Vm

LWa

Cb

c

d

a

ARR2A

KSGNCI

1

aseo

aAT

h0

Agricultural and Forest Meteorology 218–219 (2016) 288–297

Contents lists available at ScienceDirect

Agricultural and Forest Meteorology

j o ur na l ho me pag e: www.elsev ier .com/ locate /agr formet

alidation of the global land data assimilation system based oneasurements of soil temperature profiles

ei Wanga,b,∗, Xiuping Lia,∗, Yingying Chena,b, Kun Yanga,b, Deliang Chenc, Jing Zhoua,enbin Liua, Jia Qia, Jianbin Huangd

Key Laboratory of Tibetan Environment Changes and Land Surface Processes, Institute of Tibetan Plateau Research, Chinese Academy of Sciences, Beijing,hinaCAS Center for Excellence in Tibetan Plateau Earth Sciences, Beijing, ChinaDepartment of Earth Sciences, University of Gothenburg, Gothenburg, SwedenMinistry of Education Key Laboratory for Earth System Modeling, and Center for Earth System Science, Tsinghua University, Beijing, China

r t i c l e i n f o

rticle history:eceived 3 April 2015eceived in revised form5 December 2015ccepted 6 January 2016

eywords:oil temperature profileLDASoah land surface modelommunity land model

n situ measurements

a b s t r a c t

Soil temperature is a key parameter in the soil–vegetation–atmosphere system. It plays an importantrole in the land surface water and energy cycles, and has a major influence on vegetation growth andother hydrological aspects. We evaluated the accuracy of the soil temperature profiles from the GlobalLand Data Assimilation System (GLDAS) using nine observational networks across the world and aimedto find a reliable global soil temperature profile dataset for future hydrological and ecological stud-ies. In general, the soil temperature profile data generated by the Noah model driven by the GLDASforcing data (GLDAS Noah10 and GLDAS Noah10 v2) were found to have high skills in terms of daily,monthly, and mean seasonal variations, indicated by smaller bias and root-mean-square-error (RMSE)(both 4 ◦C). Further analysis showedthat the overestimation by GLDAS CLM10 was mainly caused by overestimation of the ground heatflux, determined by the thermal conductivity parameterization scheme, whereas the underestimation byGLDAS Noah10 was due to underestimation of downward longwave radiation from the forcing data. Thus,more accurate forcing data should be required for the Noah model and an improved thermal parameteri-

zation scheme should be developed for the CLM. These approaches will improve the accuracy of simulatedsoil temperatures. To our knowledge, it is the first study to evaluate the GLDAS soil temperatures withcomprehensive in situ observations across the world, and has a potential to facilitate an overall improve-ment of the GLDAS products (not only soil temperatures but also the related energy and water fluxes) aswell as a refinement of the land surface parameterization used in GLDAS.

© 2016 Elsevier B.V. All rights reserved.

. Introduction

Soil temperature is a key parameter in the soil–vegetation–

tmosphere transfer system. It plays an important role in theurface water and energy exchange during land–atmospherexchanges (e.g., Cheviron et al., 2005; Yang et al., 2013) becausef its strong influence on soil physical properties (Chen and Kling,

∗ Corresponding authors at: Key Laboratory of Tibetan Environment Changesnd Land Surface Processes, Institute of Tibetan Plateau Research, Chinesecademy of Sciences, Building 3, Courtyard 16, Lin Cui Road, Beijing, China.el.: +86 10 84097107; fax: +86 10 84097079.

E-mail addresses: wanglei@itpcas.ac.cn (L. Wang), lixiuping@itpcas.ac.cn (X. Li).

ttp://dx.doi.org/10.1016/j.agrformet.2016.01.003168-1923/© 2016 Elsevier B.V. All rights reserved.

1996; Or and Wraith, 1999; Grant and Or, 2004; Bachmann and vander Ploeg, 2002; Schneider and Goss, 2011). As a result, it is alsodetermined by energy flux and soil properties. Soil temperature isalso a primary control on CO2 production in most soils across differ-ent spatial and temporal scales (Lloyd and Taylor, 1994; Peterjohnet al., 1994; Raich and Potter, 1995; Rustad and Fernandez, 1998;Kirschbaum, 2000; Risk et al., 2002; Riveros-Iregui et al., 2007).Accurate knowledge of soil temperature is needed for numer-ous studies such as short-term forecasts (Godfrey and Stensrud,2008; Fan, 2009), sub-seasonal and seasonal forecasts (Mahanamaet al., 2008) and vegetation growth (McMichael and Burke, 1998).

However, soil temperature profiles remain difficult to measure atregional to global scales, although land surface temperatures canbe retrieved from satellites with acceptable accuracy (e.g., Sun andPinker, 2003; Wan, 2008; Li et al., 2013).

dx.doi.org/10.1016/j.agrformet.2016.01.003http://www.sciencedirect.com/science/journal/01681923http://www.elsevier.com/locate/agrformethttp://crossmark.crossref.org/dialog/?doi=10.1016/j.agrformet.2016.01.003&domain=pdfmailto:wanglei@itpcas.ac.cnmailto:lixiuping@itpcas.ac.cndx.doi.org/10.1016/j.agrformet.2016.01.003

L. Wang et al. / Agricultural and Forest Meteorology 218–219 (2016) 288–297 289

Table 1The observed soil temperature from nine observational networks and global land surface model outputs used in this study.

Soil temperaturedatasets

Type Land use Number ofstations

Soil layers depth (cm) Temporalresolution

Temporal extent

GLDAS CLM10 Numerical product Global product Global product 0–1.8, 1.8–4.5, 4.5–9.1,9.1–16.6, 16.6–28.9,28.9–49.3, 49.3–82.9,82.9–138.3, 138.3–229.6,and 229.6–343.3

3 h 200,201–201,212

GLDAS Noah10 Numerical product Global product Global product 0–10, 10–40, 40–100, and100–200

3 h 200,201–201,212

GLDAS Noah10 v2 Numerical product Global product Global product 0–10, 10–40, 40–100, and100–200

3 h 200,201–201,012

SGP (NorthAmerica)

In-situ observation Crop and grass 20 Stations 5, 15, 25, 35, 60, 85, 125,and 175

60 min 200,210–200,910

Mongolia(Mongolia)

In-situ observation Grass 16 Stations 3, 10, 40, and 100 30 min 200,210–200,412

MDB (Australia) In-situ observation Crop 18 Stations 3.5, 15, 45, and 75 30 min 200,210–200,805SASMAS (Australia) In-situ observation Grass and crop 14 Stations 0–5 30 min 200,602–200,712TP (China) In-situ observation Grass 38 Stations 0–5, 10, 20, and 40 30 min 201,008–201,212MAQU (China) In-situ observation Grass 20 Stations 5 30 min 200,807–200,908HOBE (Denmark) In-situ observation Crop and forest 30 Stations 0–5, 20–25, and 50–55 30 min 201,001–201,106REMEDHUS (Spain) In-situ observation Crop with some 24 Stations 5 30 min 200,504–201,212

tation

fba(taaotCa1Mislpm

gp

Fa

patchy forestSMOSMANIA

(France)In-situ observation Crop 21 S

Currently, in situ soil temperature observational networks areew and are sparsely distributed across the global. Thus, model-ased soil temperature products are considered a reasonablelternative. The Global Land Data Assimilation System (GLDAS)Rodell et al., 2004; Rodell and Kato, 2006) was produced withhe goal of realistically simulating the transfer of mass, energy,nd momentum between the soil and vegetation surfaces and thetmosphere. The GLDAS may be used to improve understandingf soil water dynamics, plant physiology, micrometeorology, andhe controls on atmosphere–biosphere–hydrosphere interactions.urrently, GLDAS drives four land surface models: Mosaic (Kosternd Suarez, 1992a, 1992b), Noah (Chen et al., 1996; Koren et al.,999; Ek et al., 2003; Betts et al., 1997), the Community Landodel (CLM; Dai et al., 2003) and the Variable Infiltration Capac-

ty model (VIC; Liang et al., 1994). The output variables includeoil moisture and temperature at multiple layers, and all majorand surface water and energy fluxes. However, soil temperaturerofile data are only available from the CLM model and the Noah

odel.The objective of this study is to fully investigate the accuracy of

lobal soil temperature profile products from the GLDAS by com-arison with large-scale (greater than 1◦ × 1◦) and small-scale (less

ig. 1. The soil temperature observational networks used in this study, including SGP in s well as HOBE, REMEDHUS and SMOSMANIA in European.

s 5, 10, 20, and 30 60 min 200,803–201,112

than 1◦ × 1◦) in situ observations from nine networks under differ-ent geographical settings and climatic controls.

2. Material and methods

2.1. Observational networks for the soil temperature profile

Unlike other routinely available hydro-meteorological variables(e.g., air temperature, and precipitation), the soil temperature pro-file is rarely observed on a regular basis at meteorological stations.It is even harder to build and maintain a larger scale (especiallygreater than 1◦ × 1◦) observational network for multi-site measure-ments of soil temperature profiles.

Nine in situ soil temperature profile datasets derived fromlarge-scale (greater than 1◦ × 1◦) and small-scale (less than 1◦ × 1◦)observational networks were used in this study (Table 1 andFig. 1). Three networks were distributed through the Coordinated

Enhanced Observing Period (CEOP) (Bosilovich and Lawford, 2002;Koike, 2004; Lawford et al., 2006), located in the Southern GreatPlains (SGP, USA), Murray-Darling Basin Murrumbidgee (MDB,Australia) and Mongolian Plateau (Mongolia, Kaihotsu et al., 2003),

USA, Mongolia in Mongolia, TP and MAQU in China, MDB and SASMAS in Australia,

2 st Me

rEmGragen(CRH2

2

saoRtsptth

voCN1es(X

iT1o(femNfdcddztr

p(waEt4i

90 L. Wang et al. / Agricultural and Fore

espectively. The aim of the CEOP, proposed under the Globalnergy and Water Cycle Experiment (GEWEX) and the World Cli-ate Research Program (WCRP), was to coordinate the data fromEWEX Continental Scale Experiments (CSEs) and those from other

egions. The remaining six in situ observational networks arerchived by the International Soil Moisture Network (http://ismn.eo.tuwien.ac.at/ismn/), which is an international cooperation tostablish and maintain a global in situ soil moisture database. Theseetworks are located in the Skjern River Catchment in DenmarkHOBE, Bircher et al., 2011), the central part of Tibetan Plateau inhina (TP, Yang et al., 2013), the source water region of the Yellowiver in China (MAQU, Su et al., 2011), Zamora of Spain (REMED-US), eastern Australia (SASMAS, Smith et al., 2012; Young et al.,008), and southern France (SMOSMANIA, Albergel et al., 2008).

.2. Soil temperature profiles from GLDAS

The GLDAS was developed to generate optimal fields of landurface states and fluxes using land surface modeling and datassimilation techniques, by integrating satellite- and ground-basedbserved data to drive four offline Land Surface Models (LSMs,odell et al., 2004). Observation-based meteorological forcings andhe best available analyses from atmospheric data assimilationystems are employed to force the models globally at multi-le resolutions. The GLDAS data are archived and distributed viahe website of the Goddard Earth Sciences Data and Informa-ion Services Center (http://disc.sci.gsfc.nasa.gov/hydrology/data-oldings).

Because the available soil temperature outputs are only pro-ided by the CLM model and the Noah model, these two modelutputs are evaluated against the observed soil temperature. TheLM used in the GLDAS includes superior components from theational Center for Atmospheric Research (NCAR) LSM (Bonan,998) and the Biosphere–Atmosphere Transfer Scheme (Dickinsont al., 1986). The CLM model developed by a consideration ofubgrid-heterogeneity has been widely used in previous studiese.g., Niu and Yang, 2006; De Lannoy Gabriëlle et al., 2006; Tian andie, 2008).

The Noah LSM (Chen et al., 1996; Koren et al., 1999) was pos-tioned within the GEWEX Continental-Scale International Project.he Noah has been used operationally within NCEP models since996, and also has been shown to be effective in reproducingbserved energy and water budget without the complexity of tilingRobock et al., 2003) and also estimates the soil temperature for dif-erent soil layers and time scales (Xia et al., 2013). Although somenhancements were made in GLDAS version 2.0 including Noahodel version upgrade, the largest difference between GLDAS 1.0oah simulations and GLDAS 2.0 Noah simulations mainly arises

rom the change to the input forcing data. Both CLM and Noah areriven by the same forcing data and parameters, including landover and soils, in GLDAS version 1.0, in which the input forcingata are from constantly updated meteorological data. The largestistinctions between the two versions arise from the parameteri-ation schemes in the model itself. In the latest GLDAS version 2.0,he input forcing data have been updated to the Princeton meteo-ological forcing dataset (Sheffield et al., 2006).

In this study, we evaluate the 3-h 1.0◦ × 1.0◦ soil temperaturerofiles produced by the Noah (hereafter Noah10) and the CLMhereafter CLM10; Dai et al., 2003) from GLDAS Version 1.0, asell as the Noah from GLDAS Version 2.0 (hereafter Noah10 v2)

gainst the observed values (Chen et al., 1996; Koren et al., 1999;k et al., 2003; Betts et al., 1997). The CLM10 model produces soilemperature profiles at 10 layers and both Noah10 models produce

layers. More detailed information about the model data is shownn Table 1.

teorology 218–219 (2016) 288–297

2.3. Evaluation approach

Evaluation of the GLDAS products is a challenging task, becauseconventional measurements of soil temperature profiles are usu-ally made at individual points, while the GLDAS provides griddedvalues for a much larger spatial extent (e.g., 1◦ × 1◦). In this study,we aimed to use the large- (greater than 1◦ × 1◦) and small-scale(less than 1◦ × 1◦) observational networks of soil temperature pro-files for evaluating the GLDAS estimates with different land surfacemodels, by averaging multi-site in situ measurements at differentsoil depths. Only large-scale networks with uniform distributionwere selected in our study to avoid the impact of station distri-bution on the evaluation. For the small-scale (less than 1◦ × 1◦)network, there is no impact of station distribution on the evalua-tion results, because all stations were located within the same gridcell.

In GLDAS, the division of the soil layer is different betweendifferent land surface models. From the surface to 100 cm soillayer, the CLM model has 8 soil layers (0–1.8, 1.8–4.5, 4.5–9.1,9.1–16.6, 16.6–28.9, 28.9–49.3, 49.3–82.9, and 82.9–138.3 cm), andboth Noah10 models have 3 layers (0–10, 10–40, and 40–100 cm).To better match the depths between the in situ observations andthe GLDAS simulations, the averaged values for the uppermostthree layers of CLM (0–9.1 cm), the uppermost four layers of CLM(0–16.6 cm), and the values for the first layer of Noah (0–10 cm)were evaluated against the average of the observations for 0–5 cmand 0–10 cm. The soil temperature for the second layer of Noah(10–40 cm) and the averaged values for the fourth to the sixth lay-ers of CLM (9.1–49.3 cm) were evaluated against the average of theobservations at 10–40 cm. Finally, the averaged values for the sev-enth to eighth layers of CLM (49.3–138.3 cm) and the third layer ofNoah (40–100 cm) were also evaluated against the average of theobservations at 40–100 cm. Weighted averages were produced forcorresponding soil layers of the in situ observations and the GLDASsimulations for 0–5, 0–10, 10–40, and 40–100 cm.

The model performance is evaluated using the mean bias (BIAS),the root-mean-square-error (RMSE), and the coefficient of variation(CV). These are defined as follows:

BIAS =N∑

i=1

Ai − BiN

(1)

RMSE =

√√√√ N∑i=1

(Ai − Bi)2N

(2)

CV =

√√√√1n

N∑i=1

(Ai − Ā)2

Ā(3)

where N represents the number of months of the study period, andAi and Bi are from the daily or monthly observed soil temperatureand the GLDAS values, respectively.

Because most in situ observational data are only available afterthe year 2000, we evaluated the soil temperature profiles of GLDASbeginning in 2000, focusing on daily, monthly and seasonal timescales.

3. Results

Soil has three important thermal properties: thermal conductiv-

ity, thermal diffusivity and volumetric heat capacity, all of whichare crucial in determining the land surface energy and water bud-get. Soil thermal conductivity can be calculated using the thermaldiffusivity and volumetric heat capacity, while soil volumetric heat

http://ismn.geo.tuwien.ac.at/ismn/http://ismn.geo.tuwien.ac.at/ismn/http://ismn.geo.tuwien.ac.at/ismn/http://ismn.geo.tuwien.ac.at/ismn/http://ismn.geo.tuwien.ac.at/ismn/http://ismn.geo.tuwien.ac.at/ismn/http://ismn.geo.tuwien.ac.at/ismn/http://ismn.geo.tuwien.ac.at/ismn/http://disc.sci.gsfc.nasa.gov/hydrology/data-holdingshttp://disc.sci.gsfc.nasa.gov/hydrology/data-holdingshttp://disc.sci.gsfc.nasa.gov/hydrology/data-holdingshttp://disc.sci.gsfc.nasa.gov/hydrology/data-holdingshttp://disc.sci.gsfc.nasa.gov/hydrology/data-holdingshttp://disc.sci.gsfc.nasa.gov/hydrology/data-holdingshttp://disc.sci.gsfc.nasa.gov/hydrology/data-holdingshttp://disc.sci.gsfc.nasa.gov/hydrology/data-holdingshttp://disc.sci.gsfc.nasa.gov/hydrology/data-holdings

st Me

cWiCiemwipt

3

didrdGs(

mnftctNe

ccuv

Fa5

L. Wang et al. / Agricultural and Fore

apacity is determined by soil composition and soil porosity (Vanijk, 1963). As a result, the calculation of soil thermal conductiv-

ty is important to accurately estimate the soil temperature. In theLM model, the soil thermal conductivity parameterization scheme

s from Farouki’s scheme (Farouki, 1981), while this parameter isstimated using Johansen’s scheme (Johansen, 1975) in the Noahodel. It is expected that a different simulated soil temperatureill be estimated by each scheme across the world. In the follow-

ng we will evaluate the soil temperature from the GLDAS modelroducts among nine networks at the daily, monthly and seasonalime scales.

.1. Evaluation of soil temperature at the daily scale

To evaluate the model performance of soil temperature at theaily time scale, the cumulative distribution functions act on the

n situ observations and the GLDAS model outputs for four soilepths (0–5, 0–10, 10–40, and 40–100 cm) (shown in Fig. 2). Theesults show that the three GLDAS outputs can reflect the generalistribution of soil temperature at the different soil depths. TheLDAS products can therefore reproduce the daily variation foroil temperature, as indicated by the higher correlation coefficientslarger than 0.90), with an exception of CLM10 at TP.

For the uppermost soil layers (0–5 or 0–10 cm), the CLM10odel consistently overestimated the soil temperature for all

etworks except SASMAS and SMOSMANIA. The soil temperaturerom the Noah10 and the Noah10 v2 models clearly reproducedhe daily variability of observations, dominated by correlationoefficients greater than 0.90. However, they underestimated soilemperature in all seasons for all networks except for SGP. Theoah10 v2 products display the best performance with the small-st BIAS and RMSE relative to the other two models.

For the middle soil depths (10–40 cm), the three GLDAS outputs

onsistently overestimated the soil temperature at the SGP networkompared with the observations; however there were concurrentnderestimates at SMOSMANIA. The soil temperature from bothersions of Noah10 underestimated the observed values at the TP

Obs CLM10 No

Soil temper

TP (0-10cm)

-30 -20 -10 0

Mongolia (0-10c

SGP (0-10cm)

MDB (0-10cm)

MAQU (0-5cm)

REMEDHUS (0-5cm)

SASMAS (0-5cm)

HOBE (0-5cm)

SMOSMANIA

(0-5cm)

-40 -30 -20 -10 0 10 20 30 40 50

0.8

0.6

0.4

0.2

0.0

1.0

0.8

0.6

0.4

0.2

0.0

0.8

0.6

0.4

0.2

0.0

0.8

0.6

0.4

0.2

0.0

0.8

0.6

0.4

0.2

0.0

ig. 2. Distribution of daily soil temperature (unit: ◦C) from GLDAS and in-situ measuremnd MDB networks for the different soil depth for 0–5, 0–10, 10–40 and 40–100 cm. The N46 for HOBE, 427 for MAQU, 2832 for REMEDHUS, 699 for SASMAS, 1401 for SMOSMAN

teorology 218–219 (2016) 288–297 291

and MDB networks, while the CLM10 produced overestimates atthese networks.

For the deeper soils (40–100 cm) at the SGP network, the threemodel products exhibited better simulated performance with acorrelation coefficient of 0.99 compared with the observations,although they consistently overestimated the values, especially inthe case of the CLM model.

The GLDAS Noah10 and Noah10 v2 therefore underestimatedthe soil temperature for most of the networks, while the CLM10tended to overestimate the soil temperatures at all networks exceptfor SMOSMANIA, especially for the peak values. An overestimationof the thermal conductivity in the CLM model, already found at theTP (Chen et al., 2012), may be one of the reasons for higher soiltemperatures in the studied regions.

3.2. Evaluation of monthly soil temperature profiles

Figs. S1–S5 (shown in the Appendix) display the observedmonthly soil temperature profiles and the difference between theobservations and the GLDAS monthly soil temperature profiles atthe SGP, Mongolia, TP, MDB and SMOSMANIA networks. The resultsfrom the observations display a clear seasonal variation of soil tem-perature from the topsoil to deeper soils. The temperature alsogradually decreases from the topsoil to the deeper soil layers.

Although there are different vegetation covers in the SGP (crop-land and grassland) and the Mongolia (grassland), the three modelsdisplay similar distributions of soil temperature (in Figs. S1 andS2). In general, both versions of the Noah products (Noah10 andNoah10 v2) performed better than the CLM10 product in describ-ing soil temperature profiles, with a BIAS of 0.3–1.7 ◦C and RMSEof 0.9–2.0 ◦C at the SGP and a BIAS of between −0.9 and −0.2 ◦Cand a RMSE of 0.9–1.3 ◦C at Mongolia (also see Table 2). However,both Noah10 products performed an overestimation in the warm

seasons and an underestimation in the cold seasons; the CLM10products consistently overestimated the soil temperatures.

As the highest elevation observational network of the ninenetworks, the TP network is characterized by low biomass, high soil

ah10 Noah10_v2

ature (oC)

10 20 30 40

m)

-40 -30 -20 -10 0 10 20 30 40 50

SGP (10-40cm)

TP (10-40cm)

MDB (10-40cm)

SGP (40-100cm)

SMOSMANIA

(10-40cm)

ents at the Mongolia, SGP, TP, MAQU, SASMAS, HOBE, REMEDHUS, SMOAMANIAumber of data points is 779 for Mongolia, 2491 for SGP, 884 for TP, 2037 for MDB,

IA, respectively.

292 L. Wang et al. / Agricultural and Forest Meteorology 218–219 (2016) 288–297

Table 2Evaluation of monthly GLDAS soil temperature (unit: ◦C) with in situ observations. Bold values are the lowest error metrics. Nrepresents the number of data points used in the evaluation.

Observations GLDAS BIAS RMSE

SGP (N = 85) 0–10 cm soil layer CLM10 (0–10.0 cm) 4.4 4.8NOAH10 (0–10 cm) 1.3 1.7NOAH10 v2 (0–10 cm) 1.7 2.0

10–40 cm soil layer CLM10 (10–40 cm) 4.2 4.5NOAH10 (10–40 cm) 0.7 1.1NOAH10 v2 (10–40 cm) 1.2 1.5

40–100 cm soil layer CLM10 (40–100 cm) 4.1 4.4NOAH10 (40–100 cm) 0.3 0.9NOAH10 v2 (40–100 cm) 0.9 1.3

Mongolia (N = 27) 0–10 cm soil layer CLM10 (0–10.0 cm) 5.0 6.1NOAH10 (0–10 cm) −0.9 1.3NOAH10 v2 (0–10 cm) −0.2 0.9

TP (N = 29) 0–10 cm soil layer CLM10 (0–10.0 cm) 7.0 7.8NOAH10 (0–10 cm) −2.8 3.2NOAH10 v2 (0–10 cm) −1.6 2.4

10–40 cm soil layer CLM10 (10–40 cm) 6.9 7.6NOAH10 (10–40 cm) −4.4 4.5NOAH10 v2 (10–40 cm) −1.0 1.2

MAQU (N = 14) 0–5 cm soil layer CLM10 (0–5.0 cm) 1.3 3.6NOAH10 (0–10 cm) −2.7 3.0NOAH10 v2 (0–10 cm) −3.8 4.5

MDB (N = 68) 0–10 cm soil layer CLM10 (0–10.0 cm) 3.3 4.8NOAH10 (0–10 cm) −1.5 1.8NOAH10 v2 (0–10 cm) −0.5 1.0

SASMAS (N = 23) 0–5 cm soil layer CLM10 (0–5.0 cm) −3.5 3.8NOAH10 (0–10 cm) −1.8 2.3NOAH10 v2 (0–10 cm) −1.4 1.9

HOBE (N = 18) 0–5 cm soil layer CLM10 (0–4.5 cm) 0.1 0.7NOAH10 (0–10 cm) −0.3 1.3NOAH10 v2 (0–10 cm) 0.1 1.3

REMEDHUS (N = 93) 0–5 cm soil layer CLM10 (0–5.0 cm) 1.0 2.6NOAH10 (0–10 cm) −1.2 1.6NOAH10 v2 (0–10 cm) −0.1 1.1

SMOSMANIA (N = 46) 0–10 cm soil layer CLM10 (0–10.0 cm) −2.2 2.3NOAH10 (0–10 cm) −2.7 2.8NOAH10 v2 (0–10 cm) −1.8 2.0

metaBcNt

svtouo(s

tdsa

10–40 cm soil layer

oisture dynamic range, and a typical soil freeze–thaw cycle (Yangt al., 2013). At this network, the CLM10 consistently overestimatedhe soil temperature with both the BIAS and RMSE greater than 7 ◦C,nd didn’t reproduce the repeating soil freeze–thaw cycle (Fig. S3).oth Noah10 models had better simulations of soil freeze–thawycles, but followed by negative biases. The soil temperatures ofoah10 v2 show improvements over those of Noah10, although

he Noah10 v2 data covers a shorter period.At the MDB network (Fig. S4), both Noah10 models show

imilar soil temperature profiles compared with the in situ obser-ations, and the Noah10 v2 products show a smaller BIAS (−0.5 ◦C)han the Noah10 products (BIAS = −1.5 ◦C). The CLM10 productsverestimated the soil temperature from January to August, andnderestimated it for the other months, followed by a mean BIASf 3.3 ◦C and RMSE of 4.8 ◦C. However, at the SMOSMANIA networkFig. S5), all three model products consistently underestimated theoil temperature, especially for the cold seasons.

In short, the comparisons of soil temperature profiles at five of

he nine studied networks reveal that all three models can repro-uce the gradual variation of soil temperature with the variation ofoil depth. However, the CLM10 model overestimated soil temper-ture profiles, especially in the warm seasons at most networks.

CLM10 (10–40 cm) −2.1 2.3NOAH10 (10–40 cm) −3.1 3.5NOAH10 v2 (10–40 cm) −2.2 2.8

In addition, both Noah10 models display similar and comparabledistributions of soil temperature compared with the observationsand the Noah10 v2 shows better performance than the Noah10.

3.3. Evaluation of seasonal mean soil temperature

The mean seasonal variation of soil temperature and itsstandard deviation (as indicated by error bar) at the nine networksfor 0–5 cm, 0–10 cm, 10–40 cm, and 40–100 cm was evaluatedbetween the observations and the model results (Fig. 3). Table 2lists the error metrics for these monthly values.

All three land surface models reproduced the seasonal vari-ation of soil temperature but display different bias and range ofvariation compared with the observed values. The three modelssystematically overestimated soil temperature at SGP, possiblybecause of the higher sensitivity of the thermal conductivityparameterization schemes at this network. However, both Noah10

models underestimated the soil temperature at all networksexcept for SGP. The performance of the Noah10 v2 was better thanthat of the Noah10 model with a smaller bias, although it coversa shorter period. The performance of CLM10 was slightly poorer

L. Wang et al. / Agricultural and Forest Meteorology 218–219 (2016) 288–297 293

F S ande ation o

te

aebsbb

4

fdtum

patszicctcgpgnvamss

ig. 3. Averaged seasonal mean variation of soil temperature (unit: ◦C) from GLDArror bar (unit: ◦C) (i.e., standard deviation of soil temperature) represents the vari

han both Noah10 models, as indicated by larger bias and range,specially in the warm seasons.

Overall, the three models displayed the different simulatedbilities around the world. The CLM model showed a general over-stimation except for SASMAS and SMOSMANIA networks, whileoth Noah10 model products performed better followed by con-istent underestimations and the Noah10 v2 model results displayetter performance. In addition, the three models display smalleriases in the uppermost soil layers than in the deeper soil layers.

. Discussions

These uncertainties of soil temperature in models originaterom various sources, including the input meteorological forcingata and model parameterization schemes. We investigate poten-ial reasons for CLM10s overestimations and both Noah10 models’nderestimations in the following sections, based on the inputeteorological forcing data and model parameters.Because same forcing data were used, the different model

arameterization schemes between the CLM and the Noah modelsre likely to cause discrepancies in their simulated soil tempera-ures. The soil thermal conductivity is one of the most importantoil thermal properties. Different thermal conductivity parameteri-ation schemes from Johansen (1975) and Farouki (1981) were usedn the Noah10 and the CLM10 models, respectively. Both thermalonductivity parameterization schemes overestimated the thermalonductivity, which was greater for the Farouki (1981) scheme fromhe CLM10 model in particular (Chen et al., 2012). Higher thermalonductivity parameterization scheme tends to generate a largerround heat flux, which thus causes the overestimation of soil tem-eratures from top to deeper soil layers. A comparison of monthlyround heat flux from the CLM10 and Noah10 outputs at the nineetworks is presented in Fig. 4. The CV (coefficients of variation)alues are shown in the figures to emphasize the variation of the

mplitude. In general, the CLM10 (with larger CV values) overesti-ates the amplitude of ground heat flux more than Noah10 (with

maller CV values) in the summer (warm) and winter (cold) sea-ons, especially for areas with short vegetation (SGP, MDB, MAQU,

in-situ measurements at nine networks for 0–5, 0–10, 10–40 and 40–100 cm. Thef soil temperature during the study period of each network.

and REMEDHUS). At the daily scale, in periods without much dif-ference in soil moisture between day-time and night-time, we alsofound that the CLM10 model overestimated the amplitude of theground heat flux compared with the Noah10 model around mid-day (warm) and midnight (cold) for regions with short vegetation(SGP, TP, MDB, MAQU, and Mongolia, shown in Fig. 5). Therefore,the difference in thermal conductivity parameterization betweenthe two land surface models is likely to be responsible for theirdifferent performance in soil temperature simulations.

The major difference between the Noah10 and Noah10 v2 mod-els was the input forcing data. The radiation flux is a dominantfactor in soil temperature. The flux used in GLDAS Noah10 isfrom the Air Force Weather Agency Radiation, while the flux inGLDAS Noah10 v2 is mainly based on the NASA World ClimateResearch Programme/Global Energy and Water—Cycle Experiment(WCRP/GEWEX) Surface Radiation Budget (SRB) datasets (Cox et al.,2006). The GEWEX/SRB downward short—and longwave radiation,measured based on an elevation correction from Yang et al. (2006),have a consistently reliable performance in both flat and high-land areas. In the current study, taking the MDB network as anexample, we compared the input and output radiation variablesbetween Noah10 and Noah10 v2 (Fig. 6). Comparable downwardshortwave radiation was found between Noah10 and Noah10 v2.However, taking the GEWEX/SRB data as a reference, the downwardlongwave radiation flux from Noah10 showed larger underesti-mates, which leads to overestimates in the net longwave radiation,and underestimates in the simulated net total radiation. Therefore,Noah10 tends to produce lower soil temperatures than Noah10 v2,because of the lower net total radiation flux. The same phenomenonoccurred in the Mongolia, SASMAS, and REMEDHUS networks,where the GLDAS Noah10 v2 outputs were available.

Even with updated forcing data, Noah10 v2 still slightly under-estimated soil temperatures in most sparsely-vegetated networks(e.g., MDB, TP, SMOSMANIA, Mongolia, SASMAS, and REMEDHUS).

This may be caused by an overestimate of the thermal roughnesslength by the Noah model in bare soil, sparsely vegetated or areaswith short vegetation (Hogue et al., 2005; LeMone et al., 2008;Wang et al., 2011; Chen et al., 2011). This overestimation of thermal

294 L. Wang et al. / Agricultural and Forest Meteorology 218–219 (2016) 288–297

201008 201108 201208200210 200410 200610200210 200410 200610 200810

200 803 20 090 3 201003 201103

CLM10 Noah1 0

-20

-10

0

10

20SGP

-20

-10

0

10

20TP

-20

-10

0

10

20MDB

-20

-10

0

10

20SMOSMANIA

-20

-10

0

10

20MAQU

-40

-20

0

20

40Mongolia

-20

-10

0

10

20REMEDHUS

-20

-10

0

10

20SASMA S

2010 01 2010 07 201101-20

-10

0

10

20HOBE

200 807 200901 200 907

200504 200704 200904 201104200210 200310 200410 200602 200 612 20 0710

CLM10: 39.31Noah10: 15.47

CLM10 : -13.30Noah10: 2.08

CLM10 : 24 .41Noah10: 6.10

CLM10 : -13.85Noah10: 68 .49

CLM10 : 25 .63Noah10: 9.36

CLM10: -7.95Noah1 0: 53.46

CLM10 : 12 .37Noah10: 2.03

CLM10: 4.36Noah 10: -19 .25

CLM10: 11.50Noah 10: 72 .84

Time

Gro

und

heat

flux

(w m

-2)

F oah10s

rfl

featLrs

Fs

ig. 4. Comparison of monthly ground heat flux from the CLM10 output and the Nhown in the figures.

oughness length resulted in an underestimation of ground heatux, and therefore an underestimated soil temperature.

In addition to the parameterization scheme itself and the inputorcing data, the land cover is also another factor to directly influ-nce the soil temperature, because the fluxes and storages of energynd water at the land surface are strongly tied to the properties of

he vegetation. Mean seasonal variations of LAI from the Globaland Surface Satellite (Zhao et al., 2013) products are shown toeflect the conditions of vegetation growth (shown in Fig. 7). Thesehow that all in situ networks are covered by short vegetations (LAI

-60

-30

0

30

60

90

120SGP

-100

-50

0

50

100

150

200TP

-40

0

40

80SMOS MANI A

-40

0

40

80HOBE

-100

-50

0

50

100

150 Mongolia

-60

-30

0

30

60

90

120REMEDHU S

CLM10

00 03 06 09 12 15 18 21 00 03 06 09 Hour

mea

n gr

ound

hea

t flu

x (w

m-2

)

CLM: 70 .35Noah:59.28

CLM: 40 .03Noah:78.34

CLM: 15 .00Noah:43.98

CLM: 82.74Noah:29.55

CLM: -1415.Noah:116.66

ig. 5. Comparison of 3-h mean ground heat flux from the CLM10 output and the Noah1hown in the figures.

output in the nine networks (unit: W m−2). The coefficients of variation (CV) are

is lower than 4), and even bare ground in the cold seasons. Gener-ally, the CLM10 overestimated the amplitude of ground heat fluxmore than both Noah10 models, especially in the summer (warm)and winter (cold) seasons for the shorter vegetation regions (SGP,TP, MDB, MAQU, and Mongolia). However, in the richer vegetationregions (HOBE, SMOSMANIA and REMEDHUS), the CLM10 model

performance improved followed by decreased biases (shown inTable 2). This suggests that the biases of soil temperature causedby the thermal conductivity parameterization scheme in the CLM10are decreasing and weaken due to the effect of rich vegetation.

-80

-40

0

40

80

120

160MDB

-100

-50

0

50

100

150 MAQU

-40

0

40

80SASMAS

Noa h10

12 15 18 21 00 03 06 09 12 15 18 21(UTM)

CLM: 398 .49Noah:17.54

CLM: 86 .62Noah:66.34

CLM: 180 .12Noah:21.10

80 CLM: -2575 .42Noah:227.68

0 output in the nine networks (unit: W m−2). The coefficients of variation (CV) are

L. Wang et al. / Agricultural and Forest Meteorology 218–219 (2016) 288–297 295

SRB Noah1 0 Noah1 0_v 2

2003 2004 2005 2006 20072003 2004 2005 2006 2007

0

50

100

150

200Net longwave radiatio n

0

100

200

300Net shortwave radia tio n

0

50

100

150

200Net radia tio n

50

100

150

200

250

300

350

400Down sho rtwave radiatio n

260

300

340

380

420Down longwave radiation

350

450

550

650

750Down tota l radiat ion

Input Output

Time

Rad

iati

on v

alue

s (w

m-2

)

Fig. 6. Comparison of monthly downward shortwave, longwave and total radiation from the input forcing data as well as net shortwave, longwave and total radiation(W m−2) from the output results from the Noah10 model and the Noah10 v2 model, respectively, at the MDB. The triangle symbols represent the radiation data from theNASA World Climate Research Programme/Global Energy and Water-Cycle Experiment (WCPP/GEWEX) Surface Radiation Budget (SRB).

0.0

1.0

2.0

3.0

4.0

5.0 HOBE SMOSMANIA RE MEDHUS

0.0

1.0

2.0

3.0

4.0

5.0 Mongoli a TP MAQU

0.0

1.0

2.0

3.0

4.0

5.0 SGP MD B SASMAS

LA

I

1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12Month

und G

5

e

Fig. 7. Mean seasonal variation of Leaf Area Index (LAI) from the gro

. Conclusions

Soil temperature profiles from the GLDAS model outputs werevaluated by comparison with soil temperature profiles from nine

lobal Land Surface Satellite product over the 2002 to 2012 periods.

observational networks located around the world, at the daily,monthly, and seasonal time scales.

The CLM10 product produced overestimates of soil temper-ature profiles, except for SASMAS and SMOSMANIA networks.

2 st Me

Tvtge

dtvaCiudvmh

A

Rua(AtGsbvEdULdbsSt

A

t

R

A

B

B

B

B

B

C

96 L. Wang et al. / Agricultural and Fore

he Noah10 and its updated version (Noah10 v2) generally pro-ided realistic estimates of soil temperature profiles, althoughhey underestimated all soil temperature values except for SGP. Ineneral, Noah10 v2 performed better than the Noah10 model forstimating the soil temperature.

We conclude that three model products can reproduce theepth-averaged daily, monthly, and mean seasonal variation of soilemperature from the topsoil to the deeper soils, together with theertical soil temperature profiles, but there were clear discrepancynd shortcomings of the different models. The overestimation byLM10 was mainly caused by the thermal conductivity parameter-

zation scheme, while the underestimation by Noah10 was due tonderestimation of downward longwave radiation from its forcingata. Although the performance of different land surface modelsaries between networks and between soil layers, the Noah10 v2odel results are most applicable for future studies because of its

igher accuracy.

cknowledgments

This study was financially supported by the National Key Basicesearch Program of China (2013CBA01800), the National Nat-ral Science Foundation of China (Grants 41322001, 41405076,nd 41190083), Key Technologies R&D Program of China2013BAB05B03), and the Hundred Talents Program of Chinesecademy of Sciences (Dr. Lei Wang). Jing Zhou was supported by

he National Natural Science Foundation of China (41401080). TheLDAS data used in this study were acquired as part of the mis-ion of NASA’s Earth Science Division and archived and distributedy the Goddard Earth Sciences (GES) Data and Information Ser-ices Center (DISC). The original data for the CEOP (Coordinatednhanced Observing Period) soil moisture and soil temperatureatasets in this study were provided by Prof. I. Kaihotsu (Hiroshimaniversity), Prof. Toshio Koike (University of Tokyo) and Prof. Scotoehrer (NCAR/Earth Observing Laboratory). The soil temperatureataset over the Tibetan Plateau used in this study was providedy the Data Assimilation and Modeling Center for Tibetan Multi-pheres, Institute of Tibetan Plateau Research, Chinese Academy ofciences. The remaining in-situ observational networks were fromhe International Soil Moisture Network (ISMN).

ppendix A. Supplementary data

Supplementary material related to this article can be found, inhe online version, at doi:10.1016/j.agrformet.2016.01.003.

eferences

lbergel, C., Rüdiger, C., Pellarin, T., Calvet, J.C., Fritz, N., Froissard, F., Suquia, D.,Petitpa, A., Piguet, B., Martin, E., 2008. From near-surface to root-zone soilmoisture using an exponential filter: an assessment of the method based onin-situ observations and model simulations. Hydrol. Earth Syst. Sci. 12,1323–1337.

achmann, J., van der Ploeg, R.R., 2002. A review on recent developments in soilwater retention theory: interfacial tension and temperature effects. J. PlantNutr. Soil Sci. 165, 468–478.

etts, A.K., Chen, F., Mitchell, K.E., Janjic, Z.I., 1997. Assessment of the land surfaceand boundary layer models in two operational versions of the NCEP Eta modelusing FIFE data. Mon. Weather Rev. 125, 2896–2916.

ircher, S., Skou, N., Jensen, K.H., Walker, J.P., Rasmussen, L., 2011. A soil moistureand temperature network for SMOS validation in Western Denmark. Hydrol.Earth Syst. Sci. 8, 9961–10006, http://dx.doi.org/10.5194/hessd-8-9961.

onan, G.B., 1998. The land surface climatology of the NCAR land surface modelcoupled to the NCAR community climate model. J. Clim. 11, 1307–1326.

osilovich, M.G., Lawford, R., 2002. Coordinated enhanced observing period (CEOP)international workshop. Bull. Am. Meteorol. Soc. 83, 1495–1499.

hen, D., Kling, J., 1996. Apparent thermal diffusivity in soil: estimation fromthermal records and suggestions for numerical modeling. Phys. Geogr. 17,419–430.

teorology 218–219 (2016) 288–297

Chen, F., Mitchell, K., Schaake, J., Xue, Y., Pan, H.L., Koren, V., Duan, Q.Y., EK, M.,Betts, A., 1996. Modeling of land surface evaporation by four schemes andcomparison with FIFE observations. J. Geophys. Res. 101, 7251–7268.

Chen, Y.Y., Yang, K., He, J., Qin, J., Shi, J., Du, J., He, Q., 2011. Improving land surfacetemperature modeling for dry land of China. J. Geophys. Res. 116 (D20104,http://dx.doi.org/10.1029/2011JD015921).

Chen, Y.Y., Yang, K., Tang, W.J., Qin, J., Zhao, L., 2012. Parameterizing soil organiccarbon’s impacts on soil porosity and thermal parameters for Eastern Tibetgrasslands. Sci. Chin. Earth Sci. 55, 1001–1011.

Cheviron, B., Guérin, R., Tabbagh, A., Bendjoudi, H., 2005. Determining long-termeffective groundwater recharge by analyzing vertical soil temperature profilesat meteorological stations. Water Resour. Res. 41, W09501.

Cox, S.J., Stackhouse Jr., P.W., Gupta, S.K., Mikovitz, J.C., Zhang, T., Hinkelman, L.M.,Wild, M., Ohmura, A., 2006. The NASA/GEWEX surface radiation budgetproject: overview and analysis. In: 12th Conference on Atmospheric Radiation,Madison, WI, USA, pp. 10–14.

Dai, Y.J., Zeng, X.B., Dickinson, R.E., Baker, I., Bonan, G.B., Bosilovich, M.G., Denning,A.S., Dirmeyer, P.A., Houser, P.R., Niu, G.Y., Oleson, K.W., Schlosser, C.A., Yang,Z.L., 2003. The Common Land Model (CLM). Bull. Am. Meteorol. Soc. 84,1013–1023.

De Lannoy Gabriëlle, J.M., Houser, P.R., Pauwels, V., Verhoest, N., 2006. Assessmentof model uncertainty for soil moisture through ensemble verification. J.Geophys. Res 110, http://dx.doi.org/10.1029/2005JD006367.

Dickinson, R.E., Henderson-Sellers, A., Kennedy, P.J., Wilson, M.F., 1986.Biosphere–atmosphere transfer scheme (BATS) for the NCAR CommunityClimate Model. In: NCAR Tech. Note NCAR/TN-275 + STR., pp. 69.

Ek, M.B., Mitchell, O.C.E., Lin, Y., Rogers, E., Grunmann, P., Koren, V., Gayno, G.,Tarpley, J.D., 2003. Implementation of Noah land surface model advances inthe National Centers for Environmental Prediction operational mesoscale Etamodel. J. Geophys. Res. 108, 8851.

Fan, X., 2009. Impacts of soil heating condition on precipitation simulations in theWeather Research and Forecasting model. Mon. Weather Rev. 137, 2263–2285.

Farouki, O.T., 1981. The thermal properties of soils in cold regions. Cold Reg. Sci.Technol. 5, 67–75.

Godfrey, C.M., Stensrud, D.J., 2008. Soil temperature and moisture errors inoperational Eta model analysis. J. Hydrometeorol. 9, 367–387.

Grant, S.A., Or, D., 2004. Relationship between temperature sensitivity of capillarypressure and soil particle size. Geophys. Res. Lett. 31, L07501.

Hogue, T.S., Bastidas, L., Gupta, H., Sorooshian, S., Mitchell, K., Emmerich, W., 2005.Evaluation and transferability of the Noah model in semiarid environments. J.Hydrometeorol. 6, 68–84.

Johansen, O., 1975. Thermal Conductivity of Soils. University of Trondheim,Trondheim (Ph.D. Thesis).

Kaihotsu, I., Yamanaka, T., Oyunbaatar, D., Ganbold, T., Davaa, G., Koike, T., 2003.Long-term monitoring by water cycle stations in the central part of theMongolian Plateau. In: Proceedings of 1st International APHW (Asia PacificAssociation of Hydrology and Water Resources) Conference, Kyoto, Japan, pp.117–120.

Kirschbaum, M.U.F., 2000. Will changes in soil organic carbon act as a positive ornegative feedback on global warming. Biogeochemistry 48, 21–51.

Koike, T., 2004. The coordinated enhanced observing period—an initial step forintegrated global water cycle observation. WMO Bull. 53, 1–8.

Koren, V., Schaake, J., Mitchell, K., Duan, Q.Y., Chen, F., Baker, J.M., 1999. Aparameterization of snowpack and frozen ground intended for NCEP weatherand climate models. J. Geophys. Res. 104, 19569–19585.

Koster, R.D., Suarez, M.J., 1992a. A comparative analysis of two land surfaceheterogeneity representations. J. Clim. 5, 1379–1390.

Koster, R.D., Suarez, M.J., 1992b. Modeling the land surface boundary in climatemodels as a composite of independent vegetation stands. J. Geophys. Res. 97,2697–2715.

Lawford, R., Bosilovich, M., Eden, S., Benedict, S., Brown, C., Gruber, A., Houser, P.,Hsu, K., Huang, J., Lau, W., Meyers, T., Mitchell, K., Peters-lidard, C., Roads, J.,Rodell, M., Sorooshian, S., Tarpley, D., Williams, S.F., 2006. U.S. contributions tothe CEOP. Bull. Am. Meteorol. Soc. 87, 927–939.

LeMone, M.A., Tewari, M., Chen, F., Alfieri, J., Niyogi, D., 2008. Evaluation of theNoah land-surface model using data from a fair-weather IHOP 200 day withheterogeneous surface fluxes. Mon. Weather Rev. 136,4915–4941.

Li, Z., Tang, B., Wu, H., Ren, H., Yan, G., Wan, Z., Trigo, I.F., Sobrino, J.A., 2013.Satellite-derived land surface temperature: current status and perspectives.Remote Sens. Environ. 131, 14–37.

Liang, X., Lettenmaier, D.P., Wood, E.F., Burges, S.J., 1994. A simple hydrologicallybased model of land surface water and energy fluxes for GSMs. J. Geophys. Res.99, 14415–14428.

Lloyd, J., Taylor, J.A., 1994. On the temperature dependence of soil respiration.Funct. Ecol. 8, 315–323.

Mahanama, S.P.P., Koster, R.D., Reichle, R.H., Suarez, M.J., 2008. Impact ofsubsurface temperature variability on surface air temperature variability: anAGCM study. J. Hydrometeor. 9, 804–815.

McMichael, B.L., Burke, J.J., 1998. Soil temperature and root growth. HortScience33, 947–951.

Niu, G.Y., Yang, Z.L., 2006. Effects of frozen soil on snowmelt runoff and soil waterstorage at a continental scale. J. Hydrometeor. 7, 937–952.

Or, D., Wraith, J.M., 1999. Temperature effects on soil bulk dielectric permittivitymeasured by time domain reflectometry: a physical model. Water Resour. Res.35, 371–383.

http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0005http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0005http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0005http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0005http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0005http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0005http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0005http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0005http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0005http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0005http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0005http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0005http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0005http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0005http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0005http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0005http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0005http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0005http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0005http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0005http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0005http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0005http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0005http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0005http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0005http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0005http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0005http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0005http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0005http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0005http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0010http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0010http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0010http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0010http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0010http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0010http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0010http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0010http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0010http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0010http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0010http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0010http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0010http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0010http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0010http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0010http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0010http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0010http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0010http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0010http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0010http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0010http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0010http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0010http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0015http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0015http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0015http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0015http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0015http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0015http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0015http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0015http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0015http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0015http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0015http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0015http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0015http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0015http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0015http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0015http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0015http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0015http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0015http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0015http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0015http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0015http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0015http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0015http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0015http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0015http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0015http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0015dx.doi.org/10.5194/hessd-8-9961dx.doi.org/10.5194/hessd-8-9961dx.doi.org/10.5194/hessd-8-9961dx.doi.org/10.5194/hessd-8-9961dx.doi.org/10.5194/hessd-8-9961dx.doi.org/10.5194/hessd-8-9961dx.doi.org/10.5194/hessd-8-9961dx.doi.org/10.5194/hessd-8-9961dx.doi.org/10.5194/hessd-8-9961http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0025http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0025http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0025http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0025http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0025http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0025http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0025http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0025http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0025http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0025http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0025http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0025http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0025http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0025http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0025http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0025http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0025http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0025http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0025http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0025http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0025http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0025http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0025http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0030http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0030http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0030http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0030http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0030http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0030http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0030http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0030http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0030http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0030http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0030http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0030http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0030http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0030http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0030http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0035http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0035http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0035http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0035http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0035http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0035http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0035http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0035http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0035http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0035http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0035http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0035http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0035http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0035http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0035http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0035http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0035http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0035http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0035http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0035http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0040http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0040http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0040http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0040http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0040http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0040http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0040http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0040http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0040http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0040http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0040http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0040http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0040http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0040http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0040http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0040http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0040http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0040http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0040http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0040http://dx.doi.org/10.1029/2011JD015921http://dx.doi.org/10.1029/2011JD015921http://dx.doi.org/10.1029/2011JD015921http://dx.doi.org/10.1029/2011JD015921http://dx.doi.org/10.1029/2011JD015921http://dx.doi.org/10.1029/2011JD015921http://dx.doi.org/10.1029/2011JD015921http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0050http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0050http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0050http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0050http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0050http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0050http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0050http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0050http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0050http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0050http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0050http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0050http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0050http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0050http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0050http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0050http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0050http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0050http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0050http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0050http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0050http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0050http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0050http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0055http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0055http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0055http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0055http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0055http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0055http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0055http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0055http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0055http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0055http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0055http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0055http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0055http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0055http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0055http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0055http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0055http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0055http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0055http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0060http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0060http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0060http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0060http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0060http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0060http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0060http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0060http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0060http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0060http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0060http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0060http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0060http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0060http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0060http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0060http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0060http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0060http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0060http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0060http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0060http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0060http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0065http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0065http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0065http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0065http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0065http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0065http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0065http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0065http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0065http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0065http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0065http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0065http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0065dx.doi.org/10.1029/2005JD006367dx.doi.org/10.1029/2005JD006367dx.doi.org/10.1029/2005JD006367dx.doi.org/10.1029/2005JD006367dx.doi.org/10.1029/2005JD006367dx.doi.org/10.1029/2005JD006367dx.doi.org/10.1029/2005JD006367http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0075http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0075http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0075http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0075http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0075http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0075http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0075http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0075http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0075http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0075http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0075http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0075http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0075http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0075http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0075http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0075http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0075http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0075http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0075http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0075http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0075http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0080http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0080http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0080http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0080http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0080http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0080http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0080http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0080http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0080http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0080http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0080http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0080http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0080http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0080http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0080http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0080http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0080http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0080http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0080http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0080http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0080http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0080http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0080http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0085http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0085http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0085http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0085http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0085http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0085http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0085http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0085http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0085http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0085http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0085http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0085http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0085http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0085http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0085http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0085http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0085http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0085http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0085http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0085http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0085http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0085http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0090http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0090http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0090http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0090http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0090http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0090http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0090http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0090http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0090http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0090http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0090http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0090http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0090http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0090http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0090http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0090http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0095http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0095http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0095http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0095http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0095http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0095http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0095http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0095http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0095http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0095http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0095http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0095http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0095http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0095http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0095http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0095http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0100http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0100http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0100http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0100http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0100http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0100http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0100http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0100http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0100http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0100http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0100http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0100http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0100http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0100http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0100http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0100http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0105http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0105http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0105http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0105http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0105http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0105http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0105http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0105http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0105http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0105http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0105http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0105http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0105http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0105http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0105http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0105http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0110http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0110http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0110http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0110http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0110http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0110http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0110http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0110http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0110http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0110http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0115http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0120http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0120http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0120http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0120http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0120http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0120http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0120http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0120http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0120http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0120http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0120http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0120http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0120http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0120http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0120http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0120http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0120http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0120http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0120http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0120http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0120http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0125http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0125http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0125http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0125http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0125http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0125http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0125http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0125http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0125http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0125http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0125http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0125http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0125http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0125http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0125http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0125http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0125http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0125http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0125http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0125http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0125http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0130http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0130http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0130http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0130http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0130http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0130http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0130http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0130http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0130http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0130http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0130http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0130http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0130http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0130http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0130http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0130http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0130http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0130http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0130http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0130http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0130http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0135http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0135http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0135http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0135http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0135http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0135http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0135http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0135http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0135http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0135http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0135http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0135http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0135http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0135http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0135http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0140http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0140http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0140http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0140http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0140http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0140http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0140http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0140http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0140http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0140http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0140http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0140http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0140http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0140http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0140http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0140http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0140http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0140http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0140http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0140http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0140http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0140http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0145http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0145http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0145http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0145http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0145http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0145http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0145http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0145http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0145http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0145http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0145http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0145http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0145http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0150http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0150http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0150http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0150http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0150http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0150http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0150http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0150http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0150http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0150http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0150http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0150http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0150http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0150http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0150http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0150http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0150http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0150http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0150http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0150http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0150http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0150http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0150http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0150http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0150http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0155http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0155http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0155http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0155http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0155http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0155http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0155http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0155http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0155http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0155http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0155http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0155http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0155http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0155http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0155http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0160http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0160http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0160http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0160http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0160http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0160http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0160http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0160http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0160http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0160http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0160http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0160http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0160http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0160http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0160http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0160http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0160http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0160http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0160http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0160http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0160http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0165http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0165http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0165http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0165http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0165http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0165http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0165http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0165http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0165http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0165http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0165http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0165http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0165http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0170http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0170http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0170http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0170http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0170http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0170http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0170http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0170http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0170http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0170http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0170http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0170http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0170http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0170http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0170http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0170http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0170http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0170http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0170http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0175http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0175http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0175http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0175http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0175http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0175http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0175http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0175http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0175http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0175http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0180http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0180http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0180http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0180http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0180http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0180http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0180http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0180http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0180http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0180http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0180http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0180http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0180http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0180http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0180http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0180http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0180http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0180http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0180http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0180http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0180http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0185http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0185http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0185http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0185http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0185http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0185http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0185http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0185http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0185http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0185http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0185http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0185http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0185http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0185http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0185http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0185http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0185http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0185http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0185http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0185http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0185http://refhub.elsevier.com/S0168-1923(16)30005-3/sbref0185

st Me

P

R

R

R

R

R

R

R

S

S

S

S

L. Wang et al. / Agricultural and Fore

eterjohn, W.T., Melillo, J.M., Steudler, P.A., Bowles, F.P., 1994. Responses of tracegas fluxes and N availability to experimentally elevated soil temperatures.Ecol. Appl. 4, 617–625