jra-‐55 based surface atmospheric data set for driving ocean-‐sea
TRANSCRIPT
OMDP extended mee,ng 14 January 2016
@JAMSTEC (Yokohama, Japan) 45min
JRA-‐55 based surface atmospheric data set for driving ocean-‐sea ice models
Hiroyuki Tsujino (JMA-‐MRI) Special thanks to numerous inputs received in the framework of
OMDP-‐JRA55 collabora,ve effort
Outline
1. Background
2. Dataset aWributes
3. Adjustment on surface state of JRA-‐55
4. Valida,on
5. Plan for version 1
1. Background
• Mo#va#on: -‐ Simula,on of recent climate extreme events (e.g., Arc,c sea ice reduc,on, GW hiatus, ongoing El Nino, …) using Ocean-‐Sea ice models, as an extension of a long-‐term simula,on, to understand them in the context of long-‐term variability
• Requirements to forcing datasets: -‐ Quality controlled long-‐term dataset, also made available near-‐real ,me -‐ Temporal and spa,al resolu,on as high as possible to facilitate high-‐ resolu,on simula,ons
• JRA55: -‐ JRA55 (Kobayashi et al. 2015) is one of the most recently conducted long-‐term reanalyses using high-‐resolu,on (~55 km) model and updated assimila,on techniques -‐ JRA55 s,ll needs some adjustments (bias correc,on) as done in CORE (to NCEP/NCAR reanalysis) and DRAKKAR (to ERA) projects.
Surface heat flux components globally averaged over the ice-‐free ocean
Green: CORE (Large and Yeager 2009) Red: JRA55 product (JRA55-‐v0.0)
Blue: JRA55-‐v0.1 -‐ 2m → 10m shie of air temperature and specific humidity using surface roughness of JRA55-‐v0.0 Purple: JRA55-‐v0.1b -‐ 2m → 10m shie of air temperature and specific humidity using Large and Yeager (2009) bulk formula
• In ocean-‐sea ice models, fluxes are diagnosed on the basis of the surface state, using another set of formulae (e.g., Large and Yeager 2009). The resultant fluxes may differ considerably.
• Fluxes are also sensi,ve to how surface state variables are shieed in height.
Project to produce the data set based on JRA55
• We started to consider producing a JRA-‐55 based data set for driving ocean-‐sea ice models in 2014
• The idea was presented in “Forcing mini-‐workshop” at Grenoble in Jan2015. Par,cipants were generally suppor,ve. OMDP-‐JRA55 collabora,on started • Adjustment method and features of interannual variability are examined
during 2015 so that the data set deserves wide use
• Japanese ac,vity is supported by a research grant from JSPS (fy2015-‐2018)
2. Dataset a6ributes
Elements Ø Downward short / long wave flux … average Ø Precipita,on (rain and snow) … average Ø 10 m vector wind … instantaneous Ø 10 m air temperature, specific humidity … instantaneous
(shieed from their original height at 2 m) Ø Sea level pressure … instantaneous
Suppor#ng data Ø Unadjusted fields of the above (named version 0.1) Ø Land-‐sea mask Ø Surface brightness temperature and Ice distribu,on Ø SST and sea ice distribu,on (COBESST; Ishii et al. 2005)
• All elements are based on “forecast mode” of JRA-‐55 • Period: 1958-‐present (updated near-‐real ,me) • Interval: 3-‐hour • TL319 regular grid (~ 55km) (originally on “reduced” grid)
Surface atmospheric states
River run-‐off to the ocean
• Unavailable from JRA-‐55
• Our plan is to run a river model opera,onally using water runoff from the land surface part of JRA-‐55 → Current status will be presented by T. Suzuki
• A global river model CaMaFlood (Yamazaki et al. 2011) is used → Details of the river model will be presented by D. Yamazaki
Versions of JRA-‐55 based surface dataset
• Version 0.0: JRA-‐55 Product (when this name is useful) • Version 0.1: Unadjusted JRA-‐55 on regular TL319 grid -‐ Zonally interpolated from the (original) reduced TL319 grid -‐ 2 m temp and humidity is shieed to 10 m using surface roughness of JRA-‐55 → 10 m values are adjusted for v0.2 • Version 0.2: Adjustment on version 0.1 (Mar 2015) • Version 0.3: Revised adjustment (Dec 2015) -‐ 2 m temp and humidity is adjusted on 2 m and then shieed to 10 m using LY09 formula -‐ Adjustment is done essen,ally on v0.0 • Version 0.4: Very low temperature is cut-‐off around Antarc,ca
Strategy
ü Evalua,on of global surface heat and fresh water flux budget, followed by re-‐adjustment for downward fluxes • Global surface heat flux budget (sea ice region excluded) is computed using
COBESST (Ishii et al. 2005) as a lower boundary condi,on and Large and Yeager (2009) formulae for bulk transfer coefficient and albedo
-‐ Proper,es of moist air taken from the textbook of Gill (1982) • Downward short/long wave are re-‐adjusted so that total heat flux ~ 0 for
1988-‐2007
• Any river run off data should be adjusted so that its long-‐term mean (1988-‐2007) is the same as CORE (1.22 Sv; Dai et al. 2009).
• Precipita,on is re-‐adjusted so that E-‐P-‐R ~ 0 for 1988-‐2007.
ü Follow the method of Large and Yeager (2009) used for CORE • Apply mul,plica,ve or offset factors to the surface state variables of JRA-‐55 • To produce forcing dataset near-‐real ,me, all elements (including radia,on
and precipita,on) are based on JRA55. We make adjustments to them. • Adjustment factors are climatological: we do not touch interannual varia,ons
3. Adjustment on the surface state of JRA-‐55
reference data
adj factor based on
Bme dependency
spaBal dependency
How is the factor used
short wave CORE 1984-‐2007 monthly (x,y) & constant
mul,ply
long wave CORE 1984-‐2007 monthly (x,y) & constant
mul,ply
precipita,on CORE 1979-‐2008 monthly (x,y) & constant
mul,ply
air temperature CORE 1979-‐2008 monthly (x,y) offset
specific humidity CORE 1979-‐2008 monthly (x,y) mul,ply
wind speed QuikSCAT nov1999-‐oct2009
constant (x,y) mul,ply
wind angle QuikSCAT nov1999-‐oct2009
constant (x,y) offset
Summary of the adjustment method for v0.2 (Mar 2015)
Blue: devia,on from CORE
reference data
adj factor based on
Bme dependency
spaBal dependency
How is the factor used
short wave adjusted CERES%
mar2000-‐feb2015
monthly (x,y) & constant
mul,ply
long wave adjusted CERES%
mar2000-‐feb2015
monthly (x,y) & constant
mul,ply
precipita,on CORE 1979-‐2008 monthly (x,y) & constant
mul,ply
air temperature JRA55-‐anl_surf# IABP-‐NPOLES
1979-‐1998 monthly (x,y) offset
specific humidity JRA55-‐anl_surf# 1979-‐1998 monthly (x,y) mul,ply
wind speed QuikSCAT* JRA55-‐anl_surf#
aug1999-‐oct2009
monthly (x,y) mul,ply
wind angle QuikSCAT* JRA55-‐anl_surf#
aug1999-‐oct2009
monthly (x,y) offset
Summary of the adjustment method for v0.3 (Dec 2015) (Aeer extensive discussions with collaborators)
Red: change from v0.2 (%) CERES-‐EBAFv2.8 Surface (Kato et al. 2013) (*) Remote Sensing Systems 0.25 x 0.25 data set version 4 (#) Screen level analysis of JRA55
3.1 Radia,on flux adjustment
• For v0.3, we use CERES-‐EBAFv2.8 Surface (Mar2000-‐Feb2015) (Kato et al. 2013) for downward radia,on adjustment
-‐ Both SW and LW radia,on should be compared with buoy measurements before they are used as a reference data as LY09 did for ISCCP-‐FD
-‐ Apply necessary adjustment to CERES
• For v0.2, we used CORE version 2 (Large and Yeager 2009). -‐ CORE version 2 radia,on is based on ISCCP-‐FD, with some adjustments applied by Large and Yeager (2009) -‐ In recent assessments, CERES is widely used as a reference data -‐ Data period of CERES has now reached 15 years
Reference data
Bias of CERES-‐EBAF 2.8 surface downward SW radia,on rela,ve to Buoys
• CERES has posi,ve bias in the tropics except for the Equator • CERES’s tropical posi,ve biases are larger in the NH than the SH • Biases are generally small in la,tudes higher than 30oN/S.
JRA55-‐v0.1 CORE CERES CERESadj
Zonal mean downward SW over the ocean
Max 4% reduc,on at 10oN
Max 3% reduc,on at 10oS
Reduce CERES SW radia,on as a func,on of la,tude (profile is given in the boWom slide: CERESadj / CERES)
CERES-‐EBAF 2.8 surface downward long wave
JRA-‐55-‐v0.1 CORE CERES
Zonal mean over the ocean
• CERES LW radia,on will not be adjusted: Used as a reference as it is
• Globally uniform ,me-‐invariant factor will be applied to both SW and LW to close surface heat flux budget in the 2nd adjustment step for JRA-‐55
bias CERES should be
reduced by…
1. Monthly mul,plica,ve factors are determined on the basis of 15yr (mar2000-‐feb2015) monthly climatology of adjusted CERES and JRA-‐55, and are applied for the en,re period
2. A globally uniform, ,me-‐invariant factor (=0.977933, i.e., 2.2% reduc,on) is applied to both SW and LW to achieve a global heat flux balance for the period 1988-‐2007
Adjustment steps for short and long wave radia,on of JRA-‐55
Mul,plica,ve factor for SW (annual mean of monthly factors) v0.2 (ref CORE)
v0.3 (ref CERESadj)
Mul,plica,ve factor for LW (annual mean of monthly factors)
3.2 Precipita,on adjustment
• Reference is CORE (1979-‐2008) (from south to north, GPCP, CMAP, GPCP, CMAP, and climatology around the Arc,c: so called GCGCS) • We did not find strong reason to replace CORE with another data set
(assessment made by J. Small@NCAR)
• Monthly factors (0.2 < f < 5.0) are computed and applied for the en,re period
• Evapora,on is computed aeer the first adjustment for all elements. Then a globally uniform, ,me-‐invariant factor (=0.969863, 3% percent reduc,on) is applied to close the freshwater budget, assuming that the global mean river run off equals 1.22 Sv for the period 1988-‐2007
Adjustment method for precipita,on is the same for v0.2 and v0.3
Mul,plica,ve factor for precipita,on (annual mean of monthly factors)
Green: CORE Red: JRA55v0.1 Blue: JRA55v0.3
Zonal mean precipita,on over the ocean
For version 0.3, we make use of surface analysis of JRA55 (anl_surf) to adjust surface atmospheric state variables (air temperature, specific humidity, vector wind)
About surface analysis of JRA55
From Sec#on 3.2.a of Kobayashi et al. (2015): -‐ Analysis of screen-‐level variables (2 m temperature, 2 m rela,ve
humidity, and 10 m winds) is performed separately from the atmospheric analysis component, using 2-‐D OI
-‐ Observa,on on islands and coast are NOT used -‐ Screen-‐level analysis fields are NOT used as ini,al condi,ons for
forecasts
3.3 Surface atmospheric state adjustment
• IABP-‐NPOLES (Jan1979-‐Dec1998) and JRA-‐55 anl_surf (surface analysis) for 2 m temperature
-‐ Mainly to avoid ripples of adjustment factors that come from CORE -‐ IABP-‐NPOLES are used to adjust temperature on sea ice in the Arc,c Ocean,
JRA55-‐anl_surf is used elsewhere
• JRA-‐55 anl_surf for 2m specific humidity
3.3.1 Air temperature and specific humidity Reference data v0.2
v0.3
• CORE for both 10 m air temperature and 10 m specific humidity -‐ 2m data from original JRA-‐55 is shieed to 10 m using surface roughness output of JRA-‐55 → v0.1 -‐ CORE and v0.1 is compared at 10 m to produce v0.2
Note: For v0.3, adjustments are done on JRA-‐55’s na,ve (reduced TL319) grid. Other reference data are also interpolated on JRA-‐55’s na,ve grid.
1. Monthly offse|ng factor for 2 m air temperature is determined on the basis of 20-‐yr (1979-‐1998; the period constrained by availability of IABP-‐NPOLES) monthly climatology
2. Monthly offse|ng factor is applied for the en,re dataset period. During this processing, rela,ve humidity is kept, i.e., 2 m specific humidity is also modified
3. Monthly mul,plica,ve factor for this intermediate 2 m specific humidity is determined using 20yr (1979-‐1998) monthly climatology of JRA55 surface analysis
4. Monthly mul,plica,ve factor is applied to 2 m specific humidity for the en,re data set period
5. Aeer the adjustment to all elements, 2 m temperature and specific humidity are shieed to 10 m using LY09 formula, meteorological variables, and SST (COBESST over the ocean, brightness temperature otherwise)
Adjustment steps for air temperature and specific humidity for v0.3
Offse|ng factor for air temperature (annual mean of monthly factors)
v0.2 (@10m): reference is CORE
v0.3 (@2m): reference is JRA55 anl_surf over sea water IABP-‐NPOLES over sea ice
Cut off of extremely low temperature around Antarc,ca for version 0.4
• Because atmospheric model of JRA-‐55 does not allow par,al sea ice cover, separate adjustment should be considered for air temperature over sea ice. This is done for the Arc,c by using IABP-‐NPOLES.
• For the Antarc,c region, we followed the adjustment employed by Large and Yeager (2004) for the CORE data set: Extremely low temperatures are cut off by using a sinusoidal fits to observed monthly minimum temperature as a func,on of la,tude south of 60oS.
Black: v0.3 Red: v0.4
Sensi#vity test will be presented by S. Urakawa later
v0.2: based on CORE v0.3: based on JRA55 anl_surf
“Error” of turbulent heat flux of v0.2 provided by P. Hyder (UKMetOffice)
(posi,ve into the ocean)
Mul,plica,ve factor for specific humidity a\er the temperature adjustment (annual mean of monthly factors)
Green: CORE, Red: JRA55v0.1, Grey: JRA55v0.2, Blue: JRA55v0.3
Zonal mean 10 m air temperature Zonal mean 10 m specific humidity
devia,on from NOCS
3.3.2 Vector Wind
• Remote sensing systems QuikSCAT v4 (Ricciardulli and Wentz, 2011) • Gaps of QuikSCAT are filled with JRA55-‐
anl_surf • QuikSCAT winds are adjusted in terms of
stability before they are blended with JRA-‐55 anl_surf vector wind
Reference data
Grey: Remote Sensing Systems QuikSCAT v4 (neutral 10m wind) Blue: Remote Sensing Systems QuikSCAT v4 (actual 10m wind) (atmospheric fields for stability adjustment taken from JRA55 surface analysis) Red: JRA55 surface analysis (actual 10m wind) Green: Blend of RSS-‐QuikSCAT v4 and JRA55 surface analysis (actual 10m wind) (this is used as a reference to adjust JRA55 forecast fields)
v0.2
v0.3
• SeaWinds on QuikSCAT Level 3 Daily Gridded Ocean Wind Vectors (JPL Version 2) • Gaps of QuikSCAT (due to rain, sea ice, and near the coast) are not filled. No
adjustment around gaps • QuikSCAT provides “equivalent neutral” 10m wind. This was used to adjust
“actual” 10 m wind of JRA-‐55, which was not appropriate.
In v0.3, first of all, zonal 1-‐2-‐1 filter is applied to unadjusted data before determining the adjustment factor
-‐ This is to remove grid noises in the lee of the Andes (become evident in wind stress curl)
Wind stress curl from a 10 km resolu,on OGCM (provided by H. Nakano)
In v0.3, full period of QuikSCAT (Aug1999-‐Oct2009) is used to make monthly adjustment factors. The monthly factors are applied for the en,re data set period. (Constant factor was used for v0.2.)
• Wind speed adjustment is based on monthly climatology • Wind angle adjustment is based on CEOF analysis on ,me series, using
10 or 11 data for each month. If the first mode (the co-‐varying mode) does not account for more than 95% of total variance (energy), adjustment angle is set to zero
• We need to check later whether the monthly factors work well or not
Rota,ng factor Mul,plica,ve factor(*) v0.2 (,me-‐invariant)
v0.3 (annual mean of monthly factors) (blend of QSCAT and JRA55-‐anl_surf)
(*) General reduc,on of wind speed factor in v0.3 is due to the shie of QuikSCAT wind to “actual wind” from “equivalent 10 m neutral wind”
4. ValidaBon
• Interannual variability of surface fluxes and relevant variables (1) There are less variability in JRA-‐55 heat fluxes than CORE Discussed later in this talk (2) Increase of JRA-‐55’s LHF aeer mid-‐1990s and later Discussed with S. Josey@NOC over Skype this evening
• Comparison with buoy observa,ons Presented by H. Tomita (next presenta#on)
• Is JRA55’s wind field in favor of coastal upwelling? Presented by J. Small (later today)
• Global heat / fresh water balance Next slide
Green: CORE, Red: JRA55v0.1 Grey: JRA55v0.2, Blue: JRA55v0.3
SW LW LAT SEN Total%
CORE 165.5 -‐53.3 -‐95.2* -‐14.4 2.6
LY09 165 -‐53 -‐96 -‐14 2
JRA55v0.0 171.5 -‐54.0 -‐112.0 -‐18.4 -‐12.8
JRA55v0.1 169.9 -‐53.7 -‐102.6 -‐17.6 -‐4.0
JRA55v0.2 164.7 -‐54.7 -‐93.7 -‐14.4 1.9
JRA55v0.3 164.4 -‐57.1 -‐92.0 -‐13.2 2.1
Global mean over 23 yrs (1984-‐2006)
Surface heat flux averaged over the ice-‐free ocean
(%) Except for the four main components, about 2 W/m2 will be lost owing to sea ice processes and temperature difference between precipita,on and evapora,on (*) Because of the use of temperature-‐dependent latent heat of vaporiza,on, the latent heat loss will be reduced compared to LY09
Green: CORE Red: JRA55v0.1 Grey: JRA55v0.2 Blue: JRA55v0.3
Evap Precip residual CORE -‐14.2 13.0 -‐1.2
LY09 (tab3) -‐14.0 12.8 -‐1.2
JRA55v0.0 -‐16.2 15.4 -‐0.8
JRA55v0.1 -‐15.3 15.4 -‐0.1
JRA55v0.2 -‐14.0 12.9 -‐1.1
JRA55v0.3 -‐13.7 12.6 -‐1.2
Global mean over 23 yrs (1984-‐2006)
Units: Sv = 109 kg s-‐1
Surface fresh water flux integrated over the ocean
Ques,ons on interannual variability: (1) There are less variability in JRA-‐55 heat fluxes than CORE
Green: CORE + COBESST (daily) Blue: JRA55v0.3 + COBESST (daily) Orange: JRA55v0.3 + Hurrell SST data set (monthly)
Sensible heat flux gets more variable when Hurrell SST is used
(but s,ll smaller than CORE)
-‐ Another SST data set is applied to JRA55v0.3 -‐
JRA55 air temperature tends to follow SST more closely than CORE
• Downward long wave is more variable for CORE (or ISCCP, green solid) • Downward and Upward long wave co-‐vary in JRA55 (blue), downward
long wave of JRA55 might be more sensi,ve to SST or surface air temperature
5. Plan for version 1 Two addi#onal changes are planned for v0.3 • We prefer version 0.4, where extremely low temperatures
around Antarc,ca are cut off. -‐ Details presented by S. Urakawa using MRI model later -‐ With this choice, global adjustment factors for radia,on and precipita,on must be revised • We will return to temporarily constant instead of
monthly adjustment factor for wind speed and direc,on -‐ Explained using following slides
Schedule • Data produc,on for the surface atmospheric states including
the above changes will be completed by mid-‐February 2016 • River run-‐off produc,on schedule TBD • Method of normal (repeat) year forcing TBD (Idea will be briefly presented by S. Yeager)
v0.3: Adjusted by monthly mean
factors
Adjusted by annual mean
factor (test version)
Risien and Chelton (2008) (Monthly climatology of
QuikSCAT)
Wind stress curl (nov1999-‐oct2009)
v0.3: Adjusted by monthly mean
factors
Adjusted by annual mean
factors
Sverdrup transport stream func,on (jan2000-‐dec2009)
Thank you for your aben#on!
Supplementary slides
Proposal of JRA-‐55 based surface atmospheric data set for driving Ocean-‐Sea ice models
• All surface atmospheric state variables needed to drive ocean-‐sea ice models are taken from JRA-‐55
• Pre-‐determined (climatological) adjustment factors are applied to the surface state variables
• River discharge to the ocean is computed by running a global river model using water runoff from land to river provided by JRA-‐55
• Data sets will be updated near-‐real ,me as long as JRA-‐55 con,nues
More on the background
• This idea was presented in “Forcing mini-‐workshop” at Grenoble in Jan2015. Par,cipants were suppor,ve. OMDP-‐JRA55 collabora,on started • Adjustment method and features of interannual variability are examined
during 2015 so that the data set deserves wide use
• Japanese ac,vity is supported by a research grant from JSPS (fy2015-‐2018)
Why do we not use JRA55 surface analysis?
-‐ By using forecast field for all elements, all fields are consistent one another (slightly contaminated by the adjustment)
-‐ JRA55 surface analysis is every 6 hours, while forecast fields are provided every 3 hours
-‐ In the difference between JRA55 forecast and analysis, point sources and their spa,al spreads are iden,fied. They are contained in the 2D-‐OI fields of JRA-‐55 surface analysis
• In version 0.3, full period of QuikSCAT (Aug1999-‐Oct2009) is used to make monthly adjustment factors. The monthly factors is applied for the en,re data set period. (Constant factors are computed for v0.2.) -‐ Wind speed adjustment is based on monthly climatology -‐ Wind angle adjustment is based on CEOF analysis on ,me series, using
10 or 11 data for each month. If the first mode does not account for more than 95% of total energy, adjustment angle is set to zero
-‐ We need to check whether the monthly factors work well or not later
𝑤↓𝐽 = 𝑢↓𝐽 +𝑖𝑣↓𝐽 𝑤↓𝑄 = 𝑢↓𝑄 +𝑖𝑣↓𝑄
𝑊=[█■𝑤↓𝐽 (Aug1999)&𝑤↓𝑄 (Aug1999)@⋮&⋮@𝑤↓𝐽 (Aug2009)&𝑤↓𝑄 (Aug2009) ]
𝑅= 𝑊↑∗ 𝑊, 𝑅:(2×2), 𝑅= 𝑅↑∗
Eigen vectors of R are CEOF modes: (1) co-‐varying (correlated) mode (2) mode orthogonal to co-‐varying mode
Correc,on angle = angle between the elements (vectors) of the co-‐varying mode
JRA55 wind
QuikSCAT wind
Merit: No special treatment is needed in low wind regions such as the transi,on between easterly and westerly
Covariance matrix
• Use of CEOF analysis to compute wind angle adjustment factor
Climate indices based on SLP green:CORE(NCEP/NCAR Reanalysis), red:JRA55-‐v0.1
AO
NPI
SOI
AAOI
Mul,plica,ve correc,on factor in per cent needed for the CERES downward SW to be adjusted toward buoys
Mul,plica,ve factor for specific humidity a\er the temperature adjustment (annual mean of monthly factors)
v0.2: based on CORE v0.3: based on JRA55 anl_surf
Comparison of zonal mean wind speed
Blue line: JRA55 v0.3 (adjusted) Red line: JRA55 v0.1 (unadjusted) (wind vector is interpolated on 1x1 la|ce “before” calcula,ng scalar wind speed) Green line: CORE Light blue open circles: Blend of RSS-‐QuikSCAT v4 and JRA55 surface analysis (this is used as a reference to adjust JRA55v0.1 toward JRA55v0.3)
Actual 10 m wind Neutral 10 m wind
Amplitude of interannual variability of wind is comparable between CORE
and JRA55
Amplitude of long-‐term varia,on of “dq” is greater for CORE than JRA55 (interannual variability is comparable)
JRA55 air temperature tends to follow SST more closely than CORE
Yu (2007) OAflux global average
Green: CORE, Blue: JRA55v0.3
Ques,ons on interannual variability (2) Reason for the increase of JRA-‐55’s LHF aeer mid-‐1990s and later
CORE, JRA55-‐v0.3, OAflux
Latent heat flux
Qsat(0m) minus Q(10m)
10 m wind speed
CORE, JRA55-‐v0.3, OAflux
Specific humidity
2 m specific humidity JRA55v0.3 minus OAflux
Dashed line: 2m
Solid line: 10m
VTPR TOVS AMSU-‐B SSM/I
AMSU-‐B SSM/I
Time-‐height cross sec,ons for the 12-‐month running mean of the global mean specific humidity increments (Kobayashi et al. 2015)
• Atmospheric model used by JRA-‐55 has posi,ve (moist) bias of surface specific humidity
• As the observa,ons become abundant, this moist bias is corrected globally (global reduc,on of surface moisture). This results in the larger latent heat loss in recent decades.
SSH from the 10 km MRI.COM simula,on (courtesy H. Nakano)
v0.1 and v0.2 are available from a server at LLNL (courtesy Paul Durack) -‐ JRA55_v0.1: <ep://gdo151.ucllnl.org/pub/JRA55_v0.1/> (Corrected downward short wave data is available from hWp://amaterasu.ees.hokudai.ac.jp/~tsujino/JRA55_v0.1.01/ courtesy Hokkaido Univ.) -‐ JRA55_v0.2: ep://gdo151.ucllnl.org/pub/JRA55_v0.2/
v0.3 and v0.4 are available from Hokkaido University (courtesy Hokkaido Univ.) -‐ JRA55_v0.3: hWp://amaterasu.ees.hokudai.ac.jp/~tsujino/JRA55_v0.3/ -‐ JRA55_v0.4: hWp://amaterasu.ees.hokudai.ac.jp/~tsujino/JRA55_v0.4/
Note: Official version 1.0 and 1.1 will be released soon (Above version will soon be obsolete). Or publica,on of the technical document should come first before the release?
Access to the data set