Embed Size (px)
Citation preview
Homogenized observations from Peru and reconstructed datasets: how well do they compare?
Martín Jacques-Coper1, Stefanie Gubler1, Mario Rohrer2, Clara Oria3, Mischa Croci-Maspoli1, Thomas Konzelmann1 1Swiss Federal Office of Meteorology and Climatology MeteoSwiss, Zurich, Switzerland 2Meteodat GmbH, Zurich, Technoparkstrasse 1, Switzerland 3Meteorological and Hydrological Service of Peru (SENAMHI), Lima, Peru
references [1] Compo G.P., Whitaker J.S., Sardeshmukh P.D., et al. (2011): The Twentieth Century Reanalysis Project, Q J R Meteorol Soc 137 (2011), 10.1002/qj.776 [2] Hofer M., Marzeion B., and Mölg T. (2012): Comparing the skill of different reanalyses and their ensembles as predictors for daily air temperature on a glaciated mountain (Peru). Clim Dyn, 39, 1969-1980, doi:10.1007/s00382-012-1501-2. [3] Kalnay E. et al. (1996): The NCEP/NCAR 40-year reanalysis project, Bull. Amer. Meteor. Soc., 77, 437-470, http://dx.doi.org/10.1175/1520-0477(1996)077%3C0437:TNYRP%3E2.0.CO;2 [4] Rienecker M.M. et al. (2011): MERRA: NASA's Modern-Era Retrospective Analysis for Research and Applications. J. Clim., 24, 3624-3648, doi:10.1175/JCLI-D-11-00015.1. [5] Uppala S.M. et al. (2005): The ERA-40 re-analysis. Q. J. R. Meteorol. Soc., 131: 2961–3012, doi: 10.1256/qj.04.176. [6] Poli P. et al. (2013): The data assimilation system and initial performance evaluation of the ECMWF pilot reanalysis of the 20th-century assimilating surface observations only (ERA-20C). ERA Report Series no. 14, ECMWF, 59 pp.
1.1/ a homogenized temperature dataset from climandes Homogenized temperature time series from Peruvian regions along the Central Andes, most of them covering the last 50 years, have been produced within the CLIMANDES project (www.senamhi.gob.pe/climandes/). Ongoing efforts aim to quantify the added value brought by the homogenization procedure and to place the homogenized station records in a broader climatological context. For that, we explore the way the quality-controlled-only (qc) and homogenized (hom) data compare with reanalyses.
1.2/ recuay: the global picture Recuay is located in the Ancash Department (Fig. 1a). Its meteorological station (09°44'S; 77°27'W, 3'444 masl) has been recording since 1964. The interannual variability of the local mean temperature (Tm) is highly correlated with the Tropical Pacific SST (inst. hom Tm [blue curve] vs ENSO3.4 [green curve]: R=0.67; Fig. 1b & 2). Using annual mean surface air temperature (σ0.995) from the 20 Century Reanalysis (20CR [1]), we compare the spatial patterns of the one-point correlation maps obtained from following time series: Fig. 2a) air temperature (σ0.995) from the closest grid point to Recuay (20CR); Fig. 2b) the quality-controlled-only station record (qc); Fig. 2c) the homogenized station record (hom).
3
2.1/ is there a similar skill of 20CR for every month? We calculate the correlation coefficient between the qc and hom time series and 20CR for every month separately (Fig. 4). An annual cycle of R is evident for Recuay.
Qualitatively, we find that the surface air temperature pattern based on 20CR only (ENSO-like spatial pattern depicted in Fig. 2a) is better reproduced using the hom instrumental record (Fig. 2c) than the qc one (Fig. 2b). A similar result arises from the comparison of the annual 1964-2007 Tm trends calculated from the qc and hom records (3 stations) with 20CR (Fig. 3).
1) interannual variability 2) annual cycle
2
2.2/ what about other reanalysis datasets? We also assess the correlation coefficients obtained from further reanalyses (Fig. 5). We confirm that, in general, their agreement with the Recuay time series is better for November-April (extended austral summer, rainy season) than for May-October (extended austral winter, dry season). This behaviour was also observed by a previous study [2]. Moreover, for most reanalyses, higher R values result from the hom (Fig. 5b) than from the qc (Fig. 5a) time series (see Fig. 5c). This result is particularly interesting in the case of 20CR and ERA-20C [6], for which no air temperature observations have been assimilated and are thus independent datasets. Lower values are found for the NCEP-NCAR Reanalysis and for ERA-40, which are based on [not necessarily homogeneous] temperature observations.
Fig. 4: Correlation coefficients calculated separately for every month between the 2m mean temperature time series extracted from the closest gridpoint to the Recuay station from 20CR and: (black) the raw (original) quality-controlled station record (qc); (dark violet) the homogenized station record (hom). The red curve indicates the R-value difference hom - qc. The red asterisk denotes statistical significance of the R-value difference at 85%. In this case, significance is reached for September due to the improvement of the R value.
2.3/ conclusions and outlook We have shown that there is potential for qualitatively and quantitatively estimating the added value of the homogenization of temperature time series from the Central Andes, based on the coherence of the spatial correlation pattern and statistical metrics (R values), respectively. Further steps will focus on more detailed climatological explanations for these findings (e.g. the annual cycle of R).
4
1
Tem
per
atu
re A
no
mal
y [°
C]
recuay
lircay
urubamba
recuay
lircay
urubamba
0.1
0.2
0.3
0.4
0.5
-0.5
-0.4
-0.3
-0.2
-0.1
0
[K/d
ecad
e]
a) a)
b)
c)
[R]
Fig. 2: One-point interannual correlation maps obtained from the mean surface Tm (σ0.995) of 20CR (ensemble mean) using annual means of: a) air temperature (σ0.995) from the closest grid point to Recuay (20CR) b) the raw (original) quality-controlled station record (qc) c) the homogenized station record (hom)
Fig. 3: 1964-2007 trend maps of surface air mean temperature (σ0.995) of 20CR using annual means of: a) qc time series b) hom time series
a) b)
a)
b) c)
5
a) b)
Peru
qc hom
qc
hom
20CR
global picture South America
qc
hom hom-qc
hom qc hom-qc
1964-2007
Fig. 5: Correlation coefficients calculated separately for every month between the time series of 2m mean temperature, extracted from the closest gridpoint to the Recuay station from every reanalysis dataset, and: a) the raw (original) quality-controlled station record (qc); b) the homogenized station record (hom). c) shows the difference between the R values: hom - qc The reanalysis datasets used are, as indicated in the legends: 1) (dark blue) NCEP-NCAR Rean. (NNR) for 1964-1998 [3] 2) (red) 20CR for 1964-2012 [1] 3) (black) MERRA for 1979-2008 [4] 4) (green) ERA-40 for 1964-2001 [5] 5) (pink) ERA-20C for 1964-2010 [6]
