Towards Seamless TCP Congestion Avoidancein Multiprotocol Environments
Going seamless - towards
36
Morse IS-EPS WCRP Seasonal Prediction Barcelona 2007 Going seamless - towards Suggested seamless ranges – open to discussion • Medium range, monthly, seasonal – WCRP-WWRP interface • Seasonal, interannual, decadal - ENSEMBLES • Seasonal, interannual, decadal, centennial Suggested seamless approaches • Application models across modelling streams – ENSEMBLES • Grand ensemble approach – THORPEX medium range • Ensemble dressing
description
Going seamless - towards. Suggested seamless ranges – open to discussion Medium range, monthly, seasonal – WCRP-WWRP interface Seasonal, interannual, decadal - ENSEMBLES Seasonal, interannual, decadal, centennial Suggested seamless approaches - PowerPoint PPT Presentation
Transcript of Going seamless - towards
Morse IS-EPS WCRP Seasonal Prediction Barcelona 2007
Going seamless - towards
Suggested seamless ranges – open to discussion
• Medium range, monthly, seasonal – WCRP-WWRP interface
• Seasonal, interannual, decadal - ENSEMBLES
• Seasonal, interannual, decadal, centennial
Suggested seamless approaches
• Application models across modelling streams – ENSEMBLES
• Grand ensemble approach – THORPEX medium range
• Ensemble dressing
WP6.1 Global changes in biophysical and biogeochemical processes – integrated analysis of impacts and feedbacks- ACC GCM - Colin Prentice - UNIBRIS, UREADMM, PIK, ULUND, METO-HC, CNRS-IPSL
WP6.2 Linking impact models to probabilistic scenarios of climate change- RCM – Tim Carter – SYKE, DIAS, DISAT, FMI, FUB, NOA, PAS, SMHI, UEA, ULUND, UNIK, UREADMM
WP6.3 - Impact modelling at seasonal to decadal timescales- s2d – Andy Morse - UNILIV, UREADMM, ARPA-SIM, JRC-IPSC, METEOSWISS, LSE, IRI, EDF, DWD
ENSEMBLES GA 2006 Lund RT6 Plenary
RT6 Assessments and Impacts of Climate Change
24 partners plus affiliated partners
The three WPs – method and model refinement and application and impact model runs with existing climate model data e.g. DEMETER, AR4 etc. And publishing papers.
Partners working within or towards a probabilistic framework.
Range of application models – some themes through all WPs e.g. crops and > 1 WP e.g. fire, wind damage
Tasks on going – and ready to move to ENSEMBLES data streams – Stream 1 - s2d and ACC and, RCM as available
ENSEMBLES GA 2006 Lund RT6 Plenary
RT6 Assessments and Impacts of Climate Change
Role of users – malaria plumes
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
1 31 61 91 121 151
Forecast Day
Ma
lari
a P
rev
ale
nc
e
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
180 30 60 90 120 150
Forecast Day
Ma
lari
a p
rev
ale
nc
e
Morse IS-EPS WCRP Seasonal Prediction Barcelona 2007
Botswana malaria forecast for February 1989, LMM driven by DEMETER multi-model
(ERA-driven model shown in red)
All plots unpublished Anne Jones, University of Liverpool
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
180 30 60 90 120 150
Forecast Day
Ma
lari
a p
rev
ale
nc
e
• November 1997 start
• Improvement in skill due to temperature correction
• If temperatures too low, delay in model is increased
95
85
65
35
15
5ERA
Tier-3 ROC AreasTier-3 ROC Areasfor November malaria forecastfor November malaria forecast
ROC Area (<0.5 = no skill), Upper Tercile event, forecast 6 month totals for Botswana grid average, () 95% confidence intervals calculated from 1000 bootstrap samples
Validated against Thomson et al (2005) Malaria Index
LMM Input Data ROC Area
ERA-40 0.78
(0.56-0.97)
Raw DEMETER 0.31
(0.11-0.56)
Bias-corrected DEMETER
T correction only
0.67
(0.29-0.80)
0.70
(0.35-0.90)
ERA-40 "persistence"
(wrong year ensemble)
0.32
(0.07-0.63)
Tier-1 ROC Areas for November rainfall forecast
Event Uncorrected Corrected
Lower Tercile 0.88
(0.67-1.0)
0.75
(0.45-1.0)
Above the median
0.65
(0.37-0.89)
0.68
(0.39-0.92)
Upper Tercile 0.72
(0.46-0.93)
0.63
(0.37-0.86)
ROC Area (<0.5 = no skill),forecast 6 month totals for Botswana grid average, () 95% confidence intervals calculated from 1000 bootstrap samples
Validated against ERA-40 rainfall totals
Effect of rainfall bias correction Effect of rainfall bias correction
•Bias correction of rainfall causes decrease in skill
ECMWF Forecast User Group Meeting, June 2006 [email protected]
15
17
19
21
23
25
27
29
1 31 61 91 121 151
Forecast Day
Me
an
T (
de
gre
es
C)
Effect of temperature bias correctionEffect of temperature bias correction
• Temperature variability not a strong driver of malaria variability in this region
• However malaria model requires realistic temperatures
• DEMETER temperatures need to be bias corrected to achieve this, because models is sensitive to biases in uncorrected data of ~ 2 degrees
DEMETER temperature forecasts for Botswana, November 1997
95
85
65
35
15
5ERA
Uncorrected Temperature Corrected Temperature
15
17
19
21
23
25
27
29
1 31 61 91 121 151
Forecast Day
Me
an
T (
de
gre
es
C)
ECMWF Forecast User Group Meeting, June 2006 [email protected]
Tier-3 ROC AreasTier-3 ROC Areas--alternative model outputsalternative model outputs
LMM Input Data LMM Malaria Anomalies
LMM Mosquito Anomalies
ERA-40 0.78
(0.56-0.97)
0.89
(0.71-1.0)
Raw DEMETER 0.31
(0.11-0.56)
0.73
(0.42-0.97)
Bias corrected DEMETER 0.67
(0.29-0.80)
0.56
(0.42-0.91)
ROC Area (<0.5 = no skill), Upper Tercile event, November forecast 6 month totals for Botswana grid average, () 95% confidence intervals calculated from 1000 bootstrap samples
Validated against Thomson et al (2005) Malaria Index
• Skill improved by using model mosquito numbers
• Bias correction decreases skill due to strong rainfall driver
• Cannot use in other areas where temperature a stronger driver
Conclusions – Botswana LMM dynamicConclusions – Botswana LMM dynamic
• DEMETER-driven forecasts were skilful, better than climatology and persistence forecasts
• Bias correction of temperature is important even if variability in temperature not important - temperatures must be "realistic" for the application model
• Bias correction of rainfall is unsatisfactory - use of daily rainfall output is problematic and need to consider other methods using monthly anomalies instead (e.g. weather generator)
• Lag in model mean malaria cases may occur outside forecast window - can be solved for Botswana using mosquito model but not applicable to other areas
ECMWF Forecast User Group Meeting, June 2006 [email protected]
Morse IS-EPS WCRP Seasonal Prediction Barcelona 2007
Role of users - Statistical Model Malaria PDFs
The probability distribution functions of predicted standardized log malaria annual incidence for the years 1992 (anomalously low incidence, left) and 1993 (anomalously high incidence, right) computed with the DEMETER multi-model ensemble forecast system are depicted in red. Observations Botswana Ministry of health in blue
from M.C. Thomson, F.J. Doblas-Reyes, S.J. Mason, R. Hagedorn, S.J. Connor, T. Phindela, A.P. Morse, and T.N. Palmer (2006). Malaria early warnings based on seasonal climate forecasts from multi-model ensembles, Nature, 439, 576-579.
Morse IS-EPS WCRP Seasonal Prediction Barcelona 2007
Role of users - Tanzania statistical malaria model
1991 1992 1993 1994 1995 1996 1997 1998 1999
Year
An
om
alie
sDEMETER median Log Malaria Weather Station
Jones et al. 2007 Fig 5d (submitted)
Statistical model ‘C3’ driven by Feb-Jul DEMETER pptn and Aug-Jan. Tmx ob. Giving box-whisker malaria prediction Apr-Sep – obs. driven control, obs. malaria - all standardised anomalies.
Vittorio MarlettoENSEMBLES GA 2006 Lund RT6 Plenary
RT6 WP6.3
Impact modelling at seasonal to decadal timescales
Regional crop yield forecastingTo develop tier-3 validation systems. Dynamic crop model run with 72 downscaled Demeter hindcast members. Comparison with local field data showed interesting skill in forecasting yield two months before harvest.
years
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
0
2000
4000
6000
8000
10000
12000
Percentile yields (5, 10, 25, 50, 75, 90, 95) obtained using weather data from Cadriano up to the end of April and downscaled Demeter output up to harvest date.Observed yieldsYields simulated only with weather data from Cadriano
kg/ha
Andrew Challinor, Tom Osborne, Tim WheelerENSEMBLES GA 2006 Lund RT6 Plenary
RT6 WP5.5
Impacts of mean temperature rise andextreme temperature events on crop yield
The relative importance of mean and extremes of temperature was examined on crop yield (Task 5.5.x). Response varied geographically, and depended upon the parameterised response of the crop. Yields were particularly threatened when the reproductive temperature threshold was exceeded, but not the development rate threshold.
Boxplots of the percentage change in yield, between the baseline and projection climates for groundnut. ‘S’ and ‘T’ refer to sensitive and tolerant genotypes ; Topt is 28 or 36 oC; ‘OF’ without high temperature stress; ‘ON’ with high temperature stress.
Paul Della-Marta et al.ENSEMBLES GA Lund RT6 Plenary
RT6 WP6.3
Impact modelling at seasonal to decadal timescales
PreWiStor: Wind storm risk and prediction
Calibration of ERA40 and S2d data to be compatible with Swiss Re wind storms
Preliminary methods over calibrate the wind gust
Unreliable calibrations are masked in white
ERA40 wind storm Calibrated wind storm
Contribution to D2B.12
Morse IS-EPS WCRP Seasonal Prediction Barcelona 2007
Going seamless - towards
Suggested seamless ranges – open to discussion
• Medium range, monthly, seasonal – WCRP-WWRP interface
• Seasonal, interannual, decadal - ENSEMBLES
• Seasonal, interannual, decadal, centennial
Suggested seamless approaches
• Application models across modelling streams – ENSEMBLES
• Grand ensemble approach – THORPEX medium range
• Ensemble dressing
R index of “reproducibility” (Stern and Miyakoda, 1995)
s
nR
where n is the standard deviation of the ensemble
members about the ensemble mean and s is the
climatological variability calculated over all ensemble members and years.
W. Stern and K. Miyakoda, 1995. Feasibility of Seasonal Forecasts Inferred from Multiple GCM Simulations. Journal of Climate: Vol. 8, No. 5, pp. 1071–1085.
“Ratio of variances” used in CEDO BRANKOVIC; T.N. PALMER, 2000. Seasonal skill and predictability of ECMWF PROVOST ensembles. Quarterly Journal of the Royal Meteorological Society, Volume 126, Number 567, July 2000 Part B, pp. 2035-2067(33)
Pan-African R scores
Rainfall (uncorrected)Temperature (bias corrected)
DEMETER multimodel ensemble, 1980-2001, Feb forecast totals/mean 1-6 no masking for zero and low CoV
Time-averaged R for Botswana Malaria incidence – February forecast
Feb 2-4
1982-2001 averages of R, malaria incidence for DEMETER multimodel using bias-corrected temperature. Mask ERA<1 case per 100 people per month, CV<0.5
R DEMETER CV
Feb 2-4
Time-averaged R for Botswana Malaria incidence – November forecast
Nov 2-4
1982-2001 averages of R, malaria incidence for DEMETER multimodel using bias-corrected temperature. Mask ERA<1 case per 100 people per month, CV<0.5
R DEMETER CV
Nov 4-6
Distributions of Malaria Incidence for each model and multi-model
“Epidemic” grid points, Feb 2-4
Distributions of Malaria Incidence for each model and multi-model
“Epidemic” grid points, Nov 4-6
Rainfall
Time-averaged R for Botswana Rainfall – February forecast
Feb 2-4
1982-2001 averages of R, malaria incidence for DEMETER multimodel using bias-corrected temperature. Mask ERA<1 case per 100 people per month, CV<0.5
R DEMETER CV
Feb 2-4
Time-averaged R for Botswana Rainfall – November forecast
Nov 2-4
1982-2001 averages of R, malaria incidence for DEMETER multimodel uncorrected normalised rainfall. Mask from incidence ERA<1 case per 100 people per month, CV<0.5
R DEMETER CV
Nov 4-6
Distributions of Rainfall for each model and multi-model
“Epidemic” grid points, Feb 2-4
Distributions of Rainfall for each model and multi-model
“Epidemic” grid points, Nov 4-6
Temperature
Time-averaged R for Botswana temperature – February forecast
Feb 2-4
1982-2001 averages of R, malaria incidence for DEMETER multimodel bias-corrected temperature. Mask from incidence ERA<1 case per 100 people per month, CV<0.5
R DEMETER CV
Feb 2-4
Time-averaged R for Botswana temperature – November forecast
Nov 2-4
1982-2001 averages of R, malaria incidence for DEMETER multimodel bias-corrected temperature. Mask from incidence ERA<1 case per 100 people per month, CV<0.5
R DEMETER CV
Nov 4-6
Distributions of temperature for each model and multi-model
“Epidemic” grid points, Feb 2-4
Distributions of temperature (bias-corrected) for each model and multi-model
“Epidemic” grid points, Nov 4-6
Morse IS-EPS WCRP Seasonal Prediction Barcelona 2007
original ERA
Role of users –MARA transmission map
Based on model Craig et al. 1999 www.mara.org.za run with climatology and observations and ERA-40
Morse IS-EPS WCRP Seasonal Prediction Barcelona 2007
Botswana malaria
MARA map
(Craig et al, 1999)
Tmin
Rain
Tmean
MARA run with ERA-40 1 deg (1958-2001)
MARA limiting variable (Tmin is min of 4 daily values) 5 degree –ve offset
slide from Anne Jones, University of Liverpool
