the CLM, though it is certainly not alone in that respect. One other key point of this paper is that this issue (anomalous dry season green up in CLM) would be really hard to uncover using a standard 'benchmarking' type of system, or even something like zonal means. We actually had a hard time recognizing that this was serious until we looked at daily CLM output (CLM is usually output monthly, though it runs on a 30 minute timestep). I'm not going to summarize the paper too much here, go read it, it's Open Access, but there is more... because tiny maps in publications make me sad, and because I want people to use this paper to convince funders to fund their work (and mine!) I've posted all of the geotiffs used to make the map figures in the paper below (with permission from Biogeosciences and from Dr. Ranga Myneni for the LAI3g-derived maps). My vision is that if you study, say, Tanzania, you could, without too much heartache, download these figures, subset to your area of interest, and then include in a proposal or paper something like "See! We need more $ to study phenology in Tanzania because look how terribly this big important model does at predicting it!?"
I think that with relatively basic skills in GIS or R you should be able to download these and check them out. If you try it and have issues, please email me and I'll try to help. They're all on a grid from -180 to 180 longitude, -90 to 90 latitude, with a grid cell size of 1.25 by 0.9375. Enjoy! Note, if you do use these, please make sure to cite my paper, Zhu et al 2013 (they're both OA), and Giglio et al 2013 for the fire data. Note that the LAI3g data is available at much higher resolution (1/12 degree), if you need it, as is the GFED4 fire data (0.25 degree). Figure 1B - % cover of drought deciduous plants in CLM Figure 6 = Avg. maximum annual LAI from 1982 to 2010 (so, I calculated the max for each year then took the mean of those 29 max values). I'm not providing the difference maps just to save space, but I'm happy to provide them if you don't want to make them yourself. Figure 6A - LAI3g Figure 6C - CLM4.5BGC Figure 6E - CLM-MOD Figure 7 = Mode # of peaks in each year in the three data sets, not including spots with an annual range of <1 LAI unit. Figure 7A- LAI3g Figure 7B - CLM4.5BGC Figure 7C - CLM-MOD Figure 8 = Point-wise correlations between the different data sets. Grid cells set to zero here = not significant. Figure 8A - LAI3g & CLM4.5BGC Figure 8B - LAI3g & CLM-MOD Figure 9 = Looking at average burned area fraction per year in GFED4 and CLM. Figure 9A - GFED4 - the data is from here. Figure 9B - CLM4.5BGC Figure 9C - CLM-MOD |
ERSAM Lab ThoughtsThis is where we occasionally post things that are not research, but more than 280 characters. If you want to know what we're working on now, this would be a good place to look. Code will be linked to here and posted on github. Archives
November 2018
Categories |