Los Reanálisis
Los Reanálisis
Los Reanálisis
Eugenia Kalnay
Universidad de Maryland
En el Hemisferio Sur, los pronsticos sin datos de satlite eran muy malos.
Decidimos empezar en 1957 (y luego extendimos a 1948) porque los mapas
mensuales eran buenos. (Luego ERA-40 hizo lo mismo)
Contenido de la charla
Asimilacin de datos
Porque se crean los reanlisis?
Algunos ejemplos de Kistler, Kalnay et al. (2001)
Los reanlisis que ya existen: NCEP-NCAR
(NCEP 1), NCEP-DOE (NCEP 2), ERA-40, NARR,
JRA-25, NASA-MERRA, y los futuros
NRC reporte sobre reanlisis: Captulo 2
Caractersticas, similaridades, differencias
Aplicaciones: estudiar circulaciones, pronsticos
experimentales
Cuidado! Algunas variables (ej. lluvia), son menos
seguras. Tambin las tendencias
Preguntas y discusiones
Asimilacin de datos
Observations (+/-3hrs)
Background or FG
Initial Conditions
6 hour forecast
(operational forecasts)
forecast
Campos grillados
(anlisis)
Repronstico de 4 das!
Analysis input:
many different
types of
observations
Analysis
output:
gridded
analyses
FGGE (1979)
IGY (1958)
Different Reanalyses
can have somewhat
different climatologies
depending on the
models and the
observational coverage.
Nevertheless, they all
capture similar
interannual anomalies
1960
1980
2004
Temperature
profile
observations
in the first
three months
for different
years
1960
1970
1980
1990
2000
no trend
Observed
temperature
1960
1970
1980
1990
2000
1980
1990
2000
large trend
Model/obs
combination
1960
1970
Current global reanalysis data are most reliable in quantities that are
most strongly constrained by the observations (e.g., temperature and
winds), and least reliable for quantities that are highly model
dependent, such as evaporation, precipitation, and cloud-related
quantities.
Substantial biases exist in various components of the atmospheric water cycle (e.g.,
precipitation, evaporation and clouds), that limit the value of current reanalysis data for
assessing the veracity of these quantities in climate models, as well as for practical
applications. There are also significant biases in other surface and near-surface quantities
related to deficiencies in representing interactions across the land-atmosphere and oceanatmosphere interfaces.
But, the comprehensive and multi-variate nature of reanalysis data provide value for
understanding the causes of surface temperature and precipitation trends beyond what can
be obtained from relatively incomplete observational datasets alone, even in the face of
the noted biases in reanalysis-based trends.
Reanalysis data play a critical role in assessing the ability of climate models to simulate
the statistics of climate the means and variances (at various time scales) of basic
variables such as the horizontal winds, temperature and pressure. In addition, the
adjustments or analysis increments (i.e., the "corrections" imposed on model states by the
observations) produced during the course of a reanalysis provide a means to identify
fundamental errors in the physical processes and/or missing physics that create climate
model biases.
Reanalyses have already had enormous benefits for climate research and
prediction, as well as for a wide range of societal applications.
NCEP-NCAR 1 (NNR):
El ms largo (1948-presente)
Un sistema operacional en 1994
T62-28 niveles (~2o)
3D-Var sin inicializacin
Salida en 2.5x2.5o en la presin, niveles mandatorios, cada 6 horas
Tambin la grilla nativa, isentrpica y parameterizacines (ej. calentamiento,
flujos de superficie, precipitacin del modelo)
El ms usado (artculo ms citado en todas las geociencias, >5000 citas)
Fcilmente accesible (NOAA CDC)
Problemas
Baja resolucion
Algunos errores: PAOBS, precipitacin espectral
Ms recientemente completado
Cubre el perodo 1979-2004
Accesible gratis
Ventajas
Reanlisis de un siglo
Compo, Whitaker, Hamill
Usa el mtodo de Ensemble Kalman Filter (EnSRF)
Conclusiones
Preguntas?