Global Fire Emission Monitoring with Fire Radiative Power in the MACC Project

Transcription

1 Global Fire Emission Monitoring with Fire Radiative Power in the MACC Project Johannes W. Kaiser, A. Heil, M.G. Schultz, G.R. van der Werf, M.J. Wooster, W. Xu, more MACC partners and Gareth Roberts Russia, August 21 LandSAF User Workshop, 15 Nov 21 Kaiser 1

2 GMES Atmosphere Component Service Part of Europe s Global Monitoring for Environment and Security initiative development of operational space-based observation strengthening of complementary in-situ observing systems development and operation of associated data and information services, based on core integrated assimilation and forecasting Three environmental services for Land, Ocean and Atmosphere A 48-partner EC-funded project called MACC: provides pilot GMES Atmosphere Component Service succeeds earlier projects GEMS and PROMOTE coordinated by ECMWF LandSAF User Workshop, 15 Nov 21 Kaiser 2

3 MACC data use & modelling validation data assimilation meteo data, CO 2, CH 4, CO, O 3, NO 2, SO 2, aerosol global and regional data and web products analysis & forecast CO LandSAF User Workshop, 15 Nov 21 Kaiser 3

4 Comparison of GEMS simulated and analysed aerosol optical depth with MODIS and MISR for July 23 Independent observations Assimilated observations Morcrette et al., 29; Benedetti et al., 29 LandSAF User Workshop, 15 Nov 21 Kaiser 4

5 Instrument Satellite Provider Version Species Type Period Status GOME ERS-2 RAL O3 Profiles MIPAS Envisat ESA O3 Profiles Active Active MLS AURA NASA V2 O3 Profiles Active OMI AURA NASA V3 O3 Total column Active SBUV NOAA-16 NOAA V8 O3 6 layer profiles Active SBUV NOAA-17 NOAA V8 O3 6 layer profiles Active SBUV NOAA-18 NOAA V8 O3 6 layer profiles Active SCIAMACHY Envisat KNMI O3 Total column Active GOME-2 METOP EUMETSAT O3 Total column/profile Passive MOPITT TERRA NCAR V4 CO Total column Active IASI METOP LATMOS CO Total column Active OMI AURA KNMI Col. 3 NO2 Tropospheric column Passive SCIAMACHY Envisat KNMI V1.1 NO2 Tropospheric column Active OMI AURA NASA V3 SO2 Total column Passive SCIAMACHY Envisat BIRA SO2 Total column Passive OMI AURA NASA V3 HCHO Total column Passive SCIAMACHY Envisat BIRA V2 HCHO Total column Passive MODIS AQUA NASA AOD Total column Active MODIS TERRA NASA AOD Total column Active SCIAMACHY ENVISAT SRON CH4 Total column Active TANSO GOSAT JAXA CH4 Total column Passive AIRS AQUA NASA CO2 Radiances Active IASI METOP EUMETSAT CO2 Radiances Active TANSO GOSAT JAXA CO2 Total column Passive LandSAF User Workshop, 15 Nov 21 Kaiser 5

6 3-day European Air Quality Forecast: NO2 (single member of the ensemble shown) MOCAGE LandSAF User Workshop, 15 Nov 21 Kaiser 6

7 modelled AOD observed FRP Modelled AOD of Greek Fire Plumes, August August 1: Emissions calculated from Fire Radiative Power observed by SEVIRI on Meteosat. Emission factors from Andreae & Merlet 21 and Ichoku & Kaufman 25. Run at 25km global resolution, which is typical for regional models. MODIS :5 LandSAF User Workshop, 15 Nov 21 Kaiser 7

8 Bottom-Up Estimation of Fire Emissions promising best accuracy: MACC real time sat. obs. E i = FRE x CF x EF i (Wooster et al. 24) E i = BA x AFL x CC x EF i (Seiler & Crutzen 198) burnt biomass (dynamic) vegetation model most established, in particular GFED (van der Werf et al. 26): MACC retrospective Ei = emission of species i [kg(species i)] BA = burnt area [m2] AFL = available fuel load [kg(biomass) / m2] CC = combustion completeness [kg(burnt fuel) / kg (available fuel)] EFi = emission factor for species i [kg(species i) / kg(biomass)] FRP = fire radiative power [W] FRE = fire radiative energy [J] = FRP(t) dt ~ const. CF = conversion factor [kg(biomass) / W(FRE)] land cover map LandSAF User Workshop, 15 Nov 21 Kaiser 8 key uncertainty (Reid et al. 29) graphics by M. Wooster

9 Conversion Factor Validation against GFEDv3 GFED3 DM (Tg DM month-1) SA 1 AG 5 TF EF SAOS AGOS PE EFOS MODIS-FRE (PJ month-1) SA LR: y=1.6x-9.7 R 2 =.81 LR: y=.8x R 2 =.76 AG LR: y=.2x+2.8 R 2 =.54 LR: y=.3x R 2 =.41 TF LR: y=1.3x-3.2 R 2 =.78 LR: y=1.3x R 2 =.78 EF LR: y=.6x-1.8 R 2 =.44 LR: y=.51x R 2 =.43 SAOS LR: y=.88x+.2 R 2 =.47 LR: y=.89x R 2 =.47 AGOS LR: y=1.6x-.16 R 2 =.76 LR: y=1.51x R 2 =.76 PE LR: y=9.4x-2.9 R 2 =.82 LR: y=9.4x R 2 =.82 EFOS LR: y=1.7x-1.1 R 2 =.88 LR: y=1.68x R 2 =.88 Heil et al. ECMWF 21 SA AG TF EF PE SAOS AGOS EFOS Conversion factor depends on land cover! FRP observations can reproduce GFED within its accuracy. lab:.368 kg DM / MJ GFASv: 1.37 kg DM / MJ LandSAF User Workshop, 15 Nov 21 Kaiser 9

10 Global Fire Assimilation System (GFASv) real-time production of fire emissions using FRP observation by MODIS & SEVIRI list of species: C, TPM, PM2.5, BC, OC, CO2, CH4, CO, SO2, NOx, NMHC used in MACC NRT production system: aerosols & greenhouse gases & campaign support (CO) Sep 29 Sep 21 LandSAF User Workshop, 15 Nov 21 Kaiser 1

11 Upcoming GFAS Version 1 observation gap filling with data assimilation land-cover specific FRP conversion.5 deg resolution GOES observations Terra MODIS # observations on 1 Oct GOES-West Imager GOES-East Imager Aqua MODIS Meteosat-9 SEVIRI LandSAF User Workshop, 15 Nov 21 Kaiser 11

12 Fires Diurnal Cycle in Americas LandSAF User Workshop, 15 Nov 21 Kaiser 12

13 GOES vs. MODIS Grid Based FRP Comparison LandSAF User Workshop, 15 Nov 21 Kaiser 13

14 More NRT FRP and AOD 21 Sep Sep Sep 29 Sep 21 SH American fire activity strongly enhanced in 21 (smoke) aerosols enhanced by factor ~3 LandSAF User Workshop, 15 Nov 21 Kaiser 14

15 21 Fires in Russia 2m temperature anomaly (K) for July 21 Fire Radiative Power (Wm -2 ) 26 July 21 ERA-Interim graphics by A. Simmons FRP on 4 August 21, graphics by S. Siemen The daily jump in the modeled smoke distribution is due to the assimilation of satellite information on the fires LandSAF User Workshop, 15 Nov 21 Kaiser 15

16 Russian Smoke Plume in Finland well Predicted Fire Radiative Power (4 Aug 1) Smoke AOD (8 Aug 1) in-situ PM1 observations (Virolahti, Finland) 1-day PM1 forecast 3-day PM1 forecast 4 Aug 11 Aug 4 Aug 11 Aug LandSAF User Workshop, 15 Nov 21 Kaiser 16

17 21 Fires in Russia: Burning throughout the Night! Jul-Aug according to FRP, probably smouldering sub-surface fires Conversion and emission factors need to be adapted dynamically. 8% peat fires 2% forest fires explain the CO and NO2 observations. LandSAF User Workshop, 15 Nov 21 Kaiser 17

18 Frequency [N grids] Gas Flaring Analysis of daily NRT MODIS-FRP data at.1 deg covering 23 to 29 Source identification based on the analysis of the stationarity and/or magnitude of the FRP signal, visual inspection using GoogleEarth/Bing Maps, MODIS CMG Land Cover Data, NOAA DMSP-OLS Gas Flaring Inventory 1,, 1, 1, 1, World s Top1 grid cells by FRE. Range:.1 to 1.48 W m-2 Histogram of time-averaged MODIS-FRP grids (in.1 Wm-2 bins) (r36x18 6,48, grid cells) MODIS-FRP [W m-2] 1.41 LandSAF User Workshop, 15 Nov 21 Kaiser 18

19 Contamination of MODIS-FRP by Volcanoes Center and approximate boundaries of.1 deg grid: W,19.35 N Histogram of time-averaged MODIS-FRP grids (in.1 Wm-2 bins) (r36x18 6,48, grid cells) Lawa flow 26 FRP [W m-2] W,19.35 N Top-1: Kīlauea Volcano, Hawai Mean FRP= 1.48 W m-2, CONSECTS=247 days Total dry matter burned equivalent: 12.5 kg m-2 or 14.1 Tg per.1 grid cell (compare: tropical forest~5 kg m-2 ) Daily mean 3-day running mean Linear Trend LandSAF User Workshop, 15 Nov 21 Kaiser 19

20 Contamination of MODIS-FRP by Gas Flares Center and approximate boundaries of.1 deg grid: E,48.25 N MODIS FIRMS last 48h FRP [W m-2] E,48.25 N Top-2: Gas Flares, Embenski area, Kazakhstan Mean FRP= 1.29 W m-2, CONSECTS=288 days Total dry matter burned equivalent: 15.1 kg m-2 Top-1: or 8.7 Kīlauea Tg per.1 Volcan, grid cell Hawai, (compare: Mean tropical FRP= forest~ kg W m-2 ) Daily mean 3-day running mean Linear Trend LandSAF User Workshop, 15 Nov 21 Kaiser 2

21 Contamination of MODIS-FRP by Gas Flares Total dry matter burned equivalent Global: 128 Tg Top 1 by FRP: 172 Tg (~1.3% of global) Top1 FRP: Source Categories Contribution to total dry matter burned equivalent (Sum Top1 FRP grid cells: 172 Tg) VOLCANO 5% 13% 3% GASFLARE INDUSTRY UNCLEAR 8% FIRES 44% Identification and Masking Out of Non-Biomass Burning MODIS FRP signals in Future LandSAF User Workshop, 15 Nov 21 Kaiser 21

22 Parameterization of Injection Height Freitas 1D Plume Model : validation Fire in Siberia July 23 -Fire radius = 3ha - FRP = 419 MW - moisture = 1 % - entrainment constant =.5 Sensitivity test to fire radius and heat flux Effect of the stable layer Height above the terrain Plume Model 2.7 km MISR 2.1 km LandSAF User Workshop, 15 Nov 21 Kaiser 22

23 Conclusions MACC atmospheric and fire products publicly available starting 23 NRT, as operational GMES service in 214 observation gap filling with data assimilation global MACC services use fire emission products since January (aerosols) and April (greenhouse gases) 21 real-time fire emission rates critically depend on availability and quality of FRP products LandSAF: SEVIRI KCL: GOES-E/-W (to be moved to IM Lisbon in MACC II) NOAA/NASA: MODIS in the future: MTSAT, FY2C, Sentinel-3, NPP VIIRS LandSAF User Workshop, 15 Nov 21 Kaiser 23

24 LandSAF User Workshop, 15 Nov 21 Kaiser 24

25 Contamination of MODIS-FRP by Gas Flares LandSAF User Workshop, 15 Nov 21 Kaiser 25

26 observation gap filling with Kalman smoother in 5-day window fire enhancement consistent with aerosol one associated with transition from El Nino to La Nina More NRT MODIS FRP over SH America LandSAF User Workshop, 15 Nov 21 Kaiser 26