+ All Categories
Home > Documents > Dynamical Downscaling Using Satellite-Gauge Based Precipitation...

Dynamical Downscaling Using Satellite-Gauge Based Precipitation...

Date post: 16-Jul-2020
Category:
Upload: others
View: 1 times
Download: 0 times
Share this document with a friend
1
Dynamical Downscaling Using Satellite Dynamical Downscaling Using Satellite- Dynamical Downscaling Using Satellite Dynamical Downscaling Using Satellite Ana Nunes 1 and Ana Nunes 1 and 1 Experimental Climate Prediction Center Scripps Inst Experimental Climate Prediction Center, Scripps Inst San Diego, La Jo San Diego, La Jo 2 E th S t Si It di i li 2 Earth System Science Interdisciplina College Park College Park, 1 S mmar 1. Summary Recent studies have shown that the use of a regional model to downscale the Recent studies have shown that the use of a regional model to downscale the large-scale analyses marginally improves simulated precipitation fields. Using large scale analyses marginally improves simulated precipitation fields. Using t llit b d d t i t i l t l dl id t satellite-based products as input, a regional spectral model carried out an extended summertime simulation over South America The authors seek to extended summertime simulation over South America. The authors seek to recover the precipitation patterns during January of 2004 when daily recover the precipitation patterns during January of 2004, when daily precipitation analyses are available from a high-resolution, satellite-gauge based precipitation analyses are available from a high resolution, satellite gauge based l i th ti tl S th A i I thi td i it ti analysis over the continental South America. In this study, precipitation assimilation is only effectuated in the same time scale as the rainfall analysis assimilation is only effectuated in the same time scale as the rainfall analysis The regional model solutions using a combined satellite-gauge scheme are The regional model solutions using a combined satellite gauge scheme are encouraging, especially in comparison to the global reanalyses. Due to the encouraging, especially in comparison to the global reanalyses. Due to the t ti l i t l d f i bl i t i thl t potential impact on land surface variables, improvements in monthly to seasona predictions are expected However daily precipitation assimilation alone predictions are expected. However, daily precipitation assimilation alone removes the model-predicted precipitation diurnal cycle which negatively removes the model predicted precipitation diurnal cycle, which negatively i h d il f li d T l hi bl 3 impacts the daily near-surface temperature amplitudes. To solve this problem, 3 ho rl precipitation estimates ill then be pro ided to the model assimilation hourly precipitation estimates will then be provided to the model assimilation scheme scheme. 4 Experiment Design 4. Experiment Design T ECPC RSM i t f d S th A i f J Two ECPC-RSM experiments were performed over South America for Januar 2004 The first experiment consisted of a continuous assimilation of the dail 2004. The first experiment consisted of a continuous assimilation of the dail satellite-gauge based precipitation estimates (3B42Corr) A control simulation satellite gauge based precipitation estimates (3B42Corr). A control simulation ith t i it ti i il ti l f d H ft th i without precipitation assimilation - was also performed. Hereafter, these regiona experiments will be referred as PA and CTL in order to designate the experiment experiments will be referred as PA and CTL in order to designate the experiment with and without precipitation assimilation respectively The two ECPC-RSM with and without precipitation assimilation, respectively . The two ECPC RSM ltd i it ti fi ld t th ith R2 3B42C 1DD GPCP (1DD accumulated precipitation fields, together with R2, 3B42Corr, 1DD-GPCP (1DD GPCP; Huffman et al 2001) and the Brazilian raingauge are displayed below GPCP; Huffman et al. 2001), and the Brazilian raingauge are displayed below . -Gauge Based Precipitation Analyses Gauge Based Precipitation Analyses Gauge Based Precipitation Analyses Gauge Based Precipitation Analyses Daniel Vila 2 Daniel Vila 2 titution of Oceanography University of California titution of Oceanography, University of California, olla, CA, USA olla, CA, USA C t Ui it fM l d ry Center, University of Maryland, MD USA MD, USA 2 Precipitation Assimilation Scheme 2. Precipitation Assimilation Scheme e N d R d (2007 ) d l d i it ti i il ti e Nunes and Roads (2007a) developed a precipitation assimilation g scheme (PA) that modifies the water vapour vertical structure in a g scheme (PA) that modifies the water vapour vertical structure in a n regional climate model that is coupled to a land-surface model This o regional climate model that is coupled to a land-surface model. This o precipitation assimilation methodology is similar to the one- y precipitation assimilation methodology is similar to the one di i l i ti l (1D V ) t hi ( Mf f t l 2005) y dimensional variational (1D-Var) technique (e.g. Mafouf et al. 2005), d except that PA allows the specification of a perturbation profile rather d except that PA allows the specification of a perturbation profile rather n than the 1D-Var which finds the minimal perturbation structure A s than the 1D-Var, which finds the minimal perturbation structure. A s. linear perturbation profile - based on the differences between e linear perturbation profile based on the differences between “b d” d dl di t d i it ti i th dd d t th e “observed” and model-predicted precipitation - is then added to the e moisture profile in order to bring the model's precipitation closer to the e l moisture profile in order to bring the model s precipitation closer to the al observed values PA not only improves the surface hydrology but also e observed values. PA not only improves the surface hydrology but also e atmospheric characteristics (Nunes and Roads 2007b). y atmospheric characteristics (Nunes and Roads 2007b). y - n 3. The Regional Climate Model n 3. The Regional Climate Model The Scripps ECPC RSM used for these experiments is a version of The Scripps ECPC RSM used for these experiments is a version of the NCEP RSM developed by Juang and Kanamitsu (1994) with a 40- the NCEP RSM developed by Juang and Kanamitsu (1994), with a 40- km resolution and 28 vertical levels. The RSM is a primitive equation km resolution and 28 vertical levels. The RSM is a primitive equation model with similar physics as the driving NCEP DOE reanalysis II model, with similar physics as the driving NCEP-DOE reanalysis II ry (R2) Global Spectral Model In this study the RSM used the Relaxed ly (R2) Global Spectral Model. In this study, the RSM used the Relaxed ly Arakawa-Schubert cumulus convection scheme (RAS; Moorthi and - Arakawa Schubert cumulus convection scheme (RAS; Moorthi and S 1992) A M t j ti d f th i l id l Suarez 1992). A Mercator projection was used for the regional grid. al ts ts 5. Discussion M 5. Discussion M D From the left panels it can be seen that the PA monthly accumulated D- From the left panels, it can be seen that the PA monthly accumulated precipitation was improved in comparison to CTL and R2 although precipitation was improved in comparison to CTL and R2, although iti bi dt td i th PA i it ti fi ld ith positive bias was detected in the PA precipitation fields, with increased values from the assimilated precipitation rates The increased values from the assimilated precipitation rates. The assimilated precipitation estimate shows more detailed and realistic assimilated precipitation estimate shows more detailed and realistic t t i i t 1DD GPCP t ki it t th structures in comparison to 1DD-GPCP, taking into account the raingauge values However time series of maximum and minimum raingauge values. However, time series of maximum and minimum temperatures (TMAX and TMIN respectively) over some selected temperatures (TMAX and TMIN, respectively) over some selected it tl h th t th i bl t i d b points, apparently, show that these variables were not improved by the assimilation scheme in comparison to the GTS observations the assimilation scheme, in comparison to the GTS observations (solid black curves) and the PA CTL and R2 respective differences (solid black curves), and the PA, CTL and R2 respective differences. Thi i ht b ltd t th l k f dl i it ti di l l This might be related to the lack of model precipitation diurnal cycle due to the assimilation scheme Higher temporal resolution rain due to the assimilation scheme. Higher temporal resolution rain rates would be necessary to improve those variables as well rates would be necessary to improve those variables as well. P ii i J 2004 Precipitation: Jan 2004 Precipitation: Jan 2004 S th A i South America Li l ti Linear correlation 1.0 coefficients corroborate that 09 coefficients corroborate that th PA ttl i it ti 0.9 the PA total precipitation 0.8 patterns are closer to PA ent patterns are closer to C GC 0.7 CTL cie 3B42Corr and 1DD-GPCP 06 CTL R2 effic than CTL and R2 and that 0.6 R2 3B42 Coe than CTL and R2, and that 0.5 3B42 n C 3B42Corr also correlates tion well with 1DD GPCP 0.4 elat well with 1DD-GPCP . 03 orre 0.3 Co 0.2 0.2 0.1 00 3B42 GPCP 0.0 3B42 GPCP 6 Selected References 6. Selected References Nunes A M B and J Roads 2007a: Influence of precipitation Nunes, A. M. B., and J. Roads, 2007a: Influence of precipitation assimilation on a regional climate model's surface water and energy assimilation on a regional climate model s surface water and energy bd t JHd t 8 642 664 budgets. J. Hydrometeor., 8, 642-664. Nunes A M B and J Roads 2007b: Dynamical influences of Nunes, A. M. B., and J. Roads, 2007b: Dynamical influences of i it ti i il ti i ld li G h R precipitation assimilation on regional downscaling. Geophys. Res. Lett 34 L16817 doi:10 1029/2007GL030247 Lett., 34, L16817, doi:10.1029/2007GL030247. Vil D LGd G l D T ll dJ R 2008 Vila D., L.G. de Goncalves, D. Toll and J. Rozante, 2008, Vila D., L.G. de Goncalves, D. Toll and J. Rozante, 2008, St ti ti lE l ti fC bi dD il G Ob ti d Statistical Evaluation of Combined Daily Gauge Observations and Rainfall Satellite Estimations over Continental South America Rainfall Satellite Estimations over Continental South America, Accepted for publication J Hidrometeor Accepted for publication J. Hidrometeor.
Transcript
Page 1: Dynamical Downscaling Using Satellite-Gauge Based Precipitation …ipwg/meetings/beijing-2008/pres/Poster... · 2016-06-07 · Dynamical Downscaling Using Satellite-Ana Nunes1 and

Dynamical Downscaling Using SatelliteDynamical Downscaling Using Satellite--Dynamical Downscaling Using SatelliteDynamical Downscaling Using Satellite

Ana Nunes1 andAna Nunes1 and1Experimental Climate Prediction Center Scripps InstExperimental Climate Prediction Center, Scripps Inst

San Diego, La JoSan Diego, La Jo2E th S t S i I t di i li2Earth System Science Interdisciplinay p

College ParkCollege Park,

1 S mmar1. SummaryyRecent studies have shown that the use of a regional model to downscale theRecent studies have shown that the use of a regional model to downscale thelarge-scale analyses marginally improves simulated precipitation fields. Usinglarge scale analyses marginally improves simulated precipitation fields. Using

t llit b d d t i t i l t l d l i d tsatellite-based products as input, a regional spectral model carried out anp p g pextended summertime simulation over South America The authors seek toextended summertime simulation over South America. The authors seek torecover the precipitation patterns during January of 2004 when dailyrecover the precipitation patterns during January of 2004, when dailyprecipitation analyses are available from a high-resolution, satellite-gauge basedprecipitation analyses are available from a high resolution, satellite gauge based

l i th ti t l S th A i I thi t d i it tianalysis over the continental South America. In this study, precipitationy y p passimilation is only effectuated in the same time scale as the rainfall analysisassimilation is only effectuated in the same time scale as the rainfall analysisThe regional model solutions using a combined satellite-gauge scheme areThe regional model solutions using a combined satellite gauge scheme areencouraging, especially in comparison to the global reanalyses. Due to theencouraging, especially in comparison to the global reanalyses. Due to the

t ti l i t l d f i bl i t i thl tpotential impact on land surface variables, improvements in monthly to seasonap p p ypredictions are expected However daily precipitation assimilation alonepredictions are expected. However, daily precipitation assimilation aloneremoves the model-predicted precipitation diurnal cycle which negativelyremoves the model predicted precipitation diurnal cycle, which negativelyi h d il f li d T l hi bl 3impacts the daily near-surface temperature amplitudes. To solve this problem, 3pacts t e da y ea su ace te pe atu e a p tudes o so e t s p ob e , 3ho rl precipitation estimates ill then be pro ided to the model assimilationhourly precipitation estimates will then be provided to the model assimilationy p p pschemescheme.

4 Experiment Design4. Experiment DesignT ECPC RSM i t f d S th A i f JTwo ECPC-RSM experiments were performed over South America for Januarp p2004 The first experiment consisted of a continuous assimilation of the dail2004. The first experiment consisted of a continuous assimilation of the dailsatellite-gauge based precipitation estimates (3B42Corr) A control simulationsatellite gauge based precipitation estimates (3B42Corr). A control simulation

ith t i it ti i il ti l f d H ft th iwithout precipitation assimilation - was also performed. Hereafter, these regionap p p , gexperiments will be referred as PA and CTL in order to designate the experimentexperiments will be referred as PA and CTL in order to designate the experimentwith and without precipitation assimilation respectively The two ECPC-RSMwith and without precipitation assimilation, respectively. The two ECPC RSM

l t d i it ti fi ld t th ith R2 3B42C 1DD GPCP (1DDaccumulated precipitation fields, together with R2, 3B42Corr, 1DD-GPCP (1DDp p , g , , (GPCP; Huffman et al 2001) and the Brazilian raingauge are displayed belowGPCP; Huffman et al. 2001), and the Brazilian raingauge are displayed below.

--Gauge Based Precipitation AnalysesGauge Based Precipitation AnalysesGauge Based Precipitation AnalysesGauge Based Precipitation Analyses

Daniel Vila2 Daniel Vila2

titution of Oceanography University of Californiatitution of Oceanography, University of California, olla, CA, USAolla, CA, USA

C t U i it f M l dry Center, University of Maryland,y y yMD USAMD, USA

2 Precipitation Assimilation Scheme2. Precipitation Assimilation Schemee N d R d (2007 ) d l d i it ti i il tie Nunes and Roads (2007a) developed a precipitation assimilationg

( ) p p pscheme (PA) that modifies the water vapour vertical structure in ag scheme (PA) that modifies the water vapour vertical structure in a

n regional climate model that is coupled to a land-surface model Thiso

regional climate model that is coupled to a land-surface model. Thiso precipitation assimilation methodology is similar to the one-y

precipitation assimilation methodology is similar to the onedi i l i ti l (1D V ) t h i ( M f f t l 2005)y dimensional variational (1D-Var) technique (e.g. Mafouf et al. 2005),

d( ) q ( g ),

except that PA allows the specification of a perturbation profile ratherd except that PA allows the specification of a perturbation profile rathern than the 1D-Var which finds the minimal perturbation structure As

than the 1D-Var, which finds the minimal perturbation structure. As. linear perturbation profile - based on the differences betweene

linear perturbation profile based on the differences between“ b d” d d l di t d i it ti i th dd d t the “observed” and model-predicted precipitation - is then added to the

ep p p

moisture profile in order to bring the model's precipitation closer to theel

moisture profile in order to bring the model s precipitation closer to theal observed values PA not only improves the surface hydrology but alsoe

observed values. PA not only improves the surface hydrology but alsoe atmospheric characteristics (Nunes and Roads 2007b).y

atmospheric characteristics (Nunes and Roads 2007b).y-n 3. The Regional Climate Modeln 3. The Regional Climate Model

The Scripps ECPC RSM used for these experiments is a version ofThe Scripps ECPC RSM used for these experiments is a version ofthe NCEP RSM developed by Juang and Kanamitsu (1994) with a 40-the NCEP RSM developed by Juang and Kanamitsu (1994), with a 40-km resolution and 28 vertical levels. The RSM is a primitive equationkm resolution and 28 vertical levels. The RSM is a primitive equationmodel with similar physics as the driving NCEP DOE reanalysis IImodel, with similar physics as the driving NCEP-DOE reanalysis IIry y g y(R2) Global Spectral Model In this study the RSM used the Relaxed

yly (R2) Global Spectral Model. In this study, the RSM used the Relaxedly

Arakawa-Schubert cumulus convection scheme (RAS; Moorthi and- Arakawa Schubert cumulus convection scheme (RAS; Moorthi andS 1992) A M t j ti d f th i l idl Suarez 1992). A Mercator projection was used for the regional grid.al ) p j g g

tsts5. DiscussionM 5. DiscussionM

D From the left panels it can be seen that the PA monthly accumulatedD- From the left panels, it can be seen that the PA monthly accumulatedprecipitation was improved in comparison to CTL and R2 althoughprecipitation was improved in comparison to CTL and R2, although

iti bi d t t d i th PA i it ti fi ld ithpositive bias was detected in the PA precipitation fields, withp p p ,increased values from the assimilated precipitation rates Theincreased values from the assimilated precipitation rates. Theassimilated precipitation estimate shows more detailed and realisticassimilated precipitation estimate shows more detailed and realistict t i i t 1DD GPCP t ki i t t thstructures in comparison to 1DD-GPCP, taking into account thep , g

raingauge values However time series of maximum and minimumraingauge values. However, time series of maximum and minimumtemperatures (TMAX and TMIN respectively) over some selectedtemperatures (TMAX and TMIN, respectively) over some selected

i t tl h th t th i bl t i d bpoints, apparently, show that these variables were not improved byp , pp y, p ythe assimilation scheme in comparison to the GTS observationsthe assimilation scheme, in comparison to the GTS observations(solid black curves) and the PA CTL and R2 respective differences(solid black curves), and the PA, CTL and R2 respective differences.Thi i ht b l t d t th l k f d l i it ti di l lThis might be related to the lack of model precipitation diurnal cycleg p p ydue to the assimilation scheme Higher temporal resolution raindue to the assimilation scheme. Higher temporal resolution rainrates would be necessary to improve those variables as wellrates would be necessary to improve those variables as well.

P i i i J 2004Precipitation: Jan2004Precipitation: Jan 2004

S th A iSouth AmericaLi l tiLinear correlation

1.0 coefficients corroborate that09

coefficients corroborate thatth PA t t l i it ti0.9 the PA total precipitation

0.8p p

patterns are closer toPAen

t patterns are closer toC G C0.7 CTLcie 3B42Corr and 1DD-GPCP

06CTLR2ef

fic

than CTL and R2 and that0.6 R23B42Co

e than CTL and R2, and that0.5

3B42

n C 3B42Corr also correlates

tion 3 Co a so co e ates

well with 1DD GPCP0.4elat well with 1DD-GPCP.

03orre

0.3Co

0.20.2

0.1

003B42 GPCP

0.03B42 GPCP

6 Selected References6. Selected ReferencesNunes A M B and J Roads 2007a: Influence of precipitationNunes, A. M. B., and J. Roads, 2007a: Influence of precipitation assimilation on a regional climate model's surface water and energyassimilation on a regional climate model s surface water and energy b d t J H d t 8 642 664budgets. J. Hydrometeor., 8, 642-664.g y , ,

Nunes A M B and J Roads 2007b: Dynamical influences ofNunes, A. M. B., and J. Roads, 2007b: Dynamical influences of i it ti i il ti i l d li G h Rprecipitation assimilation on regional downscaling. Geophys. Res. p p g g p y

Lett 34 L16817 doi:10 1029/2007GL030247Lett., 34, L16817, doi:10.1029/2007GL030247. Vil D L G d G l D T ll d J R 2008Vila D., L.G. de Goncalves, D. Toll and J. Rozante, 2008,Vila D., L.G. de Goncalves, D. Toll and J. Rozante, 2008, St ti ti l E l ti f C bi d D il G Ob ti dStatistical Evaluation of Combined Daily Gauge Observations and y gRainfall Satellite Estimations over Continental South AmericaRainfall Satellite Estimations over Continental South America, Accepted for publication J HidrometeorAccepted for publication J. Hidrometeor.

Recommended