Using satellite data to understand uncertainties in reanalyses: UERRA Richard Renshaw, Peter Jermey with thanks to Jörg Trentmann, Jennifer Lenhardt, Andrea Kaiser-Weiss (DWD) © Crown Copyright 2012 Source: Met Office

Outline 1.Why Regional Reanalysis ? 2.EURO4M Regional Reanalysis – Evaluation 3.Regional Reanalysis Plans Uncertainty Estimation in Regional ReAnalysis (UERRA) project

What is a reanalysis ? state-of-the-art NWP past observations gridded analyses

Gridded data Based on observations Incorporates model equations Physically and dynamically coherent Full set of meteorological fields We can estimate accuracy Why would anyone want a reanalysis ?

Global Reanalyses NCEP-NCAR (1995)250km ECMWF ERA-40 (2004)130km ECMWF ERA-Interim (2010)80km NOAA/CIRES 20 th C (2011)200km JMA JRA-55 (2013)60km...

ERA Interim Reanalysis Dee et al, 2011 Global atmosphere, T255 (80km), 60 vertical levels 12-hour 4D-Var present

© Crown copyright Met Office 1. Why Regional Reanalysis ?

Evidence from operational NWP 25km Global vs 12km NAE

...the benefits of resolution forecast range screen temperature rms error (K) global 25km NAE 12km

...and the disadvantage of boundaries! forecast range mean sea level pressure rms error (Pa) global NAE

EU-project, April 2010 – March 2014, 9 partners Goal: LONG-TERM CLIMATE DATASETS + ASSESSMENTS OF CHANGE …describing climate variability and change at the European scale …placing high-impact extreme events in a historical context

European Regional Reanalysis: EURO4M project EURO4M project ( ) developed UM regional reanalysis, tested on 2 year period ( ). Resolution: 12km model, 24km 4D-Var Lateral boundary conditions from ERA-Interim ECMWF observation archive

Increase in resolution EURO4M: Model/DA: 12/24kmERA-Interim: Model/DA 80/125km

Observations Surface (SYNOP, buoy, etc) Upper air (sonde, pilot, wind profiler) Aircraft AMV (‘satwinds’) GPS-RO and ground-based GPS Scatterometer winds ATOVS AIRS IASI MSG clear sky radiances

Getting more from surface obs... Visibility Cloud Rainfall

© Crown copyright Met Office 2. EURO4M Regional Reanalysis Evaluation Peter Jermey

Russian heatwave, July 2010 Tmax, e-obs

KNMI, Ge Verver et aI ECAD: European Climate Assessment and Dataset Daily data from

Maximum temperature on 10 July 2010 (during the Russian heat wave) obs daily 25km grid “E-Obs”

Tmax ERA-Interim 12km EURO4M obs

© Crown copyright Met Office Climate Statistics Monthly Means MO ERAT Compare with ECA&D statistics from obs stations

© Crown copyright Met Office Precipitation Higher resolution should lead to improved representation of extremes Covers wide range of intensities, periods and scales Flooding in central Europe in 2013 caused 25 deaths and 12bn Euros damage

© Crown copyright Met Office Floods July houses damaged 20,000ha ag. land flooded 300 houses destroyed 7500ha ag. land flooded 50,000 houses flooded cost $700million 5 dead $100 million 300,000 people affected 38 dead 3 dead $300million ROMANIAMOLDOVAUKRAINE

© Crown copyright Met Office Floods July th July Accumulations SYNOP ERA-Interim UKMO 15mm Mean abs error 13mm

© Crown copyright Met Office ETS precip scores 6hr ERA-Interim HIRLAM Met Office Truth is SYNOP rain gauge data

© Crown copyright Met Office Frequency bias 6hr At low thresholds models over-represent At high thresholds models under-represent, but … … bias is reduced by increased resolution & 4DVAR assimilation

Concept, methods and results Deutscher Wetterdienst (DWD) Jörg Trentmann, Jennifer Lenhardt Evaluation of EURO4M Reanalysis data using Satellite Data

Data sets (monthly means) 28 EURO4M Reanalyses vs. CM SAF EURO4M Final Assembly – 03/2014 CM SAF products → CLARA-A1 (AVHRR Cloud Cover at 0.25°) → ATOVS (Integrated Water Vapour at 90x90 km) EURO4M product → Merged GPCC (raingauge, land)/HOAPS (SSM/I, ocean) (Precipitation at 0.5°)

Mean differences, cloud cover, MetOffice - CM SAF (AVHRR), July 2008/9 29EURO4M Final Assembly – 03/2014

Mean differences, precipitation, MetOffice - CM SAF, July EURO4M Final Assembly – 03/2014

Mean differences, water vapour, MetOffice - CM SAF (ATOVS), July 2008/9 31EURO4M Final Assembly – 03/2014

Validation Reanalysis is only useful if we know the errors Reanalysis fields are already of good quality Conventional obs have limited coverage Validation datasets need to be independent Datasets need to be good quality, with error estimates Some variables difficult to validate

© Crown copyright Met Office Regional Ensemble

Uncertainties from ensembles Calibrate for variables we can validate Get uncertainties for variables we can’t validate

Uncertainty Estimation in Regional ReAnalysis (UERRA) Project EURO4M represents just an initial step towards a full regional reanalysis capability. UERRA ( ) will provide a multidecadal, multivariate dataset of essential climate variables (ECVs) for the satellite era (1978-present). UERRA will include an ensemble regional reanalysis UERRA described as a component of a ‘pre-operational’ climate service, preparing the way for reanalysis as a central pillar of the Copernicus Operational Climate Service.

Assessing uncertainties by evaluation against independent observational datasets DWD, KNMI, MI, EDI, UEA, NMA-RO, MO EURO4M and UERRA GA March Exeter 2014 Andrea Kaiser-Weiss, DWD

WP3 Objectives To evaluate deterministic, ensemble reanalyses and downscaled reanalyses through comparison to ECV datasets, that were derived independently To establish a consistent knowledge base on the uncertainty of reanalyses across all of Europe, by adopting a common evaluation procedure for ECVs, derived climate indicators, extremes and scales of variability that are of particular interest to users To statistically assess the provided information over Europe by applying the common evaluation procedure to the reanalyses products, gridded datasets and satellite data

WP3 Objectives (cont.) To apply the common evaluation procedure for special climate features of selected sub-regions of Europe, providing feedback on the reliability of measures of uncertainty contained in reanalyses To synthesize the results of the evaluation into a general assessment of the reliability and uncertainty of regional reanalysis that guides users in the state- of-the-art application of the datasets produced in WP2

Summary 1. Reanalysis useful climate services but we need to know the errors. 2. Need high quality datasets for validation, plus knowledge of their errors 3. Ensemble reanalysis should allow better characterisation of uncertainties

Interest from UM partners India South Korea New Zealand

© Crown copyright Met Office Thank you for listening…