Climate change Climate change and and tropical total lightning tropical total lightning 1 , 2 , D. Buechler 3 , R. Albrecht 1 , W. Petersen W. Petersen 2 , D. Buechler 3 , R. Albrecht S. Goodman 4 4 , R. Blakeslee , R. Blakeslee 2 2 , H. Christian , H. Christian 3 3 S. Goodman 1 1 CICS/ESSIC University of Maryland, College Park, MD CICS/ESSIC University of Maryland, College Park, MD 2 NASA Marshal Space Flight Center , Huntsville, AL 2 NASA Marshal Space Flight Center , Huntsville, AL 3 University of Alabama in Huntsville, Huntsville, AL 3 University of Alabama in Huntsville, Huntsville, AL 4 NOAA/STAR-NESDIS, Camp Springs, MD 4 NOAA/STAR-NESDIS, Camp Springs, MD 2009 AGU Fall Meeting 2009 AGU Fall Meeting
Motivation Evidence of global temperature increase and precipitation increase in the IPCC models: they agree on increasing temperature with increasing CO 2 concentrations; they agree on increased accumulated precipitation by the heavy and very heavy events. Temperature warming by scenario Precipitation anomalies by extreme events IPCC (2007)
Motivation On the average precipitation IPCC (2007) shows conflicting responses over the 3 main convective tropical regions: +60 mm/yr in Southeast Asia/Maritime Continent Increase in East Africa, but decrease in West Africa -30 mm/yr in South America Moreover, recently the Tropical Measuring Mission (TRMM) revealed that on a regional annual mean scale less precipitation implies more lightning : a paradox (Price, 2009; Takayabu, 2006; Petersen and Rutledge, 2001).
Motivation 11-years of TRMM measurements (1998-2008)
Objective The goal of this study is to investigate the lightning trends from the 11-years of the Lightning Imaging Sensor (LIS) onboard of TRMM satellite.
Data and Methodology Flashes from LIS orbit dataset at 0.5 o resolution: Post-boost orbits field of view were corrected by the pre-boost swath and LIS detection efficiency; Cummulated daily flashes and viewtimes using a 49-day moving window to capture a full diurnal cycle (Boccippio et at., 2000); Cummulated flash rate density pentads (5-days).
Data and Methodology Flashes from LIS orbit dataset at 0.5 o resolution: Post-boost orbits field of view were corrected by the pre-boost swath and LIS detection efficiency; Cummulated daily flashes and viewtimes using a 49-day moving window to capture a full diurnal cycle (Boccippio et at., 2000); Cummulated flash rate density pentads (5-days). Quantile linear regression calculated for Regional Boxes:
Data and Methodology Quantile linear regression: Method to estimate the change (trend) of flash rate density (FRD) quantiles as a function of the year; A quantile is a point taken from the inverse cumulative distribution function of the FRD so that, for examples, the 0.7 quantile is the value such that 70% of the pentad FRD have FRD below this value (70 th percentile); A linear quantile regression model (Koenker and Bassett, 1978) assumes that the regressand y (in our case FRD pentads) is linearly dependent on K explanatory variables, and the τ th quantile of the error term ε τ (t) = 0: K y t = 0 t ∑ k = 1 k x tk t Q ∣ x t1 , ... , x tk = 0 K Q y t ∣ x t1 , ... ,x tk = 0 ∑ k = 1 k x tk
Data and Methodology The coefficients β k ( τ ) are estimated for 19 different quantiles ( τ =0.05,0.10,...,0.95) using each time the entire dataset of a regional box. All statistics were performed using the software R and its quantile regression package quantreg (Koenker, 2009).
Results South East United States + Golf of Mexico (Land):
Results South East United States + Golf of Mexico (Land):
Results South East United States + Golf of Mexico (Land):
Results South East United States + Golf of Mexico (Ocean):
Results Southern South America (Land):
Results Central Africa (Land):
Results Maritime Continent (No Mask):
Results Summary for Q(0.95): Summary for Q(0.95): Circles = Land trend (or No Mask) Triangles = Ocean trend Blue = negative trend on the 95 th percentile positive trend on the 95 th percentile Red = positive trend Red ( only trends with 95% confidence is shown)
Results When looking for trends by season...
Discussion and Conclusions 11-years is a very small period be considered a climate trend. But , if it is a sign of global change, how can we explain a decrease in the high flash rate densities?
Discussion and Conclusions Decrease in the convective mass flux (Betts 1998; Held et al., 2006; Vecchi and Soden 2007): Following Clausius-Clapeyron (C-C): d ln e s L = 2 ≡ T dT RT where α ( T ) ≈ 0.07 K -1 , that is: e s ↑7% for each 1-K increase in T The global-mean precipitation P is given by the convective mass flux M c and the typical boundary layer mixing ratio q : P = M c q and following C-C: M c = P P − 0.07 T M c
Discussion and Conclusions All IPCC AR4 models show a 10-20% decrease in mass flux by the year 2100: Vecchi and Soden (2007) Decrease in convective mass flux could be interpretated as decrease in updrafts, decrasing clould electrification.
Discussion and Conclusions Moreover, in warmer climates, IPCC AR4 models projected elevated cloud base heights (CBHs): Yoshida et al (2009) presented that theoretically there should be a fifth power relationship for lightning activity and cold-cloud depth (D): NSFC ∝ dQ 5 dt = B ' ' D where: NSFC = number of lighting flashes per sencond per convective cloud dQ/dt = charging rate B'' = constant and scale independent D = distance between freezing level and cloud top (cold-cloud depth). ↑CBHs → ↓D → ↓NSFC and lightning.
Discussion and Conclusions OR some instrument limitations: If convective strength is increasing and therefore there is also an increase in the cold cloud depth: Thick cloud depths (>13km) decreases the LIS detection efficiency for flashes. More investigation on the causes of negative flash rate density trends around the tropics is needed, as well as its the interannual variabilty (seasons).
Recommend
More recommend
