1. Convection Products
NWC SAF 2015 Users’ Workshop February Madrid AEMET HQ, Madrid (Spain)
Jean-Marc Moisselin, Frédéric Autones Météo-France –Nowcasting Department 42, av. Gaspard Coriolis Toulouse France

2. Convection Products at a glance
Two convection products: CI and RDT
CI = Convection Initiation
= probability for a cloudy pixel to become a thunderstorm
RDT = Rapidly Developing Thunderstorm
= detection, tracking, description and forecast of thunderstorms in object mode

3. Overview
RDT CI
Links between Products
MTG Context

4. RDT: data fusion for description of convection
INPUT DATA: MULTISOURCE
MSG data (5 IR channels + VIS)
NWP data
PGE11
RDT
Lightning Data
Other NWCSAF products
OUTPUT DATA: MULTILEVEL DESCRIPTION OF CONVECTION
Main description of cell: Yes/No convection diagnosis, cell-development phase, position, surface, T, gap to tropopause, cloud type and phase, cloud top pressure. Displacement Relevant trends are calculated
Overshooting Tops, Lightning Activity, Convective Index, Rainfall Activity

5. 3-steps algorithm of RDT5
3-steps algorithm of RDT
STEP1: Detection (in order to detect cells)
- Using vertical profile of 10.8µm BT
- Cells (towers) are detected at each slot
- Vertical extension: at least 6°C
STEP2: Tracking (in order to recognize each cell in the previous slot)
- Analysis of cloud cells overlap: each cell of the previous slot is advected
- Merges and splits are taken into account
- Trends of various parameters are calculated  
STEP3: Discrimination (in order to identify convective cells). Statistical process
- Made complex by the unbalanced populations, the wide variety of scales and evolution-phases of systems
- Highly improved by the use of a set of 5 IR-channels as predictors, by the use of NWP data
- Very highly improved by the use of lightning data  4
STEP4: Forecast 5

6. 6
Visualization of RDT through SYNERGIE (Météo-France forecasters’ workstation)
Development-phases:
Yellow: First detection of convective system
Red: Developing system
Purple: Mature system
Blue: Decaying system
Orange: After a split of systems
Tracking:
- motion vector
- trajectory
Attributes: 6

7. Evolution of RDT product
Since IOP ( )
Pursued in CDOP, CDOP2, proposal for CDOP3
Recent evolutions
v2011: use of NWP data
v2012: main cloud phase of the cell, highest convective rain rate inside the cell, second vertical level description
v2013: overshooting tops
v2016: advection scheme + change in NWCSAF Library + new output format + CTRAJ
CDOP3: CDOP2 continuation
GEO imager for MSG, MTG and a set of other meteorological satellites

8. RECOMMENDATION: USE NWP DATA!
v2011: impact of NWP data
How:
CONVECTIVE INDEX calculated for each pixel - A mask is built at the beginning of RDT process. Allows to focus on areas of interest - Provides an additional predictor
Consequences: - New attribute
- Strong reduction of the false alarms during intermediate and winter seasons
Improvement of early detection
EXAMPLE
25 May 2009, 12h15 UTC.
v2011 benefits from a better tuning in warmer categories, with higher early detection (cells over Italy diagnosed 30 min when v2011 and v2010 releases are compared)
v2010: without NWP data
v2011: with NWP Nata
RECOMMENDATION: USE NWP DATA!

9. v2012: 2nd level description
When cell-extension is too large, it is interesting to have the depiction of another level additionally to « Base of Tower » level.
An outline related to the « Top of Tower » has been added
2nd contour: specific attributes
main contour: general attributes, including tracking attributes

10. v2013: OTD (Overshooting Tops Detection)
OTD Inside each RDT cell
Temperature of coldest pixel, BTD WV6.2-IR10.8, WBTD WV6.2-WV7.3, reflectance VIS0.6, gap to NWP tropopause.
Morphologic criteria to confirm a spot of cold temperatures and to determine the pixels that belong to an OT
HRV for tuning/validation

11. RDT: validation Subjective validation by Météo-France
various case studies, use of topical case for each release.
Objective validation by Météo-France
Results fulfil the target accuracy requirements over a large domain and for an extended period: detection is superior to 70% and 25% of convective systems are diagnosed before lightning activity.
Validation by users
any feedback is welcome

12. RDT: v2016 and following releases
v2016: advection scheme + change in NWCSAF Library + new output format + CTRAJ + use of CMIC
v2017: ovelapping CDOP2 and CDOP3. Improved links between other products (Ci as predictor). Himawari : impact of new channels and higher resolution. New tuning.
CDOP3: CDOP2 continuation – v2017, then version ready for MTG Day1, then preparation of a Day2 version
GEO imager for MSG, MTG and a set of other meteorological satellites
WNW displacement of “A”
“A” cell has disappeared. Bad forecast (False Detection)
20:10
21:10 A
v2016 B
“B” at the expected location. Even if change in morphology is not forecast
WSW displacement of cell “B”

13. Overview
RDT CI
Links between Products
MTG Context

14. Convection Initiation (CI) New NWC SAF product released in v201614
Convection Initiation (CI) New NWC SAF product released in v2016
Low probability to develop into a thunderstorm
High probability to develop into a thunderstorm
The convection probability for each pixel is based on:
BT or BTD values or trends, e.g. BDT µm. Some relevant Parameters of Interest in « Best Practice Document For EUMETSAT Convection Working Group » Editors J. Mecikalski, K. Bedka, M. Marianne König
NWCSAF products: Clear Air Products, Cloud Products, Wind Products
NWP data
Past positions and characteristics of pixel
NWCSAF Clear Air Product, ie Precipitable Water parameters, Instability Indices based on Physical Retrieval
NWCSAF Cloud Product, ie type of cloud, vertical extension, liquid water/ice path
NWCSAF Wind Product, ie convergence, vorticity
Model product, ie instability index
Past positions and characteristics of pixel 14

15. CI – Main principles
The CI product will be elaborated in two main steps.
The first step concerns the selection of pixels of interest. We propose to exclude the pixels that are too cold or are already thunderstorm (using RDT tracks): “too much” mask. We also exclude cloud-free pixels (using NWCSAF cloud products): “not enough” mask. Masks are combined
Then cloud tower are identified having at least 3°C of vertical extension and a surface lower than km²
The second step concerns the probability calculation. Accordingly to literature on the subject three categories of predictors will be considered:
Vertical extension of the cloud,
Ice presence,
Cloud growing rate
In order to compute the trends we take into account cloud-displacement and calculate the past position of the cloud. For that purpose the HRW (High Resolution Wind) product of NWCSAF will be use additionally for an object-based movement detection.
Verification/tuning: RDT / radar / lightning data

16. Pixels of Interest selection – Preliminary results

17. Overview
RDT CI
Links between Products
MTG Context

18. Links between products: RDT case
RDT product:
Cloud products:
To operate RDT on cloudy areas
Use of CMIC for v2016
For the advection scheme. Complementary or additionally to cell speed estimate HRW/AMV can provide useful information: for new cells without speed estimate, in case of uncertainty in speed estimate, in case of merge or split. Foreseen for v2016
RDT diagnosis: to validate RDT forecast. Foreseen for v2016
Lightning data: to validate RDT diagnosis (operated in full configuration but without lightning data).
RDT diagnosis (motion vector, cooling rate): to contribute to RDT forecast
CRR: an attribute for RDT, a possibility to set the convection diagnosis to Yes if CRR above a threshold. Since v2012
CI: predictor for RDT discrimination scheme. If a pixel has a high probability to become a thunderstorm it increase the chance of the corresponding cell to be defined as convective. Foreseen for v2017

19. Links between products: CI case
CI product
CI [0-60’] to be coherent with CI [0-30’] (p0-90’>p0-60’>p0-30’)
Cloud products: to operate CI on cloudy areas (mask)
RDT (diagnosis): to operate CI on non-thunderstorm areas (mask)
RDT (diagnosis - full configuration but without CI): to validate CI (additionally to radar and lightning data)
Some predictors of this new pixel-based product are based on trends. These trends have to be calculated following the cloud track. For that we will use the HRW/AMV in the low layers of the atmosphere.

20. Overview
RDT CI
Links between Products
MTG Context

21. MTG Context LI is eagerly expected
LI instrument is eagerly expected to improve many components of RDT:
Statistical scheme,
Real time mode,
Enhancement of characteristics for a more complete description of convection,
Monitoring.
image credit: ESA

22. MTG Context FCI is eagerly expected
Number of channels:
Resolution:
Better estimate of morphological parameters and small scale phenomena
Spectral accuracy: better estimate of BT input data of RDT
RSS Challenge
The lack of channels in RSS would mean for RDT a lack of predictors and a lower quality
FCI
In previous years, RDT algorithm has always taken a lot of advantages from the increase of the number of channels
e.g. for FCI: 0.91µm (total column precipitable water)
MSG SEVIRI
Meteosat7
RDT Cell with OT

23. Conclusion CI and RDT product
Algorithm: RDT algorithm has been developed before CI algorithm
Meteorology / conceptual model: CI occurs before.
Two products different in terms of presentation
CI: image mode
RDT: object mode
Strong links between these two products: input data, exclusion mask, validation
Strong links with other NWCSAF products: cloud products, CRRR, HRW/AMV
Together they offer a complete description of convective systems in various phases