Jinshi-Ben-meeting 2020-04-21

[jinshijian]

Land model project

• Global Rh sites distribution
• Rh and Ra temporal coverage in South and North hemisphere
• Rh and Ra vs. soil temperature by month
• Test whether measure method affect Rh
• Test whether the partition method affects Rh : Rs ratio

[jinshijian]

• Rh : Rs ratio by measure month in North and South hemisphere

[bpbond]

I would just like to say your data visualization skills have improved so much this last 14 months!! 👏

Super interesting to split by NH/SH; there are some VERY interesting graphs above.

[jinshijian]

Thanks, that is because I learned a lot from you.

[jinshijian]

Surrogates manuscript review comments

Reviewer #1:

• I suggest the portion of the paper focused on assessing soil water content and precipitation be strengthened or dropped from the paper. A key premise of the paper is that soil moisture is a strong control over soil respiration; however, the authors did not find a strong or consistent relationship between soil respiration and soil moisture across their study sites. Because of this, it is not very meaningful to then ask the question of whether precipitation is a good surrogate for soil moisture in explaining soil respiration.

• In addition, the introduction neglects to mention several considerations for the use of precipitation in place of soil water content. For example, lines 82-98 omit the importance of actual soil properties - such as texture and depth - that, together with climate, result in soil moisture patterns over space and time. This is expanded upon somewhat in the Discussion, but again, this paragraph (L 358-371) could use some more detail, such as the actual range in soil taxonomy across these sites.

• The authors should re-think the use of model R2 to group sites. R2 is highly influenced by the range of independent variables that describe the predictor variable, and I don't think it is appropriate to use R2 for comparing model fit that have drastically different ranges of input data. This makes it difficult to interpret Figure 2, because the groupings are based partly on the trends in R2 which may simply reflect differences in annual or seasonal temperature range of the measurements going into the models rather than a real difference in respiration dependence on air or soil temperature.

• Finally, here are some line-by-line edits, most of which are minor usage and spelling corrections.

• [ ] L32-34: I realize these have a strict word limit, but these are too simple. Can they be expanded upon at all?

• [x] L 65: I think somewhere it would be good to point out the importance of temporal scale when suggesting that air temperature is a good surrogate for soil temperature. If the temporal frequency of measurements is sub-daily, i.e., hourly, and no temporal averaging is done, soil and air temperature may not be correlated. Fortunately that is not the case with these global or regional-scale models. But I think that needs to be explicitly stated for clarity.

• [ ] L 123-125: Unfortunately, volumetric soil water content from different studies may not be comparable to each other, depending on the method used. Different types of sensors can vary widely in accuracy, and this is complicated by using them in different soil types. Further, they can be difficult to calibrate accurately. So comparing volumetric moisture measurements from one study to another can be difficult - and could perhaps explain why you are not finding strong relationships with SWC in your study. See: Vaz, C.M.P., Jones, S., Meding, M., Tuller, M., 2013. Evaluation of Standard Calibration Functions for Eight Electromagnetic Soil Moisture Sensors. Vadose Zone Journal 12, vzj2012.0160. https://doi.org/10.2136/vzj2012.0160

• [ ] L176-180: As in my comments above, I do not think this analysis is particularly useful.

• [x] L405: Calling Group C 'puzzling' kind of contradicts what you say later in this sentence, when you offer three perfectly reasonable explanations. You could add a fourth: roots getting water primarily from bedrock cracks rather than soil moisture, and then respiring: Relative contribution of groundwater to plant transpiration estimated with stable isotopes | Scientific Reports [WWW Document], n.d. URL https://www.nature.com/articles/s41598-017-09643-x (accessed 4.20.20).

Reviewer #2:

• However, the depth and quality of Discussion is a bit disappointing. It appears that the authors may not be fully aware of the work that has been done on moisture-related variability in Rs (there is a lot of it, so it is easy to miss pieces of it). As the result, the discussion of this aspect is a bit crude and superficial. While more nuanced presentation may not change their main conclusion about the replaceability of SWC with P in scaling Rs in global simulations, it would certainly increase the value of this publication to the reader.
• In addition, I have a two questions about the underlying data that should be clarified in order for one to trust the data and ensuing analysis.
• [ ] 1. Evaluating whether Tair can be substituted for Tsoil, and Precipitation for SWC depends on (i) what is the relationship between these scalars to one another, and (ii) how the monthly respiration fluxes were derived. The first point is discussed in the manuscript (Figure 6), whereas the second one is not. Thus, it is hard to say how much of the observed correlation is attributable to the use of similar upscaling models in creating the database, and now analyzing it.
• [ ] 2. Figures S11, S13 and S14 report daily and per-second soil CO2 effluxes, even though the data set is supposed to be of monthly fluxes. This raises concerns whether other reported relationships might refer to instantaneous rather than monthly aggregated fluxes. As the shape of the relationship of fluxes to environmental variables changes with the interval of aggregation, this is of critical importance to the analysis.
• [ ] Due to these issues, I am unable to recommend that the study be accepted for publication at this point. However, in principle, I find the analysis of potential value to modeling community. If the underlying causes for observed patterns of variability are discussed in detail (see the numerous comments on Discussion in the annotated manuscript), I can see this as a valuable contribution. Absent this, the two main conclusions that air temperature correlates well with soil temperature, and that soil moisture does not correlate well with precipitation, may be a bit too little to warrant publication in AFM.

[jinshijian]

Q10 of Ra and Rh analysis

[bpbond]

金世(this is my Chinese name, means a golden world)

I want to ask you about this.

[jinshijian]

Measure time and collar depth results update

[jinshijian]

[jinshijian]

Question: how to remove the regression line from Agriculture and wetland

[bpbond]

For land models work, we decided won't make our own global RH data product (yet). Base land models comparisons for now:

• global total over time
• rh across months of year
• rh by latitude
• site-specific rh
• temperature sensitivity

[bpbond]

For collar height and insertion depth and area analysis:

• Use a weighted regression if using those data
• Maybe use a robust linear model (e.g. MAAS:::rlm)
• Plot collar depth versus Rs faceting by RC_annual; we would expect bigger slopes (stronger effect) in places with lots of root contribution to Rs?

Some perhaps relevant papers: http://dx.doi.org/10.5194/bg-14-1603-2017 http://dx.doi.org/10.1007/s00442-002-0870-3 http://dx.doi.org/10.1016/j.agrformet.2003.12.001 http://dx.doi.org/10.1016/j.agrformet.2008.07.012

[jinshijian]

Linear regression weight by obs Facet by RC_annual (will be very strong evidence) Measure interval: depend on the climate by biome (how variate is the weather)

[bpbond]

Question: how to remove the regression line from Agriculture and wetland

Two ways come to mind. One would be

+ geom_smooth(data = filter(x, ! Ecosystem_type %in% c("Agriculture", "Wetland")), ...)

The other would be fit the models outside of ggplot first, and then use geom_line() to add them.

[jinshijian]

Thanks!