AGU Advances
Peer Review History of
Cropland Carbon Uptake Delayed and Reduced by 2019 Midwest Floods
Yi Yin
1
, Brendan Byrne
2
, Junjie Liu
3,1
, Paul Wennberg
1,4
, Kenneth J. Davis
5,6
, Troy
Magney
1,7
, Philipp Köhler
1
, Liyin He
1
, Rupesh
Jeyaram
1
, Vincent Humphrey
1
, Tobias
Gerken
5
, Sha Feng
5
, Joshua P. Digangi
8
, Christian Frankenberg
1,3
1
Division of Geological and Planetary Sciences, California Institute of Technology, Pasadena, CA
2
NASA Postdoctoral Program Fellow, Jet Propulsion Laborato
ry, California Institute of Technology, CA, USA
3
Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA
4
Division of Engineering and Applied Science, California Institute of Technology, Pasadena, CA, USA
5
Department of Meteorology and
Atmospheric Science, The Pennsylvania State University, University Park,
PA, USA
6
Earth and Environmental Sciences Institute, The Pennsylvania State University, University Park, PA, USA
7
Department of Plant Sciences, University of California, Davis, Davis,
CA, USA
8
Langley Research Center, National Aeronautics and Space Administration, Hampton, VA, USA
Files Uploaded Separately
Original Version of Manuscript (
2019AV000140
)
First Revision of Manuscript
[Accepted]
(
2019AV000140
R
)
Peer
Review Comments on
2019AV000140
Reviewer #1
Yin et al.’s manuscript demonstrate the detectability of the 2019 flood impact on vegetation productivity
in the US Midwest cropland, combining space-based observations, solar-induced chlorophyll fluorescence
(SIF) by TROPOMI and XCO
2
by OCO-2, and aircraft CO
2
measurements. They present that the extreme
2019 Midwest flood resulted in the shift of SIF seasonal cycle and the reduction of peak value over the
cropland. Based on TOROPOMI SIF, the anomalies of gross primary production and net CO
2
flux (NEE) are
estimated. The estimated positive NEE anomaly in June-July 2019 are examined for the consistency with
the two independent CO
2
measurements, one from ground-based aircraft measurements, ACT-America,
other from space-based measurements, OCO-2, employing an atmospheric transport model.
The manuscript is well structured and reasonably concise. It is a timely topic to utilize space-based
observations to constrain net carbon fluxes on a regional scale and country scale. It is also of broad interest
to monitor impacts of extreme events on terrestrial carbon cycle at a near-real time. Thus, this study is
suitable for AGU Advance. For all the above reasons, I suggest the paper be accepted pending minor
revision. My comments are mainly regarding the top-down and bottom-up analysis on NEE anomaly in
June-July 2019.
Minor comments
In Figure 3(C), besides the summertime difference of OCO-2 XCO
2
between the zonal and the eastern US
domain (Corn Belt and its downwind area) means, more differences are apparent. For example, there are
some off-zero (positive) periods for both means, and negative period (May) in the eastern US domain
mean, which is lower than the zonal mean. OCO-2 XCO
2
are de-trended. What cause these differences
between 2019 and 2018? Are they noise or uncertainty in this analysis? The zonal mean is also enhanced
in July, as well as the eastern US domain. If these zonal enhancements are real, they should be a
background contribution to XCO
2
over the domain of interest and could be a bias to the top-down ΔXCO
2
analysis.
The agreement of slopes for both OCO-2 and ACT-America is impressive. However, the slope for OCO-2
XCO
2
(0.37) has larger uncertainty than ACT-America CO
2
(0.07). The authors do not explicitly mention
possible contributions to the XCO
2
slope uncertainty. Long-term transport could be one possibility.
Another possibility might be inherent in the optimized 2018 NEE. Figure S5 shows that OCO-2 data-model
mismatch of June-July for 2018 and 2019. Since 2018 OCO-2 XCO
2
data are assimilated to estimate the
2018 fluxes, one could expect the mean data-model mismatch for 2018 to be close to zero. However,
Figure S5 shows that the mean data-mismatch of 2018 is -0.64 ppm. This off-zero value is relatively large,
given the mean data-model mismatch for 2019 of 0.49 ppm. What is the implication of the 2018 negative
(off-zero) mean to top-down ΔXCO
2
for 2019?
The definition of cropland (US Midwest) area seems not consistent. According to the caption of figure 5,
the observations and modelled CO
2
and XCO
2
are sampled over the area of 100-60 ̊W and 20-65 ̊N. It is a
much larger area than the one discussed earlier as Midwest cropland. For example, in Figure 3 (c) and
Page 12, Line 380-381, the domain of 75-95 ̊W and 35-50 ̊N, is discussed. Is there any reason why the
domain of analysis is expanded for the top-down and bottom-up analysis? North of 50 ̊N and up to 65 ̊N,
it is Canada and covers boreal ecosystem there, which is not apparently downwind area of the Corn Belt.
According to Page 12, Line 387, “Bottom-up ΔCO
2
and ΔXCO
2
estimates are all positive”. Why are negative
bottom-up ΔCO
2
and ΔXCO
2
seen in Figure 5?
Specific comments
Page 2, Lines 40-42: The enhancements, up to 10 ppm of XCO
2
in OCO-2, and ~1 ppm of CO
2
in ACT-
America, in Abstract, are not read clearly in the main text, section 5 (page 13).
Figure 3 (b): gCm
-1
day-1 should be gCm
-2
day
-1
.
Page 15, Line 453: “it in unclear” should be “it is unclear.”
Page 15, Line 477: What s detection limit of the satellite XCO
2
observations.
R
eviewer #
2
The manuscript by Yin et al. documented an interesting study on using multi
-
source
observation and simulated data to quantify the impact of 2019 spring wet anomaly in
the U.S. Midwest. The key idea is to use area
-
aggregated TROPOMI SIF data in both 2018
an
d 2019 to quantify the shift in seasonal trajectories and change in peak magnitudes,
and then related the SIF
-
based quantification of GPP & NEE changes to the changes in
boundary layer atmospheric CO2 and column
-
average CO2 estimates from both bottom
-
up an
d top
-
down approaches. The authors found that flood
-
induced delay in crop
planting shifted the 2019 SIF seasonal cycle by 16 days and reduced the peak value by
~15% compared to 2018, and the decrease of cropland NEE was consistent with increase
in atmosphe
ric CO2. Overall, this work is interesting and provides new insights on using
satellite
-
based observations to quantify the perturbations of hydro
-
climatological
extremes on regional
-
to
-
global carbon cycles. The manuscript is well organized and
written. I h
ave the following comments for the authors to consider.
Firstly, although the authors put "2019 Midwest floods" in the title, the definition of
floods in this manuscript seems to be not clear. Delayed planting could also be due to
field inundation or too
wet field condition (but not necessarily floods). I would suggest
the authors to consider change it to something like "wet anomaly". Also, "2019 Midwest
floods" include all the flood events happened in 2019 and over the Midwest, say the one
around the midd
le of March or those after July, which are not necessarily a direct factor
contributed to late planting.
Secondly, the authors did not provide any uncertainty information to the reported
numbers. Without the uncertainty information, we cannot really inter
pret those
increasing/decreasing rates. For TROPOMI or OCO
-
2 SIF products, what is the signal to
noise ratio in the early growing season, say in June? I would like to see some comparison
between uncertainty in SIF retrievals and the SIF difference between
2018 and 2019.
Thirdly, the definition of "crop production" and "crop yield" seems weird to me. At least,
those definitions are not aligned to the general understanding of ag
-
community. This
could be resolved by providing detailed definition or calculatio
n methods in the
appendix.
Fourthly, I would suggest the authors add an overall framework graph for the processing
of multi
-
source data and linked them with scientific goals. Considering more than 10
datasets are used in this study, having such a figure o
r table would be valuable for
readers.
Lastly, from the USDA reports, we see that delayed planting is mainly for soybean (the
authors talked about this around L443 and Table S2). For soybean, the planting or
harvested area decreased about 14%; yield decre
ased about 7.3%; and final production
decreased about 20% for soybean. For corn, planting or harvested area did not decrease
at all but with a slightly increase; yield decreased about 5.3%; and final production
decreased about 5.3%. Are there any clues fro
m the observations on this crop
-
specific
trigger?
Detailed comments:
L48: J. Liu et al., 2018
-
>Liu et al., 2018. Please check it across the whole paper, say in L70,
L82, L127.
L55: Corn Belt never goes to "southern U.S."
L56: >40% is not accurate here
. From the USDA reports, the ratios of US production to
total world production should be less than 38%.
L67: there should be a period before "SIF"?
L153: TWS from GRACE is not a climate variable.
L184: Did you use the simulated NEE in 2018 from these th
ree models? If you take the
data from a existing dataset, please clarify that and acknowledge the data source in the
acknowledge part. Otherwise, please describe the simulation protocols.
L186: Why not assimilating TROPOMI here? From my understanding, ass
imilating
TROPOMI footprint data could provide better constraints.
L187: it would be good to show the optimized scaling factors in the supplementary
materials. That will be critical for diagnose model uncertainties.
L198: there should be a period before
"linear"?
L208: how did you differentiate SIF for corn and soybean here? If not, it's actually a
mixing of these slopes.
L269
-
L270: could poor growth condition also be a factor for this decline??
L279: is this really instantaneous SIF without day length
correction?
L282: how did you define planting date using the USDA progress report (the date with
50% area planted)? To my knowledge, the USDA report is weekly, which means the
uncertainty there is about 1 week. Please clarify these points in the manuscri
pt.
L360: "crop production" is misleading here. USDA defines the crop production as the
yield times harvested area and the unit is in tons. Are you talking about grain biomass
per m2 here or NPP?
L380: were all longitudes or only those over the Corn Belt
included in the zonal mean?
L364: "crop yield" of which crop? Again, crop yield has specific meaning in agricultural
domain and it's grains in bushels per acre or kg/ha.
L371
-
L372: harvest index=grain biomass/AGB. This statement is not accurate.
L412:
Figure. 5
-
>Figure 5
L513: why is the tropomi data only available before July of 2019 at this site? Will the
authors make the data available to the public?