AOSC 620: Special Lecture Lightning R. Dickerson North Dakota Thunderstorm Experiment 1 Lightning Nomenclature Flash – the entire lightning process whether staying within the cloud (intra-cloud) or striking the ground (cloud-to-ground). Stroke – cloud-to-ground components only. Leader – discharge creating a conductive channel. Corona – composed of streamers and limited to the region around an electrode. ~ambient temperature. Plasma – gas with >1% ions or electrons. 2 >90% Thanks to M. Uman Rare but powerful 3 4 5 Global electric circuit – classic view In fair weather the current is downward (positive ions down and neg ions or electrons up) with a global current of ~1kA. Thunderstorms induce a current of neg charges downward (lightning and precp) and positive charges upwards from the tops of clouds. 6 7 Basics of negative cloud-to-ground lightning Stepped leader Length of steps ~50 m Mean current 100-200 Amp First return stroke Peak current 30kA Radius 1-2 cm T ~30,000 K Overall Flash Duration ~250 ms Strokes/flash 3-5 Charge transfer ~ 20 C Energy 109-1010 J 8 Impact of Lightning-NO Production on Tropospheric Photochemistry and Radiative Forcing Dale J. Allen University of Maryland College Park, MD 20742 With contributions from: K. Pickering, NASA-GSFC H. Huntrieser, DLR E. Bucsela, SRI R. Pinder, EPA M. Martini, PNNL C. Liaskos, A. Ring, & Y. Li, UMCP October 17, 2013 Photo of northeast Colorado STEREO storm Courtesy of K. Pickering Outline • Lightning formation and the resulting production of nitrogen oxide (LNOx) • Impact of LNOx on atmospheric composition and radiative forcing • Parameterization of LNOx production in global/regional models – Estimation of LNOx production per flash – Relation between flash rate and meteorological variables – Specification of vertical distribution of LNOx • Impact of LNOx on air quality and nitrogen deposition • “When LNOx is not enough.” – Why was UT NOx so high during the INTEX field campaign? – Why do models underestimate tropospheric NO2 columns at rural locations? Lightning initiation... Lightning is most common in clouds that contain both water and ice (-10 to -40°C) -40°C -10°C In these clouds, strong updrafts carry water droplets upward where they collide with ice crystals, forming graupel and transferring a positive charge to the ice crystals and a negative charge to the graupel. As the relatively heavier graupel falls, an increasingly negative charge builds up in the lower portion of the cloud until the electrical potential becomes sufficient to initiate a lightning discharge or leader from the lower portion of the cloud. This leader and subsequent branches may travel upward (intracloud flash) or downward to the surface (cloud-to-ground flash). O2 N2 N2 N2 N2 N2 O2 N2 O2 N2 O2 N2 N2 Air is ~78%N2 and ~21% O2 N2 + O2 NO + NO Zel’dovich mechanism (Zel’dovich and Raizer 1966) Lightning stroke high T & high P N2 + O ↔ NO + N (1) N + O2 ↔ NO + O (2) N2 + O2 ↔ 2(NO) (2b) Reaction rate (1) is a strong function of temperature. NO is “frozen out” when lightning channel cools rapidly. NO from Lightning Copyright © 2013 R. R. Dickerson & Z.Q. Li 13 Lightning, Reminder In the absence of industrial processes, lightning is a major source of odd nitrogen (NOx) and thus nitrate (NO3-) in the atmosphere. Even today, lightning is a major source of NOx in the upper free troposphere. N2 + O2 2NO How can this happen? Let’s calculate the Gibbs Free Energy for the reaction for 298 K and again for 2000 K. G° = – nRT ln (Keq) Gf° = Hf° 0.0 for N2 and O2 Gf° (NO) = 20.7 kcal mole-1 Hf° (NO) = 21.6 kcal mole-1 R = 1.98 cal mole-1 K-1 G° = 2*(20.7) = 41.4 kcal mole-1 Keq = exp (– 41.4E3/1.98*298) = 3.4 E-31 Copyright © 2011 R. R. Dickerson æ P2 ö Keq = ç NO ÷ è PN 2 PO2ø 14 G° = 2*(20.7) = 41.4 kcal mole-1 æ P2 ö Keq = ç NO ÷ è PN 2 PO2ø = exp (-G°/RT) = 3.4 E-31 Assume PN2 = 0.8 atm; PO2 = 0.2 atm PNO = - Keq × 0.8× 0.2 = 2.3E-16 atm (pretty small) Let’s try again at a higher temperature (2000 K). Remember H and φ are independent of temperature. GT ≈ H° – Tφ° 41.4 = 43.2 – 298 φ° φ° = +6.04E-3 kcal mole-1 K-1 GT ≈ 43.2 – 2000* 6.04E-3 = 31.12 kcal mole-1 Copyright © 2013 R. R. Dickerson & Z.Q. Li 15 G(2000) ≈ 31.12 kcal mole-1 Keq = = PNO2/PN2*PO2 = exp (-GT/RT) = exp (-31.12E3/1.98*2000) = 3.87E-4 If the total pressure is 1.00 atm PNO = 7.9E-3 atm = [0.79% by volume] You can show that the mole fraction of NO at equilibrium is nearly independent of pressure. Try repeating this calculation for 2500 K; you should obtain Keq = 3.4E-3 and [NO] = 2.3%. In high temperature combustion, such as a car engine or power plant, NO arises from similar conditions. Copyright © 2013 R. R. Dickerson & Z.Q. Li 16 Where does G(T) = 0? GT ≈ H° – Tφ° 41.4 = 43.2 – 298 φ° φ° = +6.04E-3 kcal mole-1 K-1 GT ≈ 43.2 – T* 6.04E-3 = 0 T ≈ 7000 K But at 7000K N2, NO, and O2, forced by entropy, all begin to decompose to their constituent atoms. Project left to the student: Calculate the Keq for NO = N + O At 7000 K. With 20% O2 in the atmosphere, the maximum [NO is ~10% Copyright © 2013 R. R. Dickerson & Z.Q. Li 17 BL Climatological OTD/LIS 0.5° x 0.5° Flash Rate Distribution (Flashes km-2 yr-1) Courtesy of GHCC lightning team Constructed using 1995-2000 data from the Optical Transient Detector (global coverage) & 2000-2011 data from the Lightning Imaging Sensor (Equatorward of 35°) [GHCC] Why is this LNOx important ? LNOx indirectly affects local air quality and global climate through its influence on ozone and hydroxyl radical (OH) [OH is often referred to as the troposphere’s detergent] NOx [NO+NO2]: • Is a primary pollutant found in photochemical smog • Is a precursor for tropospheric ozone formation TROPOSPHERIC OZONE: • Is the third most important GHG • Impacts the Earth’s radiation budget and can cause changes in atmospheric circulation patterns • Toxic to humans, plants & animals Photo courtesy of University of California at Berkley *(review by Schumann and Huntrieser, ACP, 2007) Global NOx sources [TgN a-1] 53* < 1 NOx = NO+NO2 30-60 (source: Lee et al., AE, 1997) < 1 Ozone production efficiency (OPE) in the PBL as a function of NOx mixing ratio assuming full oxidation of NMHCs [Liu et al., 1987] Note decrease in OPE for high values of NOx. Most LNOx production occurs at altitudes where NOx amounts are low and OPE is high. Therefore the LNOx source is more important than one might infer based on its % of the total NOx source. [Grewe, 2007] UT NOx in summer increases by 61-75% when LNOx is added Calculations use output from NASA GMI model and assume LNOx source of 500 moles per flash in extratropics [Allen et al., 2010] UT ozone over US in summer increases by 20-33% due to LNOx Contribution of lightning-NO to UT O3 increases from June-August and is especially large during August 2006 Calculations made using NASA GMI model and assume an extratropical LNOx source of 500 moles per flash NASA’s Chemistry and Climate Model (NASA CCM) • Global modeling system for evaluating the impact of anthropogenic and/or natural perturbations on atmospheric composition & climate. • Includes NASA GCM, NASA Global Modeling Initiative CTM, and GOCART aerosol module. It can be run in a free-running mode or in a replay mode where analysis increments are used to simulate particular time periods. • The GMI CTM, which can also be run off-line using met fields from the GEOS-4 or GEOS-5 GCM/DAS. It solves for the distribution of trace gases in the atmosphere using a stratospheric or combined stratospheretroposphere chemical mechanism. • The GOCART aerosol module can also be run off-line or within the CCM framework. When combined with the GMI CTM, it can be used to evaluate the impact of aerosols on atmospheric chemistry and climate. • In order to investigate the sensitivity of atmospheric composition to LNOx production, 4 “replay” CCM simulations (no LNOx, HalfLNox, stdLNOx , and DoubleLNOx.) were run for 2006 & 2007. Magnitude of LNOx source in these runs was 0.0, 2.5, 5.0, and 10.0 Tg N / yr corresponding to ~0, 125, 250, and 500 moles per flash. • We begin by looking at the impact on ozone and OH. Zonally-averaged ozone and OH for September 2007 as a function of magnitude of LNOx source Upper-tropospheric ozone (OH) amounts in the tropics increase by nearly 50% (100%) as the global LNOx source is increased from a lower bound of 125 moles per flash (“Half LNOx”) to an upper bound of 500 moles per flash (“Double LNOx”). C. Liaskos et al. (JGR, 2015). Change in zonally-averaged ozone relative to standard simulation Change in zonally-averaged OH relative to standard simulation Sensitivity of radiative forcing from ozone to magnitude and spatial pattern of lightning-NO production (NASA-GSFC CCM December 2007) 2.5 Tgs 5.0 Tgs Co-located With Climatological LNOx source 5.0 Tgs Co-located With Convection 10.0 Tgs Note: RF by NO2 itself is relatively minor due to its short lifetime. Changes in RF from methane due to LNOx are not considered here. Instantaneous radiative forcing (IRF) from ozone due to North American (NA) anthropogenic emissions (left) and NA lightning-NO production (right) for Jun 1-Jul 17 (top) and Jul 18-Aug 31 (bottom) of 2004. Calculated by Martini et al. [2011] using University of Maryland CTM (UMD-CTM). Ratio of IRF from ozone due to North American LNOx vs. IRF from ozone due to North American anthropogenic emissions IRF ratios calculated using UMD-CTM for Jun 1 – Jul 17, 2004 [Martini et al., 2011] Requirements for Specifying Lightning NO Production in Global/Regional Chemical Transport and Climate Models 1) Estimates of NO production per IC and CG flash (DeCaria et al., 2000; 2005; Ott et al. (2007; 2010) 2) Flash rate distribution as a function of space and time that is consistent with the driving model’s convection (e.g., Allen and Pickering, 2002; Allen et al., 2010). 3) Vertical distribution of LNOx at the end of storm (e.g., Pickering et al., 1998; Ott et al., 2010). Table 1. Sample of previous investigations of lightning NOx production for individual storms Method Moles NO/flash (Notes) Reference Theoretical Laboratory Theoretical/lab/ltng. obs. Aircraft data, cloud model Aircraft data, cloud model Aircraft data, cloud model Aircraft data, cloud model Aircraft data Aircraft data Aircraft data Satellite (GOME) Satellite (SCIAMACHY) Satellite (OMI) Satellite (OMI) Satellite (OMI) 1100 (CG), 110 (IC) 103 484 (CG), 35 (IC) 345-460 (STERAO-A) 360 (STERAO-A, EULINOX) 590-700 (CRYSTAL-FACE) 500 (Mean midlatitude) 500 – 600 (SCOUT-O3/ACTIVE) 70-210 (TROCCINOX) 121 – 385 (SCOUT-O3/ACTIVE) 70 – 179 (AMMA) 32-240 (Sub-Tropical) 33 – 50 (global, mostly marine) 87-246 (TC4 – tropical marine) 174 (TC4 mean) 440 (Central US, Gulf of Mex.) 72 – 92 (DC3) Price et al., 1997 Wang et al., 1998 Koshak et al., 2013 DeCaria, et al., 2005 Ott et al., 2007; 2010 Ott et al., 2010 Ott et al., 2010 Cummings et al., 2013 Huntrieser et al. 2008 Huntrieser et al., 2009 Huntrieser et al., 2011 Beirle et al., 2006 Beirle et al., 2010b Bucsela et al., 2010 Bucsela et al., 2010 Pickering et al., in prep Pickering et al., in prep Specification of NO production per flash Summary of Five Midlatitude and Subtropical Storms Orville et al., 2002 Means: 500 moles/flash 0.94 ratio For global rate of 44 flashes/sec, this implies ~9 Tg N/yr (a bit high) Ott et al., 2010, JGR OMI-based estimation of LNOx production per flash 1. 2. 3. Use back trajectories to determine if enhanced OMI NO2 columns can be associated with upstream flashes. Portion of column due to LNOx (VLNOx) is determined by subtracting the sum of the measured stratospheric slant column and estimated tropospheric background slant column from the measured total slant column and dividing the resulting “lightning” slant column by an air mass factor (AMF) appropriate for a region impacted by convection. Symbolically, VLNOx = ( Stotal – Vstrat ·Astrat – Vtrop_background ·Atrop ) / ALNOx Production per flash can then be estimated by dividing the moles of NO due to LNOx by the number of upstream flashes. • Note: AMFs for NO2 are determined using Goddard Modeling Initiative (GMI) CTM output and the TOMS radiative transfer model (TOMSRAD). • Note: The tropospheric background is estimated using OMI measurements from surrounding days with minimal lightning activity. OMI LNOx for May 30, 2012 during DC3 field mission. Large maximum is evident over southern Appalachians downwind of previous day’s storms over Oklahoma [14.0 Mmoles LNOx within box] Back trajectories suggest transport times of 9 – 18 hours; mean = 16.1 hours Total (CG + IC) contributing flashes along trajectories = 181,743 1.4 x 107 moles/181,743 flashes = 77 moles LNOx/flash Example over the Gulf of Mexico “Observed” flashes (World Wide Lightning Location Network, WWLLN, after adjusting for detection efficiency) during 6-hrs before OMI overpass OMI-based estimate of lightning-NO production per flash Cloud radiance fraction OMI sees enhanced LNOx over western portion of cloud system but low values over eastern portion Remaining Issues • VLNOx distribution is usually noisy and occasionally negative in regions where OMI shows high cloud radiance fractions (CRFs). Therefore, we will try screening out regions with CRF > ~50%. • Flash rate detection efficiencies from the World Wide Lightning Location Network (WWLLN) are 5-15% over much of the globe including Africa where flashes are most plentiful. Therefore, we are investigating other lightning networks (ENTLN, GLD360) and focusing on regions with DEs greater than 15%. • Tropospheric background is difficult to pin down. Is it best defined using OMI data alone or is model output needed (ratios of simulations with and without LNOx)? What is the impact of day-to-day variation in biomass burning on this background? LNOx production per flash is sensitive to environment; Storms with substantial wind shear produce more NO per flash (e.g., tropical (low-shear) vs non-tropical systems) c) 04 February 2005, 18 UTC Wind 200 hPa (~12 km) Tropical d) 18 February 2005, 21 UTC Subtropical Plot taken from Jan 2011 AMS presentation by Heidi Huntrieser NO production per flash varies with flash component length which in Comparison of "flash component" lengths in tropical and subtropical TS turn varies with vertical wind shear (Huntrieser et al., ACP, 2008) 2x larger 1.6 km 3.1 km TROCCINOX: subtropical flash component length = 2x tropical Institut für Physik der Atmosphäre Flash rates in the NASA GMI model are parameterized in terms of UT mass flux. How well does this approach capture observed flashes? Number to right of year shows total flash rate (flashes s-1) over region shown. Spatial distributions of “observed” flashes are well captured by mass-flux based parameterization. Magnitudes ~20% low. Vertical distribution of LNOx production is expected to show mid-trop peak due to CG flashes and UT peak due to IC flashes. Here is the LMA-based distribution that was used as a guide for partitioning LNOx production in CMAQ Segment altitude distribution of CG+IC flashes in vicinity of North Alabama Lightning Mapping Array (LMA) [Koshak et al., 2010] EPA’s /Community Multi-scale Air Quality (CMAQ) CTM • Regional photochemical model that solves the constituent continuity equation. In this study, it was run offline using meteorological fields from WRF. • CMAQ was originally designed to help regional air quality planners predict the distribution of surface layer ozone and particulate matter. • • These simulations use the CB05 chemical mechanism. Calculates biogenic emissions of isoprene and monoterpenes using BEIS. • The lightning-NO algorithm used in these simulations was released operationally in March 2011. • Simulations of 2006 air quality performed at EPA as part of this NASA Air Quality Decision Support Project and the AQ Model Evaluation International Initiative (AQMEII) http://aqmeii.jrc.ec.europa.eu/doc/AQMEII_activity_description.pdf • These 2006 simulations use chemical boundary conditions from GEMS (European-led assimilation effort) http://ozone.meteo.be/meteo/view/en/1550484-GEMS.html • Bias between modeled and OMI tropospheric NO2 column for CMAQ simulations without (top) and with (bottom) LNOx. Adding LNOx reduces mean negative bias for summer 2006 from 35-45% to 10-20% Comparison uses OMI DOMINO product However, an interesting urban/rural bias remains … Even after adjusting the OMI vertical column using a priori information from CMAQ, CMAQ has a high-bias over urban areas and a low-bias over rural areas (summer) . Low-bias over rural areas suggests that the chemical lifetime of NO2 [Henderson et al., 2011; Canty et al., 2015] and/or the medium-range transport of NOx [Gilliland et al., 2008] is underestimated. Of course bias is also sensitive to magnitude of mostly rural LNOx & soil-NOx. Does LNOx affect air quality? Mean summer 2006 enhancement of 8-hr maxO3 due to LNOx O3 enhancement (ppbv) calculated using CMAQ assuming 500 mole per flash source) Impact of LNOx on wet deposition of nitrate Adding LNOx eliminates low-bias Adding LNOx & adjusting for precip bias leads to better fit Adjusting for precip Bias leads To better fit Eastern US: Longitudes east of 100W How well does CMAQ simulate the wet deposition of nitrate (OxN) and ammonium (redN) as measured by the NADP network? oxN Biases NE: +5% SE: +19% West –(5-10)% Note: Adding LNOx causes Wetdep(OxN) 50%↑ Totdep(OxN) 22%↑ redN biases NE: +20% SE: +40% West: -(20-25) Totdep(N) 11%↑ How much do uncertainties in chemistry contribute to models’ inability to capture the high amounts of UT NOx measured during INTEX-A (An upper bound) • CMAQv4.7.1 with CB-05 chemistry and AERO5 aerosols was used to simulate the summer of 2004. • Three simulations: 1) standard chemistry without lightning-NO, 2) standard chemistry with lightning-NO, and 3) updated chemistry with lightning-NO • Updated chemistry: Main change: Organic nitrate (ON) yield from the oxidation of paraffins (PAR) was reduced from 15% to 3%. This decrease in ON production reduces NO consumption, increases the NOx lifetime, and is in better agreement with observations (Henderson et al., 2011). Comparison of GMI-calc NOx with INTEX-A DC-8 Flight 5 GMI with No Light GMI with lightning Lightning-NO emissions appear responsible for observed 16 EST peak in UT NOx; However, magnitude of peak is underestimated by GMI simulation Comparison of GMI-calc ozone with INTEX-A DC-8 Flight 5 GMI with No Light GMI with Lightning Lightning-NO emissions appear responsible for observed 16 UT peak in ozone. Vertical extent of peak is overestimated by GMI simulation. Comparison of GMI-calc NOx profiles with INTEX-A profiles GMI-calc NOx profiles with or without lightning-NO greatly underestimate observed UT NOx amounts. Hudman et al. [2007] found a similar bias using GEOS-CHEM. 250 mole per flash lightning source used here Lightning-NO increases UT NOx by 75-100 pptv (X4 increase) Adjusting chemistry increases UT NOx by 20-30 pptv reducing model biases by 5-16% if data are unbiased and by 10-33% if data are assumed to be 30% too high. Simulation could be re-run using chemistry modifications from Canty et al. (2013). However, a 120-500 pptv bias remains (60-300 pptv if assume MPN interference) NO2 measurements taken by R. Cohen. Black line with asterisks shows measured mean assuming 30% interference from MPN [Browne et al., 2011] Comparison of GMI-calc ozone profiles with INTEX-A profiles GMI-calc O3 profiles for sims with lightning-NO agree reasonably well with INTEX-A O3 profiles despite the low-bias in NOx. Summary • Upper tropospheric ozone (OH) amounts in the tropics increase by nearly 50% (100%) as the global LNOx source is increased from a lower bound of 125 moles per flash to an upper bound of 500 moles per flash. • Cloud-to-ground and intracloud flashes produce ~same amount of NO per flash. However, flashes from storms with midlatitude characteristics usually produce more NO than storms with tropical characteristics possibly due to wind-shear driven differences in flash channel length. • LNOx is responsible for 60-75% (20-33%) of upper tropospheric NOx (ozone) over the United States during the summer time. Therefore, it must be included in simulations when comparing with satellite data; however, the magnitude of the LNOx source is likely too low to resolve model/OMI differences in tropospheric NO2. Look to the chemistry. Summary (continued) •Observations of LNOx from OMI can be used in conjunction with flash rate information to reduce uncertainties in production per flash. The method must be used with caution due to uncertainties introduced by retrievals in partly cloud scenes, biomass burning, and relatively low detection efficiencies. •Lightning-NO must be included in simulations that evaluate wet deposition. Over the U.S., it increases wet deposition of nitrate by 50% and total deposition of nitrogen by 11% in CMAQ. It also changes a 30% low-bias with respect to NADP deposition to a 2% high-bias. •Ozone with a LNOx source does reach the surface, especially over the southwestern U.S. where boundary layers are high. According to CMAQ, on poor AQ days (8-hr max ozone>60 ppbv), LNOx contributes >6.5 ppbv to 8hr-ozone at 10% of western sites and 3% of eastern sites •Uncertainty in the LNOx spatial distribution and magnitude, contribute to uncertainties in the RF from ozone. Additional simulations that examine the methane RF for different LNOx sources are warranted.