Observing Convection with FY-4A Satellite · 2018-06-05 · Observing Convection with FY-4A...
Transcript of Observing Convection with FY-4A Satellite · 2018-06-05 · Observing Convection with FY-4A...
Observing Convection with FY-4A Satellite
Qin Danyu
Sun Fenglin, Shou Yixuan, Wu Chunqiang, Cao Dongjie
National Satellite Meteorological Center (NSMC), CMA
The CWG Workshop Meeting, Ljubljana, Slovenia, 17-19 April 2018
GIIRS: Geo. Interferometric Infrared Sounder
AGRI: Advanced Geosynchronous Radiation Imager
LMI: Lightning Mapping Imager
SEP: Space Environment Package
FY-4A: New Era of China GEO Satellite
• The FY-4A satellite was launched successfully on 11 Dec 2016
• It’s position is now at 104.7°E
• Data have been broadcasted by CMACast since 12 Mar 2018.
Spectral
Coverage
Spectral
Band (µm)
Spatial
Resolution (Km) Sensitivity
Main
Applications
VIS/NIR
0.45~0.49 1 S/N≥90 (ρ=100%) Aerosol
0.55~0.75 0.5~1 S/N≥200 (ρ=100%) Fog, Clouds
0.75~0.90 1 S/N≥5(ρ=1%)@0.5Km Vegetation
1.36~1.39 2
S/N≥200 (ρ=100%)
Cirrus
1.58~1.64 2 Cloud,Snow
2.10~2.35 2~4 Cirrus,Aerosol
Middle-
wave IR
3.50~4.00 2 NEΔT≤0.7K(300K) Fire
3.50~4.00 4 NEΔT≤0.2K(300K) Land surface
5.80~6.70 4 NEΔT≤0.3K(260K) WV
6.90~7.30 4 NEΔT≤0.3K(260K) WV
Long-wave
Infrared
8.00~9.00 4 NEΔT≤0.2K(300K) WV,Clouds
10.3~11.3 4 NEΔT≤0.2K(300K) SST
11.5~12.5 4 NEΔT≤0.2K(300K) SST
13.2~13.8 4 NEΔT≤0.5K(300K) Clouds,WV
Spectral Parameters
(Normal mode)
Range Resolution Channels
LWIR: 700-1130 cm-1 0.8 538
S/MIR:1650-2250 cm-1 1.6 375
VIS : 0.55- 0.75 μm
Spatial Resolution LWIR/MWIR : 16 Km SSP
VIS : 2 Km SSP
Operational Mode China area 5000 5000 Km2
Mesoscale area 1000 1000 Km2
Temporal Resolution China area <1 hr
Mesoscale area <½ hr
Sensitivity
(mW/m2srcm2)
LWIR: 0.5-1.1 S/MIR: 0.1-0.14
VIS: S/N>200(ρ=100% )
Calibration accuracy 1.5 K (3σ) radiation
Calibration accuracy 10 ppm (3σ) spectrum
Quantization Bits 13 bits
Spatial resolution about 7.8Km at SSP
Sensor size 400300 2
Wave-length at center 777.4nm
Band-width 1nm±0.1nm
Detection efficiency >90%
False-alarm ratio <10%
Dynamic range >100
SNR >6
Frequency of frames 2ms
Quantization 12 bits
Measurement Error 10%
Characteristics of Payloads (Specification & Main Usage)
AGRI’s Main Usage:
Acquire multiple band, high
temporal resolution, high
radiation accuracy images
of Earth’s surface,
atmosphere and cloud
GIIRS’s Main Usage:
Acquire atmospheric
temperature and humidity
profile structures under
clear condition
LMI’s Main Usage:
Acquire lightning distribution
maps for a certain coverage
AGRI
GIIRS
LMI
Three steps of FY-4A IOT :
1st stage :20161226-20170630,testing mainly for satellite function and performance,
spatial segment of FY-4A is turned over to end users
2nd stage:20170630-20170930,testing mainly for L1 products
3rd stage: 20171001-20171230,testing mainly for L2+ products .
Now operational used
Stages of in-orbit testing (IOT) for FY-4A
AGRI
Cyclone in Australia Haze in the Bay
of Bengal
Vortex in the South Pole Area
Cellular Clouds
in the South Pole Area
Tropical Cyclone
(local area)
Tibet and its surrounding areas
Tropical Cyclone
(wide area) Frontal Cloud across mid China Snow Cover Monitoring
In north China
卡努 KHANUN(1720)
2017.07.06
Binary typhoons
TALIM (1718)
and DOKSURI(1719)
2017.06.24
The Meiyu frontal cloud system over Yangzi river basin, it is a stationary frontal system, usually can last a few days.
Clear Sky Masks Cloud Type Cloud Optical Depth Cloud Liquid Water Path Cloud Ice Water Path Cloud Particle Size Distribution Cloud Phase
Cloud Top Temperature Cloud Top Height Cloud Top Pressure
FY-4A L2+ product examples
Aerosol Detection Rainfall Rate/QPE Atmospheric Motion Vector Downward Long wave
Radiation:Surface
Upward Long
wave Radiation:
Surface
Reflected
Shortwave Radiation
Land Surface (Skin)
Temperature Sea Surface (Skin)
Temperature Land Surface Emissivity Land Surface Emissivity
FY-4A Convection Product
For the Rapid Developing Convection product,
What we did during the IOT
Solar zenith angle correction for VIS albedo
Multi channel thresholds tuning base on real data
Optimize the convection detection by testing their
size expanding rate, colder pixels increase rate ...
Combine to use partial filter and overlap method for
multi targets tracing according to the combined
observation mode of FY-4A
BJT 08,19 09 10 11 12,13 14 15,16 17,18
Param(k) 1.4 1.3 1.0 0.8 0.6 0.7 0.8 1.0
The k values vs Beijing time(BJT)
Ac=A*(sec θ)k
A. Sun zenith angle correction
Without sun zenith angle correction With sun zenith angle correction With sun zenith angle correction but k=1
B7 B8
B2 B1 B3 B4
B5 B6
B9 B10 B11 B12
B13 B14
Tuning include:
BTD=6.2µ-10.8µ
BTD=6.2µ-7.1µ
BTD=8.5µ-10.8µ
BTD= (8.5µ-10.8µ)-(10.8µ-
12.0µ)
BTD=12.0µ-10.8µ
BTD=13.5µ-10.8µ
B. Multi channel thresholds tuning base on real data
C. Choose one of the two
tracing methods
according to the
combined observation
mode
Baseline observation every hour one
FD(15min)
Every 3 hours , two more FD observation(15
min),Deriving AMV
During 17-19 (BJT), AGRI is suspended to
ensure its safety.
All the other time RRS (5min*9=45min)
15min FD 5min RRS AGRI combined observation
Convective Initiation(pink) and Possible Convective Initiation(blue)
Summary for the FY-4A convection product
The FY-4A convection product can provide convective initiation and
rapid convection growing information to end-user, it is good used for
nowcasting
The FY-4A convective initiation results have false alarms, many CI
detections do not produce severe weather
To distinguish the convective cloud from non convective frontal cloud
system is still a challenge, but this is quite important during summer
rain season in China
1. GIIRS:BT animation of different layers in troposphere for China area
He
igh
t (k
m)
Courtesy of Han Wei
Skew T-lnP diagram over Fangshan, Beijing
during 00:00 UTC-12:00 UTC (08:00-20:00 LST)
02 Aug 2017
Convective Available Potential Energy (CAPE) map during
00:00 UTC-12:00 UTC (08:00-20:00 LST) 02 Aug 2017 (grey
to white areas represent cloud observed by FengYun-4 satellite;
the asterisk denotes the location of Fangshan, Beijing)
Fangshan, Beijing:24h precipitation 111.9mm
Horizontal distribution of a) K index; b)
Showalter index; and c) Lift index at 09:00
UTC (17:00 LST) (2hrs before the occurrence
of the rainstorm over Fangshan) (grey to white
areas represent clouds observed by FengYun-4
satellite)
(a) (b)
(c)
2.LMI:Dynamic Distribution of Lighting
A typical thunderstorm occurred in West Australia during 13 February, 2017
Lightning Event detected from FY-4A LMI
LMI lightning events about 3 hours, is displayed
over the LMI background image in June 5, 2017.
Red color indicates lightning events. The
brightest storm system is located in the south of
the Yangtze River.
Distribution of number of lightning events,
group and flash about 3 hours with 0.5º× 0.5º
resolution in June 5, 2017.
Brighter colors indicate more lightning events,
group and flash was recorded.
event
group
flash
Severe Weather,June 28th to 29th,2017
2017.06.28 (UTC) 2017.06.29 (UTC)
LMI lightning events distribution,is displayed over the AGRI 11 μm channel image in
June 28th to 29th,2017.
Red color indicates lightning events, moving with convective storms in the southern China.
AGRI AGRI+LMI AGRI+LMI+GIIRS AGRI+LMI+GIIRS+ NWP AGRI+LMI+GIIRS+ NWP+Radar or GPM…
Question:
How to combine use these new data to better identify the convective activities?
How to get added value information of severe and high impact weather, more
easily and more quickly?.
Future Plan
To use machine learning and deep learning to develop new satellite convection
products in next few years.
It needs well organization. A joint working group will be established with NSMC
scientists, AI scientists and forecasters to develop new algorithms and products
Machine learning and deep learning such new technology will introduce to FY-
4B/C satellites to generate better products and better applications.
Motivation
Focus on those RDC connected to severe storm or high impact weather .
Three modules are included in this system :
RDC Training module ( Sensitivity analysis of interest fields constructed by FY-
4/AGRI、H8/AHI channels & thresholding training for RDC);
RDC Project Running module ;
RDC product validation module。
A. New Rapid Developing Convective Clusters (RDC) Algorithm-RDC v2
RDC Training module
(In prepare)
B. Machine Learning for predicting Convective Storm and QPE of FY-4
Data
• NASA GPM IMERG 0.1°*0.1 ° grid data in a half hour resolution
(Truth for training)
• FY-4A/AGRI or Himawari-8/AHI FullDisk infrared band
measurements (TBB) FY-4A/AGRI uses 6 infrared bands or
Himawari-8/AHI uses 9 infrared bands observations for training the
model
• Numerical Weather Prediction (NWP) data (GFS 0.5°*0.5 ° /Grapes
0.25°*0.25 ° )
• Surface ancillary data (i.e., elevation, surface type)
NWP data
Global Forecast System
Methodology
• Track Convection Cells
• Co-locate GPM data, FY-4A/AGRI or Himawari-8/AHI data, and NWP data
• Extract some useful samples from matched dataset
• Train classification and regression models for predicting Convective Storm and QPE
based on Machine Learning
• Predict Convective Storm and QPE using real-time FY-4A/AGRI or Himawari-8/AHI
and NWP data and models
Rank of predicted factors
rank name score 1 dtb62max 0.10849001
2 ch13 var min 0.102537347
3 dtb73max 0.094444007
4 dtb70max 0.088137094
5 dtb96max 0.080524218
6 ch16-ch13max 0.07700988
7 area 0.063007372
8 dtb96mean 0.029540668
9 ch13 var mean 0.026863466
10 ch14-ch15min 0.01872382
11 dtb86min 0.014983453
12 dtb12min 0.011430704
13 dtb12max 0.010817736
14 dtb86max 0.009642071
15 dtb11min 0.009274285
16 ch11-ch14max 0.008871846
17 dtb11max 0.008870689
18 ch16-ch13 min 0.008594803
19 dtb62mean 0.008255241
20 dtb70mean 0.008029407
21 dtb73mean 0.006768807
22 div850max 0.005789622
23 ch13 10per warm mean 0.005747327
24 thtse925min 0.005490885
25 dtb73min 0.005393001
Random Forest
n_estimators=100
max_depth=10
max_features=10
Training sample
date: April-October,2016
total 389315 convection cloud system
Case ——07/29/2016
Observation at 1200 UTC,precipitation 16 mm/h
RF,0700 UTC LR,0900 UTC NB,1030 UTC
Courtesy of Min Min
Result TBB at 11μm
Ground rainfall
observation
Machine Learning for Predicting Convective Storm and QPE by FY-4 Data
prediction
GPM IMERG observation
Summary
Every day the forecaster has to face torrent of data from satellite, surface,
balloon, radar, lightning…. How to use these data to forecast the severe
weather is a challenge. How to pick out the valuable information from big data
automatically, more quickly and more easy to use is also an other challenge.
We have to look for new approach to solve these problems, and machine
learning and deep learning technique show great potential benefit for
applications, we need prepare for that.
Thanks for your attention!