Lightning observations from space: Time and space characteristics of optical events. Ullrich Finke, FH Hannover 5 th December, 2007

Transcription

1 Lightning observations from space: Time and space characteristics of optical events Ullrich Finke, FH Hannover 5 th December, 2007

2 Contents 1. Lightning Imaging Mission 2. Optical characteristics 3. GEO-Orbit 4. Benefits and applications 5. Conclusions

3 The Lightning Imager Mission (LI) Lightning Imager: is dedicated to the detection of lightning events Lightning: as an event is clearly defined but not well quantified Lightning is a short transient event with high dynamic range and variability lack of knowledge about radiance function Users need total lightning information from GEO for nowcasting of severe storms (flooding, hail, gusts, tornadoes) air and ground strike warning, wild fire early detection global and regional climatology, NOx-chemistry improvement of NWP-models (ice flux, latent heat release) Feasibility of space based detection Prototypes on low orbit satellites (NASA) successfully employed for 10 years a lightning detector is in preparation on GOES-R

4 Lightning and its optical radiation Lightning lightning is always initialised in the cloud - intracloud lightning (IC) some IC propagate towards ground cloud-to-ground stroke (CG) IC + CG = TL (Total Lightning) each lightning flash consists of several return strokes with with inter-stroke time ms Optical radiation Optical radiation of energy in pulses from the hot lightning channel scattering transports light to cloud surface results in pulse time delay, broadening and source area enlargement

5 Lightning observed from above clouds Lightning seen from space shuttle Notice elliptical, circular shape, many scales multiple strokes illuminate the same cloud area radiation intensity not well detected by video-cam (saturation, blooming, glowing)

6 Challenges of the LI design lightning is a short transient signal on bright background large dynamics in radiance high spatial resolution required Optimization of detection parameters frame integration time ~ 2ms increase to avoid energy splitting decrease to minimise background radiation and detect sub-flash structure and pixel size ~ 8km increase to gain more energy decrease to enhance spatial resolution, minimise background radiation and detection threshold increase to minimze false alarm rate decrease to enhance detection efficiency

7 Lightning and Background Radiance Background radiance is higher than lightning radiance (daytime) originates from clouds (solar reflection), specular reflection (solar glint), ice, snow but lightning is a short transient signal (<2ms) can be discriminated by on-board processing detecting fast changes

8 Optical spectrum Uman, 1986

9 Optical spectrum of lightning Ground based observations (Orville & Henderson, 1984) strong emission line in NIR at nm, contains 5% of the total optical energy

10 Prototype: NASA Lightning Imaging Sensor: LIS on board of TRMM optical detection CCD-array narrow optical bandpass filter 777 nm detection of transient signals against slowly changing background Parameters pixel array pixel footprint Field of view Location viewtime frame integration time Orbit inclination Operation time LIS 128 x km 670km x 670km 92 s 2ms ? proven feasibility, development of algorithms and concepts

11 Example LIS data contiguos observation time for any location on ground ~90s

12 Global Distribution of Lightning Activity Goodman et al., Glimpses of a Changing Planet, M. King, ed., Cambridge University Press Mean annual global lightning flash rate (flashes km -2 yr -1 ) derived from 8 years (Apr 1995 to Feb 2003, data from NASA OTD + LIS instruments)

13 Characteristics of the optical lightning signal Described by statistical distribution of 1. Occurence of lightning in time and space lightning frequency, density flash-duration, strokes per flash 2. Single lightning pulse parameters pulse shape (duration, rise-, decay time) spatial pattern (size, shape) radiation amplitude Methodology: use available data from above clouds lightning observation: - from airplanes (U2, Altus), low orbit satellites (LIS, OTD, Forte) determine the empirical distribution functions for the signal parameters

14 1. Statistics of lightning pulse occurence (LIS-data)

15 Temporal characteristics: lightning pulses per second distribution of the number of lightning pulses per second 90% of non-empty seconds have less than 50 pulses per second mean lightning rate s -1 for the LIS - field of view extrapolated to full disk ~ 40 s -1

16 Spatial distribution: footprint area of pulses footprint area for lightning pulses mean is 92 km 2 ~ 10km x 10km 20% of the pulses have minimal footprint area

17 Temporal distribution: Duration of the flashes single stroke flashes multi-stroke flashes 14% with minimal duration (single strokes) broad distribution within 1s (mean 0.3 s for multiple strokes)

18 Lightning strokes per flash mean number of 11.6 pulses (lightning strokes) per flash 65% of the flashes consist of more than 5 single pulses

19 Auto-correlation of lightning distribution long pulses multi-strokes auto-correlation = frequency of distances and time intervals between lightning events LIS pulse lengths <5ms strokes msec spatial dimension: <15km observation time ~ 2min

20 2. Optical characteristics of single lightning pulses

21 Optical characteristics of single lightning pulses Spatial-temporal characteristic of a single lightning pulse in space in time Characteristics are: size geometrical shape radiance distribution pulse width rise time in relation to detector parameters

22 Lightning pulse transformation: source detector Detected energy from lightning is = Integrating the radiation function convolution with instrument function e = ik, m ( i+ 1) a ( k + 1) a ( m+ 1) τ = 2 2 f ( x + y ) dxdy f ( x 2 ia + y 2 ka mτ ) Π( x, y) p( t) Π( t) p( t) dt

23 Model of the radiation function for single event: F(r,t) = A T(t) R(x,y) 1. T(t) pulse function 2. R(x,y) spatial distribution 3. A amplitude Parameters of the radiation function derived from observational data statistics: Photodiode on Forte-satellite and Altus-aircraft LIS data with 2ms, 6km resolution and integrated energy

24 Temporal pulse function (FORTE) optical pulses (Light et al. 2003) Notice: shape close to symmetry around peak (A,B,C,D,G,I) optical pulses of a multi-stroke flash (Light et al. 2003)

25 pulse function (Altus aircraft) Altus - unmanned aircraft with instruments for electric. measm. Photodiode: broadband (vis), large angle (80 ) during ACES-experiment (FL) observation statistics Altus and Forte yield: mean pulse width 650 µs mean rise time 250 µs

26 Pulse function: analytical expression Function shape: T ( t, τ ) = I 0 t 2 e t 2 / τ 2 observation: mean pulse width 650 µs mean rise time 250 µs Parameters: width ~ d and integral ~ I 0 are choosen according to observed statistical distribution generated as random variates

27 Spatial distribution: pixels per lightning pulse 1-pixel pulse 5-pixel pulse 4-pixel pulse mean number of 5 pixels for lightning pulse 27% in 1pix, 19% in 2pix pulses,...

28 Spatial pattern zoo 2-5 pixels 6-15 pixels >15 pixels

29 Discrete pattern Possible pixel pattern (polyominoes) in a square array

30 Statistics of pattern in LIS data Frequency distribution of pixel pattern in LIS-data Only compact, convex pattern are observed in real data Statistics for radiance weighted pattern: standard deviation is in x and y-direction axis ratio σ r = x 2 ( xi x) σ x = A σ y i 2 A i diameter d = σ + σ 2 x 2 y

31 Geometrical characteristics of the pattern axis ratio axis ratio circular pattern are most frequent larger pattern (blue) are even more circular

32 Pulse function: analytical expressions time function area function T ( t, τ ) = I 0 t 2 e t Pulse function: width, rise time 2 / τ 2 r R( ) R exp r 2 = πσ 0 2σ 0 Pattern function: diameter, position parameters determined from statistics, generated as random derivates

33 Distribution of energy (Altus) energy threshold for LIS: 3 µjm -2 sr -1 Energy vs. peak radiance

34 LIS: Radiance of lightning pulses from integrated energy over 2ms frame only local nighttime limited between detection threshold 3 µjsr -1 m -2 and saturation value most pulses with weak radiance, 64% of the pulses have radiance > 10µJsr -1 m -2

35 Maximum pulse radiance in flashes flashes consist of several pulses: 65% of the flashes consist of more than 5 pulses statistics of the brightest pulse per flash 90% of the flashes have pulses with radiance > 10 µjm -2 sr -1 important for required detection threshold

36 from FORTE, Altus, LIS Optical characteristics Summary pulse width function of source distance to cloud surface mean around µsec footprint area mean 92 km 2 small pattern are highly circular energy LIS: mean >50µJm -2 sr -1, median 17µJm -2 sr -1 65% of pulses with energy >10µJm -2 sr -1 90% of all flashes have pulses with energy >10µJm -2 sr -1

37 Summary lightning radiation function functions describing the optical lightning signal could be derived from the lightning statistics these functions contain random variates, which reflect the large variability of lightning parameters simple mathematical form of the functions allow for analytical integration the functions can be used for simulation of the lightning detection

38 Transition to GEO-field of view Observation geometry changes from LIS to GEO: LIS GEO altitude 400km 35,800km viewing angle from sat IFOV on Earth (ang. diameter) pixel footprint: ~ 8km depends on angle from nadir pixel matrix ~ 1300 x 1300 angular accuracy ~ 100 µrad continuous observation mean lightning rate extrapolated to full disk is ~ 40s -1

39 Field of View from GEO-orbit field of view shifted northward Field of view of lightning imagers on MTG and GOES-R East and West

40 Distortions with increasing nadir angle circular lightning clouds as seen from GEO: at large nadir angles, in the outer parts of disk: circles are distorted appear brighter

41 Stochastic Lightning generator stochastic simulation of the GEO-lightning signal using data statistics (geography, day, year) on 2 scales: storm-size and lightning-size Lightning is simulated as sum of storms each being gaussian cloud with size(t), moving center position (t), lightning rate(t) storm parameters randomly chosen according to the empirical statistics Each storm produces lightning as a cloud of photons of circular gaussian size and pulse shape and energy correlated in time and space satisfying the empirical statistics

42 Artifical Lightning Clouds lightning pulses storms

43 Simulation: example 3 hours of simulation UTC, June mean flash rate: 30s storms

44 Simulation: Example over Europe on scales 10 x 10

45 Advantageous characteristics of LI on MTG LI detects lightning worldwide detection over full disk with uniform detection efficiency detection of total lightning (Intracloud and Cloud-to-Ground) spatial resolution of 10 km optical radiance provided timeliness <1-2 min LI in combination to the other missions on MTG: provides continuously data in between the scan cycles qualifies cloud images detects convection directly benefits from cloud data to discriminate false alarms

46 Summary Optical characteristics of the lightning signal as seen from above the clouds can be derived from satellite and aircraft observation data. Statistics can be extrapolated to the GEO-field of view Lightning can be simulated basing on the empirical statistics and conceptual models.

47 End