- Email: [email protected]
Unusual phytoplankton bloom phenology in the northern Greenland Sea during 2010
Unusual phytoplankton bloom phenology in the northern Greenland Sea during 2010 Bo Qu, Albert J. Gabric, Zhifeng Lu, Li Hehe, Zhao Li PII: DOI: Reference:
S0924-7963(16)30217-2 doi: 10.1016/j.jmarsys.2016.07.011 MARSYS 2856
To appear in:
Journal of Marine Systems
Received date: Revised date: Accepted date:
22 January 2016 31 May 2016 22 July 2016
Please cite this article as: Qu, Bo, Gabric, Albert J., Lu, Zhifeng, Hehe, Li, Li, Zhao, Unusual phytoplankton bloom phenology in the northern Greenland Sea during 2010, Journal of Marine Systems (2016), doi: 10.1016/j.jmarsys.2016.07.011
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Unusual phytoplankton bloom phenology in the northern
RI P
T
Greenland Sea during 2010 QU Bo (瞿波) 1, Albert J. GABRIC2, LU Zhifeng3, Li Hehe4 ，Zhao Li5 13,4,5
.School of Science, Nantong University, Nantong, 226007, China
2
SC
. School of Environment, Griffith University, Nathan, QLD, 4111, Australia
Abstract: Arctic marine ecosystems are disproportionately impacted by global warming. Sea ice plays an
MA NU
important role in the regional climate system and the loss of perennial sea ice has diverse ecological implications. Here we investigate the causes of an unusually early and strong phytoplankton bloom in the northern Greenland Sea (20°W-10°E, 75°N -80°N) during the 2010 season. In order to better understand the anomalous bloom in 2010 we examine the correlation between satellite-derived biomass and several possible environmental factors for the period 2003-2012. Results show that the timing of sea ice melt played an important role in promoting the growth of phytoplankton. Multivariate lagged regression analysis shows that phytoplankton biomass (CHL) is correlated with ice concentration(ICE) and ice melting, as well as sea surface temperature (SST) and photosynthetically
ED
active radiation (PAR). During 2010, the spring peak in biomass came much earlier and achieved a higher value than most other years in the satellite archive record, which was due to earlier and more extensive sea ice melt in that year. Relative lower SST and PAR in spring and early summer in year 2010 associated with a persistent
PT
negative North Atlantic Oscillation (NAO) index were possible drivers of the bloom. Wind direction changed from the southeast to southwest direction in spring, possibly transporting nutrient enriched melt runoff from glaciers on Greenland and other sources from the south to northern coastal regions.
CE
Keywords: chlorophyll-a; sea ice concentration; North Atlantic Oscillation (NAO); Arctic Oscillation (AO);
1.
AC
Greenland Sea (GS)
INTRODUCTION The Greenland Sea (GS) is one of the most productive regions in the Arctic (Arrigo and
Van Dyiken, 2011) and adjacent to the world’s second largest glacier in Greenland. Glaciers in the northeast of Greenland are melting faster than expected (Glasser et al., 2011). The GS is an important area for water mass exchange between the North Atlantic Ocean and the Arctic Ocean. It is also the area to where most Arctic drifting ice is advected (Cherkasheva et al., 2014). Hence, the GS is an appropriate region for studying the relationship between sea ice and phytoplankton dynamics and where an extensive archive of in situ and satellite-derived chlorophyll data is available (Arrigo et al., 2011).
Surface Currents
E-mail: [email protected], This work is funded by National Nature Science Funding No. 41276097.
1
ACCEPTED MANUSCRIPT The surface currents in the GS are shown in Figure 1, within our study region is highlighted by the red box. The East Greenland Current (EGC) moves from north to south along Greenland’s eastern coastline bringing colder less saline Arctic water to the south. From south-east of Iceland, warmer more saline Atlantic water flows to north merging with the Norwegian current and
T
flowing into Arctic ocean. In this area, the vertical stability of the water column increases to the
RI P
north due to the input of melt water and solar heating, causing phytoplankton biomass to increase and nutrient concentration to decrease (Lara et al., 1994). Between 70°N -80°N, there is an anticlockwise gyre, affecting our study region, and at around 70°N, the EGC branches into two parts,
SC
one flowing along the west coast of Norway and east into the Barents Sea, and the other northwards to the Spitsbergen region. The Polar front is located to the east of EGC and the Arctic
MA NU
front is located west of the Norwegian Current.
Sea ice melt, runoff, iron and phytoplankton dynamics
The decline of Arctic sea ice (and concomitant decrease in surface albedo) in recent decades has resulted in a regional temperature increase in the Arctic, where sea surface temperatures have increased at twice the global average rate and could continue to increase throughout this century (Chalecki, 2007). Melting of Greenland’s ice sheet has increased six-fold
ED
over the last decade ago, according to a draft of the UN’s most comprehensive study on climate change (Stroeve et al., 2011). Greenland may add a total of 4 - 21 centimeters to global sea levels
PT
by the end of the 21stC (Schuenemann and Cassano, 2010). The indirect effect of melting sea ice is an increase in regional average temperatures, which may accelerate the melting of the Greenland ice sheet and lead to global sea level rise. Arctic sea ice concentration has retreated extensively
CE
and first-year ice is thinning, making it more vulnerable to summer melting and sea level rising . Summer sea ice could be totally gone by 2030 (Stroeve et al., 2011).
AC
Several studies have examined the impact of sea ice melt on regional phytoplankton biomass (Matrai et al., 1997; Wassmann et al., 1999; Olli et al., 2002; Qu et al., 2006; Pabi et al., 2008, Leu et al., 2011). It is suggested that decreasing sea ice extent and thickness and a concomitant increase in water column illumination, leads to an increase in phytoplankton biomass. Moline et al. (2008) point out that the less saline surface water during ice melt can stimulate primary production. The melting ice increases the area of open water, increasing absorption of solar radiation and could enhance the melt process. In July 2012, the NASA ICESCAPE project discovered large under ice blooms appeared in Arctic water due to thinning ice and proliferation of melt ponds (Arrigo et al., 2012). Phytoplankton was extremely active and growth rates were the highest ever measured in polar waters. Arrigo et al. (2012) suggest that satellite-based estimates of annual primary production in Arctic waters may be underestimated up to 10-fold due to under ice blooms. In contrast to the Southern Ocean where primary production is iron limited, the Arctic Ocean is generally land-locked, with higher levels of atmospheric deposition of micro-nutrients such as iron, which can stimulate primary production during and after ice melting (Moline et al., 2008). 2
ACCEPTED MANUSCRIPT It is suggested that glacial runoff serves as a significant source of bioavailable iron to the surrounding oceans (Bhatia et al. 2013). The Greenland ice sheets are also likely a significant source of iron to the adjacent ocean (Hawkings et al., 2014). The availability of Iron (Fe) is said to be the main factor controlling primary production in high nutrient (N, P, Si) and low CHL (HNLC)
T
systems (Martin and Fitzwater, 1988). An alternative source of Fe to the ocean from glacial runoff
RI P
is associated with glacial particles (Smith et al., 2007). These glacial melt waters provide an important supply of Fe and other micro nutrients to surface Arctic waters (Dierssen et al., 2002; Statham et al. , 2008)
SC
Primary production is high in the coastal zone of Greenland, due to the impact of melting sea ice (Rysgaard and Nielsen, 2006). Macronutrients (N, Si, and P) appear to control primary production (Nielsen and Hansen, 1999), although Fe could be a limiting nutrient (Blain et al.,
MA NU
2004). During winter storms, vertical mixing of high Fe content deep water, together with the lateral entrainment of high Fe surface water from Greenland coast could contribute to relatively high Fe concentrations at the beginning of the phytoplankton bloom period. The melt water input from Greenland glacier inputs is around 10% of sea ice melt water, although sea ice melt would occur earlier in the season (Statham et al., 2008).
Following on a previous analysis of the spike of phytoplankton biomass in the GS during
ED
2009 (Qu et al., 2014), more recent satellite data indicated elevated phytoplankton biomass in the northern Greenland Sea during 2010 when compared to data for the ten year period 2003-2012.
PT
Here we examine the factors that may be responsible for higher phytoplankton biomass in 2010. Apart from the effect of melting ice on phytoplankton biomass, other factors such as sea surface temperature (SST), wind speed (WIND), photosynthetically active radiation (PAR), and climate
CE
variability as indicated by the North Atlantic Oscillation (NAO) are also considered.
AC
2. MATERIAL AND METHODS 2.1 Data Sources
Our study region is the northern Greenland Sea (20°W-10°E, 75°N -80°N) (Figure 1), for the period 2003-2012. Due to the Arctic sunset, satellite data are only available between March and September. MODIS (Aqua) satellite, 8-day, 4-km, level 3, mapped data Aerosol Optical Depth (AOD), Chlorophyll-a (CHL) and Photosynthetically Active Radiation (PAR) global data was archived (modis.gsfc.nasa.gov/). Sea ice concentration (ICE) is from the following archive iridl.ldeo.columbia.edu/SOURCES/.NOAA/.NCEP/.EMC/.CMB/.GLOBAL/.Reyn_SmithOIv2/. Wind speed and direction, and sea surface temperature (SST) were obtained from www.remss.com/windsat.
PAR
is
derived
from
SeaWiFS,
8-day
mapped
data
(oceandata.sci.gsfc.nasa.gov/seawifs). The image analysis package data analysis system SeaDAS 6.4 (seadas.gsfc.nasa.gov/) is used to subset data for our study region. Mean values for each (1°x1°) grid cell are calculated first,
3
ACCEPTED MANUSCRIPT with missing values excluded. Regional mean values of the CHL，AOD and PAR are calculated by averaging the 150 (1°x1°) grid cells. Mean values of the wind and direction of each (1°x1°) grid are calculated from the 0.25°x0.25° grid. The weekly data windAW is chosen for wind speed data. windAW is 10 meter
T
surface wind for all weather conditions made using 3 algorithms (Meissner et al. 2009) with rainy
RI P
condition included. The missing values are excluded again in the mean value calculations. The CHL, AOD and PAR could be slightly over predicted due to the exclusion of missing values . However, the higher CHL spots obviously occurred in early summer in the study region as （day 184） 2010 and May (day 160) 2011.
SC
indicated in the satellite images (Figure 2). Figure 2 shows the CHL satellite images in early June
Eviews statistical software is used for correlation and lagged regression analysis. R
MA NU
software is used to do the partial correlation analysis and multivariate regression analysis among
AC
CE
PT
ED
mean time series of CHL, SST, WIND, ICE and PAR in the study region.
Fig. 1 Map of Study region in Greenland Sea (highlighted box is for 20°W–10°E, 75°N–80°N). White arrows indicate the surface flow directions.
(a)
(b)
Fig. 2 Satellite image of CHL (a) on day 180 (2010) and (b) day 160 (2011) around the study region.
4
ACCEPTED MANUSCRIPT 2.2 Accuracy of the Satellite data Due to the remoteness of the Arctic Ocean, the satellite data is the only means of obtaining synoptic coverage. However, the accuracy of these data directly relate to the reliability of this
T
study. Surface chlorophyll concentration (CHL) is from the MODIS AQUA sensor with data
RI P
available from 2002. NASA carried a polarization correction for MODIS productions and improved the accuracy, especially for the Arctic region. Calibrations were done for regional differences between MODIS (TERRA & AQUA) with SeaWiFS (Bailey and Werdell, 2006). At
SC
high latitudes, multiple orbits are considered for a given in situ record.
However, there is still an unavoidable error that can occur for solar zenith angles more than 70° and view angles more than 45° (Bailey and Werdell, 2006). Due to cloud and ice cover, the
MA NU
data is more accurate during summer time compared to early spring and winter. It is reported that the satellite-derived CHL underestimates the concentration in the field by a factor of 1.4 when compared with in-situ CHL data averaged over the optical penetration depth (Bailey and Werdell, 2006). The relative biases for PAR retrieval are 4.6% for all sky and 2.9% for clear sky (Su et al., 2007). Notice that MODIS PAR is cloud corrected. Satellite-to-in situ match-ups for MODISA (MODIS instrument on board Aqua) were evaluated for CHL in the western Arctic Ocean, and the
ED
errors are well within the range of errors for global data (Chaves et al., 2015). Another possible source of error is the computation of regional means due to missing data.
PT
Bearing in mind that computed study region mean values could be higher than reality. Although the errors of the satellite data in Arctic Ocean are unavoidable, more accurate satellite data in the Arctic Ocean is expected in the future. However, our results indicate that the unusual nature of the
AC
3. RESULT
CE
phytoplankton bloom happened in year 2010.
3.1 CHL and Ice Melt distributions A ten-year climatology (2003-2012) derived from 8-day data (Figure 3) shows a gradual increase in CHL from March, reaching a late spring peak on day 168 (mid-June), followed by a small dip on day 184 (early July), and a second peak on day 192 (mid-July), steadily decreasing through to the end of September. Figure 3 also shows the CHL time series in year 2010 when the spring and summer peaks of CHL were more than double the long-term average. There were unusually high peaks on day 128, and 184. It is interesting to note the high interannual variability in CHL during early spring, possibly related to interannual variability in sea ice concentration (Qu et al., 2014).
5
Mean CHL climatology (with standard deviation) compared with CHL in year 2010 in the study region
MA NU
Fig. 3
SC
RI P
T
ACCEPTED MANUSCRIPT
(75°N-80°N, 20°W-10°E).
Comparing to average zonal mean, year 2010 exhibited a strong meridional gradient (not shown in figures)..
The time series of monthly CHL and sea ice concentration (ICE) in the study region is
ED
shown in Figure 4. There are apparently strong negative correlations between CHL and ICE from April to June. CHL increased in spring with the decreased ICE. There is a clear positive trend in the amplitude of the CHL spring bloom during 2003-2012. CHL increases as sea ice concentration
PT
decreases during spring and early summer, however the timing of the CHL bloom displays considerable interannual variability, occurring in either May (2003, 2010), June (2004, 2006, 2007
CE
and 2011) or July (2005, 2008, 2009, 2012). A doublet peak of CHL in year 2010 in both spring and early summer was distinguished from other years. The general increased trend is shown in the figure with a positive slope. For spring and early summer period, the increasing rate was much
AC
more significant. Mean CHL increased 10.8% during spring (before day 168) within year 20032012 in the study region (not show in figure).
Fig. 4 Monthly mean CHL and sea ice concentration (ICE) in the study region with trend line (dashline) for CHL.
6
ACCEPTED MANUSCRIPT Ice melt is computed by the percentage change in sea ice concentration between octads. The 10-year (2003-2012) climatological mean sea ice melt is compared with that in year 2010. Ice melt in 2010 was higher than the climatological average during most of spring and from July to midAugust . In other years, ice melt did not commence until May (e.g 2007, 2008, 2009 and 2012).
T
However, 2010 saw consistent ice melt from April to August. The time series for CHL and Ice
RI P
melt in 2010 is shown in Figure 5. The first CHL peak of 2010 occurred in early May. From late
MA NU
SC
April (day 112) until mid-June (day 160), CHL tracks the change in Ice melt.
ED
Fig. 5 Mean CHL and Ice Melt time series in 2010 in the study region (75°N-80°N, 20°W-10°E).
3.2 Impacts of SST, PAR and Wind
SST gradually increases from a minimum in March to a maximum in July in 2010. SST in
PT
2010 was 8.45% lower than the climatological mean during spring and early summer. The PAR profiles had normal distributions with the similar pattern of higher in summer and lower in winter.
CE
Interestingly, similar to SST, PAR was 13.6% lower than the climatological average during spring
AC
and early summer in 2010.
Average wind speed (bars) and wind direction (curves) for year 2003-2012 and year 2010 are compared in Figure 6. Although there was a much higher wind speed occurred in early spring in year 2010, year 2010 had lower wind speed in general. Lower wind speed favors phytoplankton growth. There is an inverse relationship between wind speed and CHL (Fitch and More 2007). Wind direction (the gcurve lines in the figure) generally was from the southeast direction changing to a southwest direction in March and early April (day 71-day 99). This possibly brought melt water from south to north and together with glacial runoff water from the east coast of Greenland.
7
SC
RI P
T
ACCEPTED MANUSCRIPT
b
÷
MA NU
Fig. 6 Mean wind speed (bars) and direction (lines) for year 2003-2012 compared with those in year 2010 in the study region.
Positive (negative) NAO is usually described as a stronger (weaker) north-south pressure gradient between the Azores high and the Icelandic low (Hurrell, 1995, Qu et al., 2012). Positive NAO is associated with cool summers and mild and wet winters in Central Europe. In contrast, a
ED
negative NAO indicates the cold air comes from Europe down south and milder temperatures in Greenland (Hurrell, 1995). AO (Arctic Oscillation) is a large scale mode of climate variability, also
referred
to
as
the
Northern
Hemisphere
annular
mode
(https:
PT
//www.ncdc.noaa.gov/teleconnections/ao/). The positive AO indicates a ring of strong winds circulating around the North Pole that confining colder air across polar regions. NAO and AO
CE
indices for the period 2003 -2012 are shown in Figure 7. NAO and AO had inter-annual variations and showed similar patterns. In contrast to most other years, the NAO during 2010 was negative
AC
throughout the year (shown in red colour in the figure). The AO was mostly negative during 2010 especially in the winter. Milder conditions in Greenland would lead to more glacial ice melt from the east coast of Greenland to the GS. Stroeve et al. (2011) analyzed the winter of 2009-2010 and found sea ice by September 2010 was the third lowest in the satellite record. They also pointed out that negative AO indicated sea ice motion tends to have clockwise anomaly, with enhanced ice transport from the western to the eastern Arctic. The warm southerly winds favors ice melt and transport ice poleward, with enhanced ice export through Fram Strait in a northerly direction. That may explains the higher ice melt during 2010 in the northern Greenland Sea, and hence higher and earlier CHL blooms occurred in our study region.
8
ACCEPTED MANUSCRIPT 3 2
T
-1 -2 -3
NAO
-4 -5
AO
Ja n0 Ju 3 l-0 Ja 3 n0 Ju 4 l-0 Ja 4 n0 Ju 5 l-0 Ja 5 n0 Ju 6 l-0 Ja 6 n0 Ju 7 l-0 Ja 7 n0 Ju 8 l-0 Ja 8 n0 Ju 9 l-0 Ja 9 n1 Ju 0 l-1 Ja 0 n1 Ju 1 l-1 Ja 1 n1 Ju 2 l-1 2
RI P
NAO, AO
1 0
SC
Month
Fig. 7 NAO (solid dot) and AO (hollow dot) monthly mean for year 2003-2012. Red part indicates NAO values in
MA NU
year 2010.
3.5 Correlation between CHL and sea ice concentration (ICE) The correlation coefficients for CHL and ICE for each year are shown in Table 1. Generally, they had negative correlations with year 2009 much more significant than other years. Table 1 Cross-correlation coefficient for CHL and ICE 2005
0.03
-0.21
-0.20
2006
2007
2008
2009
2010
2011
2012
-0.33
0.11
0.07
-0.53
-0.10
0.05
-0.43
ED
2004
PT
2003
The peak of CHL is about 3 months behind of ICE in the study region. The correlation coefficients of CHL and ICE before and after shifting in year 2010 in the study region range from
CE
0.13 to 0.6. The statistical software package EViews (Pang, 2007) is used to do more accurate correlation and regression analysis between CHL and ICE.
AC
The regression equation for CHL and ICE is as follows， and it is significant: CHL = -0.04+0.01ICE
(75°N–80°N, 20°W–10°E))
(2)
Table 2 shows the ICE(-3) (ICE shifted 3 months behind) had lowest P-value (Prob.=0.0003) for CHL regression coefficient test. Hence, ICE(-3) had the most significant influence on CHL.
9
ACCEPTED MANUSCRIPT Table 2 Lagged regression analysis result for CHL and ICE (Dependent variable is CHL) Coefficient
Std. error
t-Statistic
Prob.
ICE(-1)
0.0103
0.0035
2.9376
0.0047
ICE(-2)
0.0106
0.0041
2.5925
0.0119
ICE(-3)
0.0159
0.0041
3.8404
0.0003
ICE(-4)
-0.0062
0.0035
-1.7706
The goodness of fit: R =0.59
F-statistic: 21.9454
0.0816
RI P
2
T
Variable
SC
Unit root test of CHL was performed and results show that CHL has unit root and is nonstationary sequence (Qu et al., 2014). The same method is used to test the first-order differential sequence of CHL, which is also a stationary sequence. The regression residuals sequence test
MA NU
result shows that t-test statistical value is also less than 3 corresponding critical values. Again, it shows the residuals sequence did not have unit root, it was stationary sequence. Hence, CHL and ICE were co-integrated or had a long-term equilibrium relationship. R software is used to do the partial correlation analysis among mean time series of CHL, SST, WIND, ICE and PAR in the study region. Pearson's product-moment correlation method is
ED
used. The results are listed in Table 3 in the study region.
Table 3 Partial correlation coefficient among CHL, SST, WIND, ICE and PAR WIND
ICE
PAR
0.15
-0.11
0.15
0.32
1
-0.12
-0.67
0.24
-0.12
1
0.27
-0.47
0.15
-0.67
0.27
1
0.44
0.32
0.24
-047
0.44
1
1
SST
0.15
WIND
-0.12
ICE
AC
PAR
CE
CHL
SST
PT
CHL
Table 3 indicates that CHL had more relationship with PAR (coefficient is 0.32). Other relationships were generally weak. Granger causality analysis shows that PAR more likely influenced CHL. ICE and CHL had positive relationship and they influence each other. SST had weak positive relationship with CHL. However, SST has less influence on CHL. WIND had weak negative relationship with CHL.
A multivariate regression linear model is applied using R software (R3.2.3) for years 20032012. There are different regression equations with different time lags for each year. Table 4 listed the more significant regression equations within certain variables for year 2003, 2004, 2005,, 2007, 2009 and 2011. In other years the correlations are not significant. In general, CHL had higher correlation with ICE. In some years, CHL was correlated with Ice Melt as well as PAR or SST. The standard deviations and P-values of according variables in the equations are listed. In general, most of the standard deviations are low and P-values are less than 0.05. The regressions are mostly
10
ACCEPTED MANUSCRIPT significant. The goodness of fitting (R-squared) are within 0.72-0.92. That means that the listed lagged regression equations are quite correlated. The lagged regression equation for the mean of years 2003-2012 is as follow: CHLt 9 1.4221 - 0.0265PARt
T
(4)
The residual standard error is 0.106 and R-squared value is 0.8629, P-value is 0.000127.
RI P
The CHL had more correlation with PAR for the 10 years. For year 2010, the CHL was more dependent on ICE, Ice Melt and PAR. The regression equation for year 2010 is as follow: (5)
SC
CHLt 15 -4。 2715 0.7048ICEt - 6.061IceMelt t 0.0262PARt
The CHL was 15 octads （120 days） behind ICE, Ice Melt and PAR. The goodness of fit (R-squared) is 0.9843, The P-values are less than 0.05, which means the equation (5) is
MA NU
significant. The multivariate lag regression results are listed in Table 4 for those more significant years. The standard deviation (std.) and P-value of each variable are listed underneath correspondent variables. The goodness of fit (R-squared) are within the range of 0.72-0,92. Table 4
Multivariate lag regression analysis for some significant years Regression equation and correspondent std. and P-Value
2003
CHLt 5 0.4572 0.0073ICEt - 0.0111PARt
ED
Year
Std
0.0022
0.00006
0.00027
0.00019
CHLt 7 2.8582 - 0.0319ICEt - 0.299SSTt Std
0.8823
0.0156
0.0727
P-Value
0.0089
0.0677
0.0021
CE
2004
CHLt 11 1.0505 - 0.0177 PARt - 0.0396IceMelt t
AC
2005
2007
0.0015
PT
P-Value
0.0811
Std
0.0943
0.0037
0.0210
P-Value
0.00001
0.0205
0.1015
CHLt 9 1.9804 - 0.0493PARt - 0.108IceMelt t Std
P-Value
0.1277 0.00000008
0.0343
0.0000093
0.0119
CHLt 7 2.8386 - 0.0242ICEt - 0.108IceMelt t 0.1846SSTt
2009
Std
0.6839
P-Value 0.00198 2011
0.0055
0.0114
0.0078
0.0598
0.0048
0.0366 0.0005
CHLt 7 -15.2689 - 0.2834ICEt 0.3276IceMelt t 0.2713SSTt - 0.05PARt Std
6.1244
0.1001
0.0972
0.1598
0.0184
P-Value
0.0343
0.0197
0.0083
0.1239
0.0237
3.6 Correlation between Ice melt, NAO and CHL Ice Melt had a better correlation with NAO compared with the relationship between sea ice concentration (ICE) and NAO. There was a positive correlation relationship during spring (around March to May) between Ice Melt and NAO time series, although it is not consistent positive 11
ACCEPTED MANUSCRIPT related. This result is confirmed by statistical software EViews. Figure 8 shows the weekly time series for Ice Melt and NAO in year 2010. However, from day 112-208 (March-June), they were mostly positive correlated. The correlation coefficient between Ice Melt and NAO during this period is 0.56 and it is significant. The positive correlation indicates that, with the increasing of
MA NU
SC
RI P
T
NAO, Ice Melt would increase in spring 2010.
Fig. 8 Weekly time series of Ice Melt and NAO in year 2010 in the study region ( 75°N -80°N, 20°W-10°E).
In general, Ice Melt was negatively correlated with NAO. Eviews regression analysis gives
ED
result of the regression equation for CHL, ICE-Melt and NAO in the 10 years: CHL=0.054–0.111NAO–0.012IceMelt(-3)
(6)
The regression equation is significant. Therefore, Ice Melt and NAO had 3 months’ time lag
PT
with NAO 3 months ahead of Ice Melt, suggesting NAO and Ice Melt had significant influence on
CE
CHL.
4. DISCUSSION
AC
The elevated CHL in northern part of Greenland Sea is unusual and the significant rising in year 2010 related to many oceanography and biological reasons. In early years, the unusual phenology in northern GS was discovered by a few researchers. Lara et al. (1994) discussed the mechanisms of nutrient supply and factors influencing phytoplankton distribution in the northeast GS (78°N-82°N, 20°W-0), and found that vertical stability was higher at these latitudes due to the input of melt water from Greenland, with a corresponding reduction in salinity and increased heating of the water column near 80°N. Similarly, Cherkasheva et al. (2014) did a study in the Fram Strait area (76°N -84°N, 25°W-15°E) and found that late sea ice retreat leads to a late iceassociated bloom in the northern GS. In ice-free water, solar heating should be similar in the north and south of the region. Input of melt water could make the difference. They also found that the stratification of the surface water due to increased solar radiation after ice melt coincided with an increase in CHL in late April and early May. Patrick Lockerby firstly found that spring and early summer had large extent of ice breakup north
of
80°N
in
2010
in
Greenland
Sea
(http://www.science20.com/chatter_box
/arctic_ice_may_2010). Massive ice break-up extended from the west to northeast of Greenland
12
ACCEPTED MANUSCRIPT Sea (Fram Strait).due to the rapid ice melt in bottom layer. In May 2010, the ice flow north of Greenland was extensive and almost 10 times above the normal situations. In contrast to the Southern Ocean where iron limits primary production, the Arctic Ocean is supplied with trace elements from river runoff and from the atmospheric dust deposition leading to the incorporation
T
of Aeolian sediments in sea ice. The melting of ice would thus be a possible source of iron for
RI P
primary production (Smetacek and Nicol, 2005).
Stroeve et al. (2011) also found that year 2010 had higher SST in spring and summer. The ice concentration reached the third lowest in the satellite record by September. indicating that
SC
much more ice melt occurred in year 2010. Straneo et al. (2011) examined runoff from the Helheim glaciers in year 2009 and 2010 in southeast of Greenland. Their findings indicated that a large amount of melt water and runoff entered the ocean at depth, affecting water column
MA NU
stratification.
In general， CHL increased after ice melt. However, CHL would reach to a peak in sometime during early summer. The reasons of halting the CHL are complicated. Apart from Ice melting, SST、light, wind speed and direction, grazing of zooplankton, nutrients are one of important factors. The depletion of nutrients likely halts the bloom of CHL. The nutrients include nitrate, phosphate, silicate and iron. Inverse relationship between SST and nutrients (Temperature-
ED
Nitrate relationship) were found, although the relation is non-linear 6 C (Henson et al., 2003). The nitrate consumption occurs before SST reached to 6 C (Henson et al., 2003) . With the
PT
temperature rises, nutrients are consumed by phytoplankton growth (Henson et al. 2006). In Arctic and North Atlantic Oceans, diatoms are often dominate the spring blooms ((Matrai 1997),
CE
(Henson et al. 2006)). Diatoms rely on silica to form their frustules. Once silica depleted, diatom dominance ends. Nitrate and phosphate do not become depleted during spring bloom, silica is the
AC
limiting factor of the blooms. Later in grow season (autumn and winter), nutrients regenerated from organic material synthesized, the autumn blooms usually occurs (Qu and Gabric，2010)
5. CONCLUSION We investigated the phenology of the CHL bloom in year 2010 in the northern GS (70°N80°N). The unusual CHL bloom phenology in 2010 could be due to the following combination of factors: the early ice melt, and change of wind direction between southeast and southwest direction during the melt season possibly brought melt water from the south to north and runoff water (with enriched iron content) from the Greenland coast which was advected in a northeast direction oceans (near 79°N-80°N). The consistently negative AO and NAO index in year 2010, corresponding to the milder Greenland air temperatures, which likely caused more clockwise direction glacial ice melt and drift from south to north. Relative lower SST and PAR in spring and early summer also favors phytoplankton blooms. All of above factors contributed to the water column vertical stability, hence favoring phytoplankton growth.
13
ACCEPTED MANUSCRIPT More accurate satellite data is urged for further research. Improved Arctic Ocean colour satellite data, especially cloud-free data coverage, will provide more accurate analysis on interannual variations of phytoplankton distribution in this remote high latitude Arctic Ocean.
T
ACKNOWLEDGEMENT
RI P
We acknowledge NASA’s Ocean Biology Processing Group for providing MODIS aqua, Level 3 (4-km equi-rectangular projection) 8-day mapped data for aerosol optical depth (AOD) and chlorophyll-a (CHL) And Photosynthetically Active Radiation (PAR) global data. Thanks to
SC
NASA Web SeaDAS development group for providing Ocean Colour SeaDAS Software (SeaWiFS Data Analysis System) for processing regional CHL, AOD and PAR data. NOAA Reyn-SmithOIv2 provided weekly and monthly sea-ice concentration. Sea Surface Temperature
MA NU
and Wind Speed (Wind Direction) global data are produced by NASA Remote Sensing Systems and sponsored by the NASA Earth Science MEaSUREs DISCOVER Project and the NASA Earth Science Physical Oceanography Program. Thanks to NASA http: //gdata1.sci.gsfc.nasa.gov for providing cloud cover data.
We are grateful to the Chinese National Natural Science Foundation of China (Funding No.
ED
41276097) for providing funding for this project.
REFERENCE Res. 116, C09011.
PT
Arrigo K R, van Dijken G L. 2011. Secular trends in Arctic Ocean net primary production. Journal of Geophys.
CE
Arrigo K R, Matrai P A, van Dijken G L. 2011. Primary productivity in the Arctic ocean: impacts of complex optical properties and subsurface chlorophyll maxima on large-scale estimates. Journal of Geophysical Research, 116, C11022.
AC
Arrigo K R, Perovich D K, Pickart R S et al. 2012. Massive phytoplankton blooms under arctic sea ice. Science, 336 (6087): 1408. Bailey S W, Werdell P J. 2006. A multi-sensor approach for the on-orbit validation of ocean color satellite data products. Remote Sensing of Environment, 102: 12-23. Bhatia, M.P., Kujawinski, E.B., Das, S.B., Breier, C.F., Henderson, P.B., & Charette, M.A. (2013). Greenland meltwater as a significant and potentially bioavailable source of iron to the ocean. Nature Geosci, 6, 274278 Blain S, Guieu C, Claustre H, Leblanc K. 2004. Availability of iron and major nutrients for phytoplankton in the northeast Atlantic Ocean. Limnology and Oceanography, 49(6): 2095-2104. Budikova D. 2009. Role of arctic sea ice in global atmospheric circulation: a review. Global and Planetary Change, 68: 149-163. Chalecki B. 2007. Climate Change in the Arctic and its Implications for U.S. National Security, IDEAS, April 2007. http://ui04e.moit.tufts.edu/ierp/ideas/pdfs/issue2/Chalecki_Arctic.pdf. Charlson R J, Lovelock J E, Andreae M O, Warren S G. 1987. Oceanic phytoplankton, atmospheric sulfur, cloud albedo and climate: a geophysiological feedback. Nature, 326: 655-661.
14
ACCEPTED MANUSCRIPT Chaves J E, Werdell P J, Proctor C W, Neeley A R, Freeman S A, Thomas C S, Hooker S B. 2015. Assessment of ocean color data records from MODIS-Aqua in the western Arctic Ocean. Deep Sea Research, Part II, Topical Studies in Oceanography. doi: 10.1016/j.dsr2.2015.02.011. Cherkasheva A, Nöthig EM, Bauerfeind E, Melsheimer C, Bracher A. 2014. From the chlorophyll-a in the surface layer to its vertical profile: a Greenland Sea relationship for satellite applications, Ocean Science, 9: 431-
T
445.
RI P
Coupel P, Jin H Y, Joo M, Horner R, Bouvet H A, Sicre M-A, Gascard J-C, Chen J F, Garcon V, Ruiz-Pino D. 2012. Phytoplankton distribution in unusually low sea ice concentration over the Pacific Arctic. Biogeosciences Discuss., 9: 2055-2093.
Cui S, He J, He P, Zhang F, Lin L, Ma Y. 2012. The adaptation of Arctic phytoplankton to low light and salinity in
SC
Kongsfjorden (Spitsbergen). Advances in Polar Science, 23: 19-24.
Diersson H M, Smith R C, Vernet M. 2002. Glacial meltwater dynamics in coastal waters west of the Antarctic Peninsula. Proceedings of the Ntional Academy of Sciences of the United States of America, 99: 1709-1795.
MA NU
Glasser N F, Harrison S, Jansson K N, Anderson K. 2011. Global sea-level contribution from the Patagonian Ice fields since the Little Ice Age maximum. Nature Geoscience, 4: 303-307. Ericken H, Ackley S F, Richter-Menge J A, Lange M A. 1991. Is the strength of sea ice related to its chlorophyll content? Polar Biology, 11: 347-350.
Francis J A, Chan W, Leathers D J, Miller J R, Dana E. 2009. Winter Northern Hemisphere weather patterns remember summer Arctic sea-ice extent. Geophysical Research Letters, 36, L07503. Gabric A J, Qu B, Rotstayn L, Shephard J M. 2013. Global simulations of the impact on contemporary climate of a Journal, 63: 365 – 376.
ED
perturbation to the sea-to-air flux of dimethylsulphide. Australian Meteorology and Oceanographic
Gradinger R R, Baumann M E M. 1991. Distribution of phytoplankton communities in relationship to the large
PT
scale hydrographical regime in Fram Strait 1984. Mar. Biol., 111: 311-321. Henson, S., R. Sanders, J. T. Allen, I. S. Robinson and L. Brown. 2003.. Seasonal constraints on the estimation of
CE
new production from space using temperature-nitrate relationships. Geophysical Research Letters, 30. Henson, S.A., Sanders, R., Holeton, C., & Allen, J.T. 2006. Timing of nutrient depletion, diatom dominance and a lower-boundary estimate of export production for Irminger Basin, North Atlantic. Marine Ecology Progress
AC
Series, 313:73-84.
Huang H, Thomas E G, Grainger R G. 2010. Relationship between wind speed and aerosol optical depth over remote ocean. Atmos. Chem. Phys., 10: 5943-5950. Jaiser R, Dethloff K, Handorf D, Rinke A, Cohen J. 2012. Impact of sea ice concentration changes on the Northern Hemisphere atmospheric winter circulation, Tellus, A 64: 1-11. Jassby A D, Platt M E. 1976. Mathematical formulation of the relationship between photosynthesis and light for phytoplankton. Limnol. Oceanogr., 21: 540-547. Jones P D, Jonsson T Wheeler D. 1997. Extension to the North Atlantic Oscillation using early instrumental pressure observations from Gibraltar South-West Iceland. Int. J. Climatol., 17: 1433-1450. Lara R J, KATTNER G, Tillmann U, Hirche H J. 1994. The North East Water polynya (Greenland Sea) II. Mechanisms of nutrient supply and influence on phytoplankton distribution. Polar Biol., 14: 483-490. Leu E, Soreide J E, Hessen D O, Falk-Petersen S, Berge J. 2011. Consequences of changing sea-ice concentration for primary and secondary producers in the European Arctic shelf seas: timing, quantity, and quality. Progress in Oceanography, 90: 18-32. Lu H L, Pan J, Zhu J N, Qu B, Lin D R, Zhao W H, Qian F. 2010. Research on the Relationship between Photosynthetically Active Rediation and Ice concentration in Arctic Greenland Sea. Journal of Nantong University (Natural Science Edition), 9(4): 89-94.
15
ACCEPTED MANUSCRIPT Malin G. 1997. Sulphur, climate and the microbial maze. Nature, 387: 857-859. Martin J H & Fitzwater S E. 1988. Iron deficiency limits phytoplankton growth in the northeast Pacific subarctic. Nature, 331: 341-343. Matrai P A, Vernet M. 1997. Dynamics of the vernal bloom in the marginal ice zone of the Barents Sea: Dimethylsulfide and dimethysulfoniopropionate budgets. Journal of Geophysical Research – Ocean, 102:
T
22965-22979.
RI P
Moline M A, Karnovsky N J, Brown Z, Divoky G J, Frazer T K, Jacoby C A, Torres J J, Fraser W R. 2008. High latitude changes in ice dynamics and their impact on polar marine ecosystems. Annals of the New York Academy of Sciences, 1134: 267-319.
Nielsen T G, Hansen BW. 1999: Plankton community structure and carbon cycling on the western coast of
SC
Greenland during the stratified summer situation. Part I. Hydrography, phytoplankton and bacteria. Aquat. Mar. Ecol. 16: 205-216.
Olli K, Riser C W, Wassmann P, Ratkova T, Arashkevich E, Pasternak A. 2002. Seasonal variation in vertical flux
MA NU
of biogenic matter in the marginal ice zone and the central Barents Sea. Journal of Marine Systems, 38: 189-204.
Stroeve J C, Maslanik J, Serreze M C, Rigor I, Meier W, Fowler C. 2011. Sea ice response to an extreme negative phase of the Arctic Oscillation during winter 2009/2010, Geophysical Research Letters, 38(2), doi: 10.1029/2010GL045662.
Pabi S, van Dijken G L, Arrigo K R. 2008. Primary production in the Arctic Ocean, 1998-2006. Journal of Geophysical Research-Oceans, 113(C8).
ED
Pang H. 2007. Econometrics, Beijing: Science Press, 265-284. Rysgaard S, Nielsen T G. 2006. Carbon cycling in a high-arctic marine ecosystem – Young Sound, NE Greenland. Prog. Oceanogr., 71: 426-445.
PT
Qu B, Gabric A J, Matrai P A. 2006. the satellite-derived distribution of chlorophyll-a and its relation to ice concentration, radiation and sea surface temperature in the Barents Sea, Polar Biology, 29: 196-210.
CE
Qu, B, Gabric A J. 2010. Using genetic algorithms to calibrate a dimethylsulfide production model in the Arctic Ocean. Chinese Journal of Oceanology and Limnology, 28(1): 1-10. doi: 10.1007/s00343-010-9062-x. Qu B, Lu H L, Gabric A J, Lin D R, Qian F, Zhao W H. 2011. Using MODIS satellite data to analyze the
AC
relationship between chlorophyll a and aerosol optical depth in the Greenland Sea. Chinese Journal of Polar Research, 21(1): 68-76. Qu, B, Gabric A J, Zhu J, Lin D, Qian F, Zhao M. 2012. Correlation between sea surface temperature and wind speed in Greenland Sea and their relationships with NAO variability, Water Science and Engineering, 5(3): 304-315 Qu B, Gabric, A J, Lu H, Lin D. 2014. Spike in phytoplankton biomass in Greenland Sea during 2009 and the correlations among chlorophyll-a, aerosol optical depth and ice cover. Chinese Journal of Oceanology and Limnology, 32 (2): 241 – 254. Sanjay K J, Goswam A, Saraf A K. 2008. Accuracy assessment of MODIS, NOAA and IRS data in snow cover mapping under Himalayan conditions. International Journal of Remote Sensing, 29(20): 5863-5878, doi: 10.1080/01431160801908129. Schneider W, Budéus G. 1994. The North East Water Polynya (Greenland Sea). I. A physical concept of its generation. Polar Biology, 14: 1-9. Schuenemann K C, Cassano J J. 2010. Changes in synoptic weather patterns and Greenland precipitation in the 20th and 21st centuries: 2. Analysis of 21st century atmospheric changes using self-organizing maps. Journal of Geophysical Research: Atmospheres, 155(D5). doi: 10.1029/2009JD011706.
16
ACCEPTED MANUSCRIPT Schweiger A J, Ron W L, Vavrus S, Francis J F. 2008. Relationships between Arctic sea ice and clouds during autumn. Journal of Climate, 21: 4799-4810. Screen J A, Simmonds I. 2010. Increasing fall-winter energy loss from the Arctic Ocean and its role in Arctic temperature amplification. Geophysical Research Letters, 37(16): L16707. Serreze M, Barry R. 2011. Processes and impacts of Arctic amplification global and planetary change: a research
T
synthesis. Global and Planetary Change, 77: 85-96.
RI P
Smetacek V, Nicol S. 2005. Review polar ocean ecosystems in a changing world. Nature, 437: 362-368 Smith K L, Robison BH, Helly J J, Kaufmann R S, Ruhl HA, Shaw T J, Twining B S, Vernet M. 2007. Freedrifting icebergs: hot spots of chemical and biological enrichment in the Weddell Sea. Science, 317(5837): 478-482.
SC
Statham, P J, Skidmore M, Tranter M. 2008. Inputs of glacially derived dissolved and colloidal iron to the coastal ocean and implications for primary productivity. Global Biogeochemical Cycles, 22: GB3013. doi: 10.1029/2007GB003106.
MA NU
Straneo F, Curry R G, Sutherland D A, Hamilton G S, Cenedese C, Våge K, Stearns L. 2011. Impact of fjord dynamics and glacial runoff on the circulation near Helheim Glacier. Nature Geoscience, 4: 322-327. doi: 10.1038/ngeo1109.
Su W, Charlock, T P, Rose F G. Rutan D. 2007. Photosynthetically active radiation from Clouds and the Earth's Radiant Energy System (CERES) products. Journal of Geophysical Research, 112: G02022. 1-11. doi: 10.1029/2006JG000290.
Tovar-Sánchez, A, Duarte C M, Alonso J C, Lacorte S, Tauler R, Galbán-Malagón C. 2010. Impacts of metals and
ED
nutrients released from melting multiyear Arctic sea ice. Journal of Geophysical Research: Oceans, 115(C7): C07003.
Wassmann P, Ratkova T N, Andreassen I, Vernet M, Pedersen G, Rey F. 1999. Spring bloom development in the
AC
CE
PT
marginal ice zone and the central Barents Sea. Marine Ecology, 20: 321-346.
17
ACCEPTED MANUSCRIPT Figure Captions:
Figure 1: Map of Study region in Greenland Sea (highlighted box is for 20°W–10°E, 75°N–80°N). White arrows indicate the surface flow directions.
RI P
T
Figure 2: Satellite image of CHL (a) on day 180 (2010) and (b) day 160 (2011) around the study region.
SC
Figure 3: Mean CHL climatology (with standard deviation) compared with CHL in year 2010 in the study region (75°N-80°N, 20°W-10°E). Figure 4: Monthly mean CHL and sea ice concentration (ICE) in the study region with trend line (dashline) for CHL.
MA NU
Figure 5: Mean CHL and Ice Melt time series in 2010 in the study region (75°N-80°N, 20°W-10°E) Figure 6: Mean wind speed (bars) and direction (lines) for year 2003-2012 compared with those in year 2010 in the study region.
ED
Figure 7: NAO (solid dot) and AO (hollow dot) monthly mean for year 2003-2012. Red part indicates NAO values in year 2010.
AC
CE
PT
Figure 8: Weekly time series of Ice Melt and NAO in year 2010 in the study region ( 75°N -80°N, 20°W-10°E).
18
ACCEPTED MANUSCRIPT Highlights ● The causes of an unusually early and intense phytoplankton bloom in the
AC
CE
PT
ED
MA NU
SC
RI P
T
northern Greenland Sea during 2010 are investigated. ● Earlier and more extensive sea ice melt, persistent negative NAO, and changing wind directions were the main drivers of the bloom. ● Multivariate lagged regression analysis shows the bloom was correlated with the timing of sea ice melt,PAR and SST.
19
S0924-7963(16)30217-2 doi: 10.1016/j.jmarsys.2016.07.011 MARSYS 2856
To appear in:
Journal of Marine Systems
Received date: Revised date: Accepted date:
22 January 2016 31 May 2016 22 July 2016
Please cite this article as: Qu, Bo, Gabric, Albert J., Lu, Zhifeng, Hehe, Li, Li, Zhao, Unusual phytoplankton bloom phenology in the northern Greenland Sea during 2010, Journal of Marine Systems (2016), doi: 10.1016/j.jmarsys.2016.07.011
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Unusual phytoplankton bloom phenology in the northern
RI P
T
Greenland Sea during 2010 QU Bo (瞿波) 1, Albert J. GABRIC2, LU Zhifeng3, Li Hehe4 ，Zhao Li5 13,4,5
.School of Science, Nantong University, Nantong, 226007, China
2
SC
. School of Environment, Griffith University, Nathan, QLD, 4111, Australia
Abstract: Arctic marine ecosystems are disproportionately impacted by global warming. Sea ice plays an
MA NU
important role in the regional climate system and the loss of perennial sea ice has diverse ecological implications. Here we investigate the causes of an unusually early and strong phytoplankton bloom in the northern Greenland Sea (20°W-10°E, 75°N -80°N) during the 2010 season. In order to better understand the anomalous bloom in 2010 we examine the correlation between satellite-derived biomass and several possible environmental factors for the period 2003-2012. Results show that the timing of sea ice melt played an important role in promoting the growth of phytoplankton. Multivariate lagged regression analysis shows that phytoplankton biomass (CHL) is correlated with ice concentration(ICE) and ice melting, as well as sea surface temperature (SST) and photosynthetically
ED
active radiation (PAR). During 2010, the spring peak in biomass came much earlier and achieved a higher value than most other years in the satellite archive record, which was due to earlier and more extensive sea ice melt in that year. Relative lower SST and PAR in spring and early summer in year 2010 associated with a persistent
PT
negative North Atlantic Oscillation (NAO) index were possible drivers of the bloom. Wind direction changed from the southeast to southwest direction in spring, possibly transporting nutrient enriched melt runoff from glaciers on Greenland and other sources from the south to northern coastal regions.
CE
Keywords: chlorophyll-a; sea ice concentration; North Atlantic Oscillation (NAO); Arctic Oscillation (AO);
1.
AC
Greenland Sea (GS)
INTRODUCTION The Greenland Sea (GS) is one of the most productive regions in the Arctic (Arrigo and
Van Dyiken, 2011) and adjacent to the world’s second largest glacier in Greenland. Glaciers in the northeast of Greenland are melting faster than expected (Glasser et al., 2011). The GS is an important area for water mass exchange between the North Atlantic Ocean and the Arctic Ocean. It is also the area to where most Arctic drifting ice is advected (Cherkasheva et al., 2014). Hence, the GS is an appropriate region for studying the relationship between sea ice and phytoplankton dynamics and where an extensive archive of in situ and satellite-derived chlorophyll data is available (Arrigo et al., 2011).
Surface Currents
E-mail: [email protected], This work is funded by National Nature Science Funding No. 41276097.
1
ACCEPTED MANUSCRIPT The surface currents in the GS are shown in Figure 1, within our study region is highlighted by the red box. The East Greenland Current (EGC) moves from north to south along Greenland’s eastern coastline bringing colder less saline Arctic water to the south. From south-east of Iceland, warmer more saline Atlantic water flows to north merging with the Norwegian current and
T
flowing into Arctic ocean. In this area, the vertical stability of the water column increases to the
RI P
north due to the input of melt water and solar heating, causing phytoplankton biomass to increase and nutrient concentration to decrease (Lara et al., 1994). Between 70°N -80°N, there is an anticlockwise gyre, affecting our study region, and at around 70°N, the EGC branches into two parts,
SC
one flowing along the west coast of Norway and east into the Barents Sea, and the other northwards to the Spitsbergen region. The Polar front is located to the east of EGC and the Arctic
MA NU
front is located west of the Norwegian Current.
Sea ice melt, runoff, iron and phytoplankton dynamics
The decline of Arctic sea ice (and concomitant decrease in surface albedo) in recent decades has resulted in a regional temperature increase in the Arctic, where sea surface temperatures have increased at twice the global average rate and could continue to increase throughout this century (Chalecki, 2007). Melting of Greenland’s ice sheet has increased six-fold
ED
over the last decade ago, according to a draft of the UN’s most comprehensive study on climate change (Stroeve et al., 2011). Greenland may add a total of 4 - 21 centimeters to global sea levels
PT
by the end of the 21stC (Schuenemann and Cassano, 2010). The indirect effect of melting sea ice is an increase in regional average temperatures, which may accelerate the melting of the Greenland ice sheet and lead to global sea level rise. Arctic sea ice concentration has retreated extensively
CE
and first-year ice is thinning, making it more vulnerable to summer melting and sea level rising . Summer sea ice could be totally gone by 2030 (Stroeve et al., 2011).
AC
Several studies have examined the impact of sea ice melt on regional phytoplankton biomass (Matrai et al., 1997; Wassmann et al., 1999; Olli et al., 2002; Qu et al., 2006; Pabi et al., 2008, Leu et al., 2011). It is suggested that decreasing sea ice extent and thickness and a concomitant increase in water column illumination, leads to an increase in phytoplankton biomass. Moline et al. (2008) point out that the less saline surface water during ice melt can stimulate primary production. The melting ice increases the area of open water, increasing absorption of solar radiation and could enhance the melt process. In July 2012, the NASA ICESCAPE project discovered large under ice blooms appeared in Arctic water due to thinning ice and proliferation of melt ponds (Arrigo et al., 2012). Phytoplankton was extremely active and growth rates were the highest ever measured in polar waters. Arrigo et al. (2012) suggest that satellite-based estimates of annual primary production in Arctic waters may be underestimated up to 10-fold due to under ice blooms. In contrast to the Southern Ocean where primary production is iron limited, the Arctic Ocean is generally land-locked, with higher levels of atmospheric deposition of micro-nutrients such as iron, which can stimulate primary production during and after ice melting (Moline et al., 2008). 2
ACCEPTED MANUSCRIPT It is suggested that glacial runoff serves as a significant source of bioavailable iron to the surrounding oceans (Bhatia et al. 2013). The Greenland ice sheets are also likely a significant source of iron to the adjacent ocean (Hawkings et al., 2014). The availability of Iron (Fe) is said to be the main factor controlling primary production in high nutrient (N, P, Si) and low CHL (HNLC)
T
systems (Martin and Fitzwater, 1988). An alternative source of Fe to the ocean from glacial runoff
RI P
is associated with glacial particles (Smith et al., 2007). These glacial melt waters provide an important supply of Fe and other micro nutrients to surface Arctic waters (Dierssen et al., 2002; Statham et al. , 2008)
SC
Primary production is high in the coastal zone of Greenland, due to the impact of melting sea ice (Rysgaard and Nielsen, 2006). Macronutrients (N, Si, and P) appear to control primary production (Nielsen and Hansen, 1999), although Fe could be a limiting nutrient (Blain et al.,
MA NU
2004). During winter storms, vertical mixing of high Fe content deep water, together with the lateral entrainment of high Fe surface water from Greenland coast could contribute to relatively high Fe concentrations at the beginning of the phytoplankton bloom period. The melt water input from Greenland glacier inputs is around 10% of sea ice melt water, although sea ice melt would occur earlier in the season (Statham et al., 2008).
Following on a previous analysis of the spike of phytoplankton biomass in the GS during
ED
2009 (Qu et al., 2014), more recent satellite data indicated elevated phytoplankton biomass in the northern Greenland Sea during 2010 when compared to data for the ten year period 2003-2012.
PT
Here we examine the factors that may be responsible for higher phytoplankton biomass in 2010. Apart from the effect of melting ice on phytoplankton biomass, other factors such as sea surface temperature (SST), wind speed (WIND), photosynthetically active radiation (PAR), and climate
CE
variability as indicated by the North Atlantic Oscillation (NAO) are also considered.
AC
2. MATERIAL AND METHODS 2.1 Data Sources
Our study region is the northern Greenland Sea (20°W-10°E, 75°N -80°N) (Figure 1), for the period 2003-2012. Due to the Arctic sunset, satellite data are only available between March and September. MODIS (Aqua) satellite, 8-day, 4-km, level 3, mapped data Aerosol Optical Depth (AOD), Chlorophyll-a (CHL) and Photosynthetically Active Radiation (PAR) global data was archived (modis.gsfc.nasa.gov/). Sea ice concentration (ICE) is from the following archive iridl.ldeo.columbia.edu/SOURCES/.NOAA/.NCEP/.EMC/.CMB/.GLOBAL/.Reyn_SmithOIv2/. Wind speed and direction, and sea surface temperature (SST) were obtained from www.remss.com/windsat.
PAR
is
derived
from
SeaWiFS,
8-day
mapped
data
(oceandata.sci.gsfc.nasa.gov/seawifs). The image analysis package data analysis system SeaDAS 6.4 (seadas.gsfc.nasa.gov/) is used to subset data for our study region. Mean values for each (1°x1°) grid cell are calculated first,
3
ACCEPTED MANUSCRIPT with missing values excluded. Regional mean values of the CHL，AOD and PAR are calculated by averaging the 150 (1°x1°) grid cells. Mean values of the wind and direction of each (1°x1°) grid are calculated from the 0.25°x0.25° grid. The weekly data windAW is chosen for wind speed data. windAW is 10 meter
T
surface wind for all weather conditions made using 3 algorithms (Meissner et al. 2009) with rainy
RI P
condition included. The missing values are excluded again in the mean value calculations. The CHL, AOD and PAR could be slightly over predicted due to the exclusion of missing values . However, the higher CHL spots obviously occurred in early summer in the study region as （day 184） 2010 and May (day 160) 2011.
SC
indicated in the satellite images (Figure 2). Figure 2 shows the CHL satellite images in early June
Eviews statistical software is used for correlation and lagged regression analysis. R
MA NU
software is used to do the partial correlation analysis and multivariate regression analysis among
AC
CE
PT
ED
mean time series of CHL, SST, WIND, ICE and PAR in the study region.
Fig. 1 Map of Study region in Greenland Sea (highlighted box is for 20°W–10°E, 75°N–80°N). White arrows indicate the surface flow directions.
(a)
(b)
Fig. 2 Satellite image of CHL (a) on day 180 (2010) and (b) day 160 (2011) around the study region.
4
ACCEPTED MANUSCRIPT 2.2 Accuracy of the Satellite data Due to the remoteness of the Arctic Ocean, the satellite data is the only means of obtaining synoptic coverage. However, the accuracy of these data directly relate to the reliability of this
T
study. Surface chlorophyll concentration (CHL) is from the MODIS AQUA sensor with data
RI P
available from 2002. NASA carried a polarization correction for MODIS productions and improved the accuracy, especially for the Arctic region. Calibrations were done for regional differences between MODIS (TERRA & AQUA) with SeaWiFS (Bailey and Werdell, 2006). At
SC
high latitudes, multiple orbits are considered for a given in situ record.
However, there is still an unavoidable error that can occur for solar zenith angles more than 70° and view angles more than 45° (Bailey and Werdell, 2006). Due to cloud and ice cover, the
MA NU
data is more accurate during summer time compared to early spring and winter. It is reported that the satellite-derived CHL underestimates the concentration in the field by a factor of 1.4 when compared with in-situ CHL data averaged over the optical penetration depth (Bailey and Werdell, 2006). The relative biases for PAR retrieval are 4.6% for all sky and 2.9% for clear sky (Su et al., 2007). Notice that MODIS PAR is cloud corrected. Satellite-to-in situ match-ups for MODISA (MODIS instrument on board Aqua) were evaluated for CHL in the western Arctic Ocean, and the
ED
errors are well within the range of errors for global data (Chaves et al., 2015). Another possible source of error is the computation of regional means due to missing data.
PT
Bearing in mind that computed study region mean values could be higher than reality. Although the errors of the satellite data in Arctic Ocean are unavoidable, more accurate satellite data in the Arctic Ocean is expected in the future. However, our results indicate that the unusual nature of the
AC
3. RESULT
CE
phytoplankton bloom happened in year 2010.
3.1 CHL and Ice Melt distributions A ten-year climatology (2003-2012) derived from 8-day data (Figure 3) shows a gradual increase in CHL from March, reaching a late spring peak on day 168 (mid-June), followed by a small dip on day 184 (early July), and a second peak on day 192 (mid-July), steadily decreasing through to the end of September. Figure 3 also shows the CHL time series in year 2010 when the spring and summer peaks of CHL were more than double the long-term average. There were unusually high peaks on day 128, and 184. It is interesting to note the high interannual variability in CHL during early spring, possibly related to interannual variability in sea ice concentration (Qu et al., 2014).
5
Mean CHL climatology (with standard deviation) compared with CHL in year 2010 in the study region
MA NU
Fig. 3
SC
RI P
T
ACCEPTED MANUSCRIPT
(75°N-80°N, 20°W-10°E).
Comparing to average zonal mean, year 2010 exhibited a strong meridional gradient (not shown in figures)..
The time series of monthly CHL and sea ice concentration (ICE) in the study region is
ED
shown in Figure 4. There are apparently strong negative correlations between CHL and ICE from April to June. CHL increased in spring with the decreased ICE. There is a clear positive trend in the amplitude of the CHL spring bloom during 2003-2012. CHL increases as sea ice concentration
PT
decreases during spring and early summer, however the timing of the CHL bloom displays considerable interannual variability, occurring in either May (2003, 2010), June (2004, 2006, 2007
CE
and 2011) or July (2005, 2008, 2009, 2012). A doublet peak of CHL in year 2010 in both spring and early summer was distinguished from other years. The general increased trend is shown in the figure with a positive slope. For spring and early summer period, the increasing rate was much
AC
more significant. Mean CHL increased 10.8% during spring (before day 168) within year 20032012 in the study region (not show in figure).
Fig. 4 Monthly mean CHL and sea ice concentration (ICE) in the study region with trend line (dashline) for CHL.
6
ACCEPTED MANUSCRIPT Ice melt is computed by the percentage change in sea ice concentration between octads. The 10-year (2003-2012) climatological mean sea ice melt is compared with that in year 2010. Ice melt in 2010 was higher than the climatological average during most of spring and from July to midAugust . In other years, ice melt did not commence until May (e.g 2007, 2008, 2009 and 2012).
T
However, 2010 saw consistent ice melt from April to August. The time series for CHL and Ice
RI P
melt in 2010 is shown in Figure 5. The first CHL peak of 2010 occurred in early May. From late
MA NU
SC
April (day 112) until mid-June (day 160), CHL tracks the change in Ice melt.
ED
Fig. 5 Mean CHL and Ice Melt time series in 2010 in the study region (75°N-80°N, 20°W-10°E).
3.2 Impacts of SST, PAR and Wind
SST gradually increases from a minimum in March to a maximum in July in 2010. SST in
PT
2010 was 8.45% lower than the climatological mean during spring and early summer. The PAR profiles had normal distributions with the similar pattern of higher in summer and lower in winter.
CE
Interestingly, similar to SST, PAR was 13.6% lower than the climatological average during spring
AC
and early summer in 2010.
Average wind speed (bars) and wind direction (curves) for year 2003-2012 and year 2010 are compared in Figure 6. Although there was a much higher wind speed occurred in early spring in year 2010, year 2010 had lower wind speed in general. Lower wind speed favors phytoplankton growth. There is an inverse relationship between wind speed and CHL (Fitch and More 2007). Wind direction (the gcurve lines in the figure) generally was from the southeast direction changing to a southwest direction in March and early April (day 71-day 99). This possibly brought melt water from south to north and together with glacial runoff water from the east coast of Greenland.
7
SC
RI P
T
ACCEPTED MANUSCRIPT
b
÷
MA NU
Fig. 6 Mean wind speed (bars) and direction (lines) for year 2003-2012 compared with those in year 2010 in the study region.
Positive (negative) NAO is usually described as a stronger (weaker) north-south pressure gradient between the Azores high and the Icelandic low (Hurrell, 1995, Qu et al., 2012). Positive NAO is associated with cool summers and mild and wet winters in Central Europe. In contrast, a
ED
negative NAO indicates the cold air comes from Europe down south and milder temperatures in Greenland (Hurrell, 1995). AO (Arctic Oscillation) is a large scale mode of climate variability, also
referred
to
as
the
Northern
Hemisphere
annular
mode
(https:
PT
//www.ncdc.noaa.gov/teleconnections/ao/). The positive AO indicates a ring of strong winds circulating around the North Pole that confining colder air across polar regions. NAO and AO
CE
indices for the period 2003 -2012 are shown in Figure 7. NAO and AO had inter-annual variations and showed similar patterns. In contrast to most other years, the NAO during 2010 was negative
AC
throughout the year (shown in red colour in the figure). The AO was mostly negative during 2010 especially in the winter. Milder conditions in Greenland would lead to more glacial ice melt from the east coast of Greenland to the GS. Stroeve et al. (2011) analyzed the winter of 2009-2010 and found sea ice by September 2010 was the third lowest in the satellite record. They also pointed out that negative AO indicated sea ice motion tends to have clockwise anomaly, with enhanced ice transport from the western to the eastern Arctic. The warm southerly winds favors ice melt and transport ice poleward, with enhanced ice export through Fram Strait in a northerly direction. That may explains the higher ice melt during 2010 in the northern Greenland Sea, and hence higher and earlier CHL blooms occurred in our study region.
8
ACCEPTED MANUSCRIPT 3 2
T
-1 -2 -3
NAO
-4 -5
AO
Ja n0 Ju 3 l-0 Ja 3 n0 Ju 4 l-0 Ja 4 n0 Ju 5 l-0 Ja 5 n0 Ju 6 l-0 Ja 6 n0 Ju 7 l-0 Ja 7 n0 Ju 8 l-0 Ja 8 n0 Ju 9 l-0 Ja 9 n1 Ju 0 l-1 Ja 0 n1 Ju 1 l-1 Ja 1 n1 Ju 2 l-1 2
RI P
NAO, AO
1 0
SC
Month
Fig. 7 NAO (solid dot) and AO (hollow dot) monthly mean for year 2003-2012. Red part indicates NAO values in
MA NU
year 2010.
3.5 Correlation between CHL and sea ice concentration (ICE) The correlation coefficients for CHL and ICE for each year are shown in Table 1. Generally, they had negative correlations with year 2009 much more significant than other years. Table 1 Cross-correlation coefficient for CHL and ICE 2005
0.03
-0.21
-0.20
2006
2007
2008
2009
2010
2011
2012
-0.33
0.11
0.07
-0.53
-0.10
0.05
-0.43
ED
2004
PT
2003
The peak of CHL is about 3 months behind of ICE in the study region. The correlation coefficients of CHL and ICE before and after shifting in year 2010 in the study region range from
CE
0.13 to 0.6. The statistical software package EViews (Pang, 2007) is used to do more accurate correlation and regression analysis between CHL and ICE.
AC
The regression equation for CHL and ICE is as follows， and it is significant: CHL = -0.04+0.01ICE
(75°N–80°N, 20°W–10°E))
(2)
Table 2 shows the ICE(-3) (ICE shifted 3 months behind) had lowest P-value (Prob.=0.0003) for CHL regression coefficient test. Hence, ICE(-3) had the most significant influence on CHL.
9
ACCEPTED MANUSCRIPT Table 2 Lagged regression analysis result for CHL and ICE (Dependent variable is CHL) Coefficient
Std. error
t-Statistic
Prob.
ICE(-1)
0.0103
0.0035
2.9376
0.0047
ICE(-2)
0.0106
0.0041
2.5925
0.0119
ICE(-3)
0.0159
0.0041
3.8404
0.0003
ICE(-4)
-0.0062
0.0035
-1.7706
The goodness of fit: R =0.59
F-statistic: 21.9454
0.0816
RI P
2
T
Variable
SC
Unit root test of CHL was performed and results show that CHL has unit root and is nonstationary sequence (Qu et al., 2014). The same method is used to test the first-order differential sequence of CHL, which is also a stationary sequence. The regression residuals sequence test
MA NU
result shows that t-test statistical value is also less than 3 corresponding critical values. Again, it shows the residuals sequence did not have unit root, it was stationary sequence. Hence, CHL and ICE were co-integrated or had a long-term equilibrium relationship. R software is used to do the partial correlation analysis among mean time series of CHL, SST, WIND, ICE and PAR in the study region. Pearson's product-moment correlation method is
ED
used. The results are listed in Table 3 in the study region.
Table 3 Partial correlation coefficient among CHL, SST, WIND, ICE and PAR WIND
ICE
PAR
0.15
-0.11
0.15
0.32
1
-0.12
-0.67
0.24
-0.12
1
0.27
-0.47
0.15
-0.67
0.27
1
0.44
0.32
0.24
-047
0.44
1
1
SST
0.15
WIND
-0.12
ICE
AC
PAR
CE
CHL
SST
PT
CHL
Table 3 indicates that CHL had more relationship with PAR (coefficient is 0.32). Other relationships were generally weak. Granger causality analysis shows that PAR more likely influenced CHL. ICE and CHL had positive relationship and they influence each other. SST had weak positive relationship with CHL. However, SST has less influence on CHL. WIND had weak negative relationship with CHL.
A multivariate regression linear model is applied using R software (R3.2.3) for years 20032012. There are different regression equations with different time lags for each year. Table 4 listed the more significant regression equations within certain variables for year 2003, 2004, 2005,, 2007, 2009 and 2011. In other years the correlations are not significant. In general, CHL had higher correlation with ICE. In some years, CHL was correlated with Ice Melt as well as PAR or SST. The standard deviations and P-values of according variables in the equations are listed. In general, most of the standard deviations are low and P-values are less than 0.05. The regressions are mostly
10
ACCEPTED MANUSCRIPT significant. The goodness of fitting (R-squared) are within 0.72-0.92. That means that the listed lagged regression equations are quite correlated. The lagged regression equation for the mean of years 2003-2012 is as follow: CHLt 9 1.4221 - 0.0265PARt
T
(4)
The residual standard error is 0.106 and R-squared value is 0.8629, P-value is 0.000127.
RI P
The CHL had more correlation with PAR for the 10 years. For year 2010, the CHL was more dependent on ICE, Ice Melt and PAR. The regression equation for year 2010 is as follow: (5)
SC
CHLt 15 -4。 2715 0.7048ICEt - 6.061IceMelt t 0.0262PARt
The CHL was 15 octads （120 days） behind ICE, Ice Melt and PAR. The goodness of fit (R-squared) is 0.9843, The P-values are less than 0.05, which means the equation (5) is
MA NU
significant. The multivariate lag regression results are listed in Table 4 for those more significant years. The standard deviation (std.) and P-value of each variable are listed underneath correspondent variables. The goodness of fit (R-squared) are within the range of 0.72-0,92. Table 4
Multivariate lag regression analysis for some significant years Regression equation and correspondent std. and P-Value
2003
CHLt 5 0.4572 0.0073ICEt - 0.0111PARt
ED
Year
Std
0.0022
0.00006
0.00027
0.00019
CHLt 7 2.8582 - 0.0319ICEt - 0.299SSTt Std
0.8823
0.0156
0.0727
P-Value
0.0089
0.0677
0.0021
CE
2004
CHLt 11 1.0505 - 0.0177 PARt - 0.0396IceMelt t
AC
2005
2007
0.0015
PT
P-Value
0.0811
Std
0.0943
0.0037
0.0210
P-Value
0.00001
0.0205
0.1015
CHLt 9 1.9804 - 0.0493PARt - 0.108IceMelt t Std
P-Value
0.1277 0.00000008
0.0343
0.0000093
0.0119
CHLt 7 2.8386 - 0.0242ICEt - 0.108IceMelt t 0.1846SSTt
2009
Std
0.6839
P-Value 0.00198 2011
0.0055
0.0114
0.0078
0.0598
0.0048
0.0366 0.0005
CHLt 7 -15.2689 - 0.2834ICEt 0.3276IceMelt t 0.2713SSTt - 0.05PARt Std
6.1244
0.1001
0.0972
0.1598
0.0184
P-Value
0.0343
0.0197
0.0083
0.1239
0.0237
3.6 Correlation between Ice melt, NAO and CHL Ice Melt had a better correlation with NAO compared with the relationship between sea ice concentration (ICE) and NAO. There was a positive correlation relationship during spring (around March to May) between Ice Melt and NAO time series, although it is not consistent positive 11
ACCEPTED MANUSCRIPT related. This result is confirmed by statistical software EViews. Figure 8 shows the weekly time series for Ice Melt and NAO in year 2010. However, from day 112-208 (March-June), they were mostly positive correlated. The correlation coefficient between Ice Melt and NAO during this period is 0.56 and it is significant. The positive correlation indicates that, with the increasing of
MA NU
SC
RI P
T
NAO, Ice Melt would increase in spring 2010.
Fig. 8 Weekly time series of Ice Melt and NAO in year 2010 in the study region ( 75°N -80°N, 20°W-10°E).
In general, Ice Melt was negatively correlated with NAO. Eviews regression analysis gives
ED
result of the regression equation for CHL, ICE-Melt and NAO in the 10 years: CHL=0.054–0.111NAO–0.012IceMelt(-3)
(6)
The regression equation is significant. Therefore, Ice Melt and NAO had 3 months’ time lag
PT
with NAO 3 months ahead of Ice Melt, suggesting NAO and Ice Melt had significant influence on
CE
CHL.
4. DISCUSSION
AC
The elevated CHL in northern part of Greenland Sea is unusual and the significant rising in year 2010 related to many oceanography and biological reasons. In early years, the unusual phenology in northern GS was discovered by a few researchers. Lara et al. (1994) discussed the mechanisms of nutrient supply and factors influencing phytoplankton distribution in the northeast GS (78°N-82°N, 20°W-0), and found that vertical stability was higher at these latitudes due to the input of melt water from Greenland, with a corresponding reduction in salinity and increased heating of the water column near 80°N. Similarly, Cherkasheva et al. (2014) did a study in the Fram Strait area (76°N -84°N, 25°W-15°E) and found that late sea ice retreat leads to a late iceassociated bloom in the northern GS. In ice-free water, solar heating should be similar in the north and south of the region. Input of melt water could make the difference. They also found that the stratification of the surface water due to increased solar radiation after ice melt coincided with an increase in CHL in late April and early May. Patrick Lockerby firstly found that spring and early summer had large extent of ice breakup north
of
80°N
in
2010
in
Greenland
Sea
(http://www.science20.com/chatter_box
/arctic_ice_may_2010). Massive ice break-up extended from the west to northeast of Greenland
12
ACCEPTED MANUSCRIPT Sea (Fram Strait).due to the rapid ice melt in bottom layer. In May 2010, the ice flow north of Greenland was extensive and almost 10 times above the normal situations. In contrast to the Southern Ocean where iron limits primary production, the Arctic Ocean is supplied with trace elements from river runoff and from the atmospheric dust deposition leading to the incorporation
T
of Aeolian sediments in sea ice. The melting of ice would thus be a possible source of iron for
RI P
primary production (Smetacek and Nicol, 2005).
Stroeve et al. (2011) also found that year 2010 had higher SST in spring and summer. The ice concentration reached the third lowest in the satellite record by September. indicating that
SC
much more ice melt occurred in year 2010. Straneo et al. (2011) examined runoff from the Helheim glaciers in year 2009 and 2010 in southeast of Greenland. Their findings indicated that a large amount of melt water and runoff entered the ocean at depth, affecting water column
MA NU
stratification.
In general， CHL increased after ice melt. However, CHL would reach to a peak in sometime during early summer. The reasons of halting the CHL are complicated. Apart from Ice melting, SST、light, wind speed and direction, grazing of zooplankton, nutrients are one of important factors. The depletion of nutrients likely halts the bloom of CHL. The nutrients include nitrate, phosphate, silicate and iron. Inverse relationship between SST and nutrients (Temperature-
ED
Nitrate relationship) were found, although the relation is non-linear 6 C (Henson et al., 2003). The nitrate consumption occurs before SST reached to 6 C (Henson et al., 2003) . With the
PT
temperature rises, nutrients are consumed by phytoplankton growth (Henson et al. 2006). In Arctic and North Atlantic Oceans, diatoms are often dominate the spring blooms ((Matrai 1997),
CE
(Henson et al. 2006)). Diatoms rely on silica to form their frustules. Once silica depleted, diatom dominance ends. Nitrate and phosphate do not become depleted during spring bloom, silica is the
AC
limiting factor of the blooms. Later in grow season (autumn and winter), nutrients regenerated from organic material synthesized, the autumn blooms usually occurs (Qu and Gabric，2010)
5. CONCLUSION We investigated the phenology of the CHL bloom in year 2010 in the northern GS (70°N80°N). The unusual CHL bloom phenology in 2010 could be due to the following combination of factors: the early ice melt, and change of wind direction between southeast and southwest direction during the melt season possibly brought melt water from the south to north and runoff water (with enriched iron content) from the Greenland coast which was advected in a northeast direction oceans (near 79°N-80°N). The consistently negative AO and NAO index in year 2010, corresponding to the milder Greenland air temperatures, which likely caused more clockwise direction glacial ice melt and drift from south to north. Relative lower SST and PAR in spring and early summer also favors phytoplankton blooms. All of above factors contributed to the water column vertical stability, hence favoring phytoplankton growth.
13
ACCEPTED MANUSCRIPT More accurate satellite data is urged for further research. Improved Arctic Ocean colour satellite data, especially cloud-free data coverage, will provide more accurate analysis on interannual variations of phytoplankton distribution in this remote high latitude Arctic Ocean.
T
ACKNOWLEDGEMENT
RI P
We acknowledge NASA’s Ocean Biology Processing Group for providing MODIS aqua, Level 3 (4-km equi-rectangular projection) 8-day mapped data for aerosol optical depth (AOD) and chlorophyll-a (CHL) And Photosynthetically Active Radiation (PAR) global data. Thanks to
SC
NASA Web SeaDAS development group for providing Ocean Colour SeaDAS Software (SeaWiFS Data Analysis System) for processing regional CHL, AOD and PAR data. NOAA Reyn-SmithOIv2 provided weekly and monthly sea-ice concentration. Sea Surface Temperature
MA NU
and Wind Speed (Wind Direction) global data are produced by NASA Remote Sensing Systems and sponsored by the NASA Earth Science MEaSUREs DISCOVER Project and the NASA Earth Science Physical Oceanography Program. Thanks to NASA http: //gdata1.sci.gsfc.nasa.gov for providing cloud cover data.
We are grateful to the Chinese National Natural Science Foundation of China (Funding No.
ED
41276097) for providing funding for this project.
REFERENCE Res. 116, C09011.
PT
Arrigo K R, van Dijken G L. 2011. Secular trends in Arctic Ocean net primary production. Journal of Geophys.
CE
Arrigo K R, Matrai P A, van Dijken G L. 2011. Primary productivity in the Arctic ocean: impacts of complex optical properties and subsurface chlorophyll maxima on large-scale estimates. Journal of Geophysical Research, 116, C11022.
AC
Arrigo K R, Perovich D K, Pickart R S et al. 2012. Massive phytoplankton blooms under arctic sea ice. Science, 336 (6087): 1408. Bailey S W, Werdell P J. 2006. A multi-sensor approach for the on-orbit validation of ocean color satellite data products. Remote Sensing of Environment, 102: 12-23. Bhatia, M.P., Kujawinski, E.B., Das, S.B., Breier, C.F., Henderson, P.B., & Charette, M.A. (2013). Greenland meltwater as a significant and potentially bioavailable source of iron to the ocean. Nature Geosci, 6, 274278 Blain S, Guieu C, Claustre H, Leblanc K. 2004. Availability of iron and major nutrients for phytoplankton in the northeast Atlantic Ocean. Limnology and Oceanography, 49(6): 2095-2104. Budikova D. 2009. Role of arctic sea ice in global atmospheric circulation: a review. Global and Planetary Change, 68: 149-163. Chalecki B. 2007. Climate Change in the Arctic and its Implications for U.S. National Security, IDEAS, April 2007. http://ui04e.moit.tufts.edu/ierp/ideas/pdfs/issue2/Chalecki_Arctic.pdf. Charlson R J, Lovelock J E, Andreae M O, Warren S G. 1987. Oceanic phytoplankton, atmospheric sulfur, cloud albedo and climate: a geophysiological feedback. Nature, 326: 655-661.
14
ACCEPTED MANUSCRIPT Chaves J E, Werdell P J, Proctor C W, Neeley A R, Freeman S A, Thomas C S, Hooker S B. 2015. Assessment of ocean color data records from MODIS-Aqua in the western Arctic Ocean. Deep Sea Research, Part II, Topical Studies in Oceanography. doi: 10.1016/j.dsr2.2015.02.011. Cherkasheva A, Nöthig EM, Bauerfeind E, Melsheimer C, Bracher A. 2014. From the chlorophyll-a in the surface layer to its vertical profile: a Greenland Sea relationship for satellite applications, Ocean Science, 9: 431-
T
445.
RI P
Coupel P, Jin H Y, Joo M, Horner R, Bouvet H A, Sicre M-A, Gascard J-C, Chen J F, Garcon V, Ruiz-Pino D. 2012. Phytoplankton distribution in unusually low sea ice concentration over the Pacific Arctic. Biogeosciences Discuss., 9: 2055-2093.
Cui S, He J, He P, Zhang F, Lin L, Ma Y. 2012. The adaptation of Arctic phytoplankton to low light and salinity in
SC
Kongsfjorden (Spitsbergen). Advances in Polar Science, 23: 19-24.
Diersson H M, Smith R C, Vernet M. 2002. Glacial meltwater dynamics in coastal waters west of the Antarctic Peninsula. Proceedings of the Ntional Academy of Sciences of the United States of America, 99: 1709-1795.
MA NU
Glasser N F, Harrison S, Jansson K N, Anderson K. 2011. Global sea-level contribution from the Patagonian Ice fields since the Little Ice Age maximum. Nature Geoscience, 4: 303-307. Ericken H, Ackley S F, Richter-Menge J A, Lange M A. 1991. Is the strength of sea ice related to its chlorophyll content? Polar Biology, 11: 347-350.
Francis J A, Chan W, Leathers D J, Miller J R, Dana E. 2009. Winter Northern Hemisphere weather patterns remember summer Arctic sea-ice extent. Geophysical Research Letters, 36, L07503. Gabric A J, Qu B, Rotstayn L, Shephard J M. 2013. Global simulations of the impact on contemporary climate of a Journal, 63: 365 – 376.
ED
perturbation to the sea-to-air flux of dimethylsulphide. Australian Meteorology and Oceanographic
Gradinger R R, Baumann M E M. 1991. Distribution of phytoplankton communities in relationship to the large
PT
scale hydrographical regime in Fram Strait 1984. Mar. Biol., 111: 311-321. Henson, S., R. Sanders, J. T. Allen, I. S. Robinson and L. Brown. 2003.. Seasonal constraints on the estimation of
CE
new production from space using temperature-nitrate relationships. Geophysical Research Letters, 30. Henson, S.A., Sanders, R., Holeton, C., & Allen, J.T. 2006. Timing of nutrient depletion, diatom dominance and a lower-boundary estimate of export production for Irminger Basin, North Atlantic. Marine Ecology Progress
AC
Series, 313:73-84.
Huang H, Thomas E G, Grainger R G. 2010. Relationship between wind speed and aerosol optical depth over remote ocean. Atmos. Chem. Phys., 10: 5943-5950. Jaiser R, Dethloff K, Handorf D, Rinke A, Cohen J. 2012. Impact of sea ice concentration changes on the Northern Hemisphere atmospheric winter circulation, Tellus, A 64: 1-11. Jassby A D, Platt M E. 1976. Mathematical formulation of the relationship between photosynthesis and light for phytoplankton. Limnol. Oceanogr., 21: 540-547. Jones P D, Jonsson T Wheeler D. 1997. Extension to the North Atlantic Oscillation using early instrumental pressure observations from Gibraltar South-West Iceland. Int. J. Climatol., 17: 1433-1450. Lara R J, KATTNER G, Tillmann U, Hirche H J. 1994. The North East Water polynya (Greenland Sea) II. Mechanisms of nutrient supply and influence on phytoplankton distribution. Polar Biol., 14: 483-490. Leu E, Soreide J E, Hessen D O, Falk-Petersen S, Berge J. 2011. Consequences of changing sea-ice concentration for primary and secondary producers in the European Arctic shelf seas: timing, quantity, and quality. Progress in Oceanography, 90: 18-32. Lu H L, Pan J, Zhu J N, Qu B, Lin D R, Zhao W H, Qian F. 2010. Research on the Relationship between Photosynthetically Active Rediation and Ice concentration in Arctic Greenland Sea. Journal of Nantong University (Natural Science Edition), 9(4): 89-94.
15
ACCEPTED MANUSCRIPT Malin G. 1997. Sulphur, climate and the microbial maze. Nature, 387: 857-859. Martin J H & Fitzwater S E. 1988. Iron deficiency limits phytoplankton growth in the northeast Pacific subarctic. Nature, 331: 341-343. Matrai P A, Vernet M. 1997. Dynamics of the vernal bloom in the marginal ice zone of the Barents Sea: Dimethylsulfide and dimethysulfoniopropionate budgets. Journal of Geophysical Research – Ocean, 102:
T
22965-22979.
RI P
Moline M A, Karnovsky N J, Brown Z, Divoky G J, Frazer T K, Jacoby C A, Torres J J, Fraser W R. 2008. High latitude changes in ice dynamics and their impact on polar marine ecosystems. Annals of the New York Academy of Sciences, 1134: 267-319.
Nielsen T G, Hansen BW. 1999: Plankton community structure and carbon cycling on the western coast of
SC
Greenland during the stratified summer situation. Part I. Hydrography, phytoplankton and bacteria. Aquat. Mar. Ecol. 16: 205-216.
Olli K, Riser C W, Wassmann P, Ratkova T, Arashkevich E, Pasternak A. 2002. Seasonal variation in vertical flux
MA NU
of biogenic matter in the marginal ice zone and the central Barents Sea. Journal of Marine Systems, 38: 189-204.
Stroeve J C, Maslanik J, Serreze M C, Rigor I, Meier W, Fowler C. 2011. Sea ice response to an extreme negative phase of the Arctic Oscillation during winter 2009/2010, Geophysical Research Letters, 38(2), doi: 10.1029/2010GL045662.
Pabi S, van Dijken G L, Arrigo K R. 2008. Primary production in the Arctic Ocean, 1998-2006. Journal of Geophysical Research-Oceans, 113(C8).
ED
Pang H. 2007. Econometrics, Beijing: Science Press, 265-284. Rysgaard S, Nielsen T G. 2006. Carbon cycling in a high-arctic marine ecosystem – Young Sound, NE Greenland. Prog. Oceanogr., 71: 426-445.
PT
Qu B, Gabric A J, Matrai P A. 2006. the satellite-derived distribution of chlorophyll-a and its relation to ice concentration, radiation and sea surface temperature in the Barents Sea, Polar Biology, 29: 196-210.
CE
Qu, B, Gabric A J. 2010. Using genetic algorithms to calibrate a dimethylsulfide production model in the Arctic Ocean. Chinese Journal of Oceanology and Limnology, 28(1): 1-10. doi: 10.1007/s00343-010-9062-x. Qu B, Lu H L, Gabric A J, Lin D R, Qian F, Zhao W H. 2011. Using MODIS satellite data to analyze the
AC
relationship between chlorophyll a and aerosol optical depth in the Greenland Sea. Chinese Journal of Polar Research, 21(1): 68-76. Qu, B, Gabric A J, Zhu J, Lin D, Qian F, Zhao M. 2012. Correlation between sea surface temperature and wind speed in Greenland Sea and their relationships with NAO variability, Water Science and Engineering, 5(3): 304-315 Qu B, Gabric, A J, Lu H, Lin D. 2014. Spike in phytoplankton biomass in Greenland Sea during 2009 and the correlations among chlorophyll-a, aerosol optical depth and ice cover. Chinese Journal of Oceanology and Limnology, 32 (2): 241 – 254. Sanjay K J, Goswam A, Saraf A K. 2008. Accuracy assessment of MODIS, NOAA and IRS data in snow cover mapping under Himalayan conditions. International Journal of Remote Sensing, 29(20): 5863-5878, doi: 10.1080/01431160801908129. Schneider W, Budéus G. 1994. The North East Water Polynya (Greenland Sea). I. A physical concept of its generation. Polar Biology, 14: 1-9. Schuenemann K C, Cassano J J. 2010. Changes in synoptic weather patterns and Greenland precipitation in the 20th and 21st centuries: 2. Analysis of 21st century atmospheric changes using self-organizing maps. Journal of Geophysical Research: Atmospheres, 155(D5). doi: 10.1029/2009JD011706.
16
ACCEPTED MANUSCRIPT Schweiger A J, Ron W L, Vavrus S, Francis J F. 2008. Relationships between Arctic sea ice and clouds during autumn. Journal of Climate, 21: 4799-4810. Screen J A, Simmonds I. 2010. Increasing fall-winter energy loss from the Arctic Ocean and its role in Arctic temperature amplification. Geophysical Research Letters, 37(16): L16707. Serreze M, Barry R. 2011. Processes and impacts of Arctic amplification global and planetary change: a research
T
synthesis. Global and Planetary Change, 77: 85-96.
RI P
Smetacek V, Nicol S. 2005. Review polar ocean ecosystems in a changing world. Nature, 437: 362-368 Smith K L, Robison BH, Helly J J, Kaufmann R S, Ruhl HA, Shaw T J, Twining B S, Vernet M. 2007. Freedrifting icebergs: hot spots of chemical and biological enrichment in the Weddell Sea. Science, 317(5837): 478-482.
SC
Statham, P J, Skidmore M, Tranter M. 2008. Inputs of glacially derived dissolved and colloidal iron to the coastal ocean and implications for primary productivity. Global Biogeochemical Cycles, 22: GB3013. doi: 10.1029/2007GB003106.
MA NU
Straneo F, Curry R G, Sutherland D A, Hamilton G S, Cenedese C, Våge K, Stearns L. 2011. Impact of fjord dynamics and glacial runoff on the circulation near Helheim Glacier. Nature Geoscience, 4: 322-327. doi: 10.1038/ngeo1109.
Su W, Charlock, T P, Rose F G. Rutan D. 2007. Photosynthetically active radiation from Clouds and the Earth's Radiant Energy System (CERES) products. Journal of Geophysical Research, 112: G02022. 1-11. doi: 10.1029/2006JG000290.
Tovar-Sánchez, A, Duarte C M, Alonso J C, Lacorte S, Tauler R, Galbán-Malagón C. 2010. Impacts of metals and
ED
nutrients released from melting multiyear Arctic sea ice. Journal of Geophysical Research: Oceans, 115(C7): C07003.
Wassmann P, Ratkova T N, Andreassen I, Vernet M, Pedersen G, Rey F. 1999. Spring bloom development in the
AC
CE
PT
marginal ice zone and the central Barents Sea. Marine Ecology, 20: 321-346.
17
ACCEPTED MANUSCRIPT Figure Captions:
Figure 1: Map of Study region in Greenland Sea (highlighted box is for 20°W–10°E, 75°N–80°N). White arrows indicate the surface flow directions.
RI P
T
Figure 2: Satellite image of CHL (a) on day 180 (2010) and (b) day 160 (2011) around the study region.
SC
Figure 3: Mean CHL climatology (with standard deviation) compared with CHL in year 2010 in the study region (75°N-80°N, 20°W-10°E). Figure 4: Monthly mean CHL and sea ice concentration (ICE) in the study region with trend line (dashline) for CHL.
MA NU
Figure 5: Mean CHL and Ice Melt time series in 2010 in the study region (75°N-80°N, 20°W-10°E) Figure 6: Mean wind speed (bars) and direction (lines) for year 2003-2012 compared with those in year 2010 in the study region.
ED
Figure 7: NAO (solid dot) and AO (hollow dot) monthly mean for year 2003-2012. Red part indicates NAO values in year 2010.
AC
CE
PT
Figure 8: Weekly time series of Ice Melt and NAO in year 2010 in the study region ( 75°N -80°N, 20°W-10°E).
18
ACCEPTED MANUSCRIPT Highlights ● The causes of an unusually early and intense phytoplankton bloom in the
AC
CE
PT
ED
MA NU
SC
RI P
T
northern Greenland Sea during 2010 are investigated. ● Earlier and more extensive sea ice melt, persistent negative NAO, and changing wind directions were the main drivers of the bloom. ● Multivariate lagged regression analysis shows the bloom was correlated with the timing of sea ice melt,PAR and SST.
19