In the past thirty five years NOAA, with help from NASA, has established a remote sensing capability on polar and geostationary platforms that has proven useful in monitoring and predicting severe weather such as tornadic outbreaks, tropical cyclones, and flash floods in the short term, and climate trends indicated by sea surface temperatures, biomass burning, and cloud cover in the longer term. This has become possible first with the visible and infrared window imagery of the 1970s and has been augmented with the temperature and moisture sounding capability of the 1980s. The imagery from the NOAA satellites, especially the time continuous observations from geostationary instruments, dramatically enhanced our ability to understand atmospheric cloud motions and to predict severe thunderstorms. These data were almost immediately incorporated into operational procedures. Use of sounder data in the operational weather systems is more recently coming of age. The polar orbiting sounders are filling important data voids at synoptic scales. Applications include temperature and moisture analyses for weather prediction, analysis of atmospheric stability, estimation of tropical cyclone intensity and position, and global analyses of clouds. The Advanced TIROS Operational Vertical Sounder (ATOVS) includes both infrared and microwave observations with the latter helping considerably to alleviate the influence of clouds for all weather soundings. The Geostationary Operational Environmental Satellite (GOES) Imager and Sounder have been used to develop procedures for retrieving atmospheric temperature, moisture, and wind at hourly intervals in the northern hemisphere. Temporal and spatial changes in atmospheric moisture and stability are improving severe storm warnings. Atmospheric flow fields are helping to improve hurricane trajectory forecasting. Applications of these NOAA data also extend to the climate programs; archives from the last fifteen years offer important information about the effects of aerosols and greenhouse gases and possible trends in global temperature. This talk will indicate the present capabilities and foreshadow some of the developments anticipated in the next twenty years. Summary Remote Sensing Seminar Lectures in Krakow May 2006 Paul Menzel NOAA/NESDIS/ORA

Krakow May 2006

Satellite remote sensing of the Earth-atmosphere Remote sensing of the earth atmosphere depends on the instruments characteristics (mirror sizes, detector signal to noise performance, spectral coverage and resolution) as well as the scene radiation and transmittance through the atmosphere (day or night, cloudy or clear, moist or dry). The following slides will present some of these concepts briefly. Observations depend on telescope characteristics (resolving power, diffraction) detector characteristics (signal to noise) communications bandwidth (bit depth) spectral intervals (window, absorption band) time of day (daylight visible) atmospheric state (T, Q, clouds) earth surface (Ts, vegetation cover)

Spectral Characteristics of Energy Sources and Sensing Systems

Terminology of radiant energy Energy from the Earth Atmosphere over time is Flux which strikes the detector area Irradiance at a given wavelength interval Monochromatic Irradiance over a solid angle on the Earth Radiance observed by satellite radiometer is described by The Planck function can be inverted to Brightness temperature

Definitions of Radiation __________________________________________________________________ QUANTITY SYMBOL UNITS Energy dQ Joules Flux dQ/dt Joules/sec = Watts Irradiance dQ/dt/dA Watts/meter2 Monochromatic dQ/dt/dA/d W/m2/micron Irradiance or dQ/dt/dA/d W/m2/cm-1 Radiance dQ/dt/dA/d/d W/m2/micron/ster dQ/dt/dA/d/d W/m2/cm-1/ster

 c13 Using wavenumbers c2/T Planck’s Law B(,T) = c13 / [e -1] (mW/m2/ster/cm-1) where  = # wavelengths in one centimeter (cm-1) T = temperature of emitting surface (deg K) c1 = 1.191044 x 10-5 (mW/m2/ster/cm-4) c2 = 1.438769 (cm deg K) Wien's Law dB(max,T) / d = 0 where max) = 1.95T indicates peak of Planck function curve shifts to shorter wavelengths (greater wavenumbers) with temperature increase.  Stefan-Boltzmann Law E =   B(,T) d = T4, where  = 5.67 x 10-8 W/m2/deg4. o states that irradiance of a black body (area under Planck curve) is proportional to T4 . Brightness Temperature c13 T = c2/[ln(______ + 1)] is determined by inverting Planck function B

B(max,T)~T5 B(max,T)~T3 B(,T) B(,T) B(,T) versus B(,T)

  c13 c1 Using wavenumbers Using wavelengths c2/T c2 /T B(,T) = c13 / [e -1] B(,T) = c1 /{  5 [e -1] } (mW/m2/ster/cm-1) (mW/m2/ster/m) (max in cm-1) = 1.95T (max in cm)T = 0.2897 B(max,T) ~ T**3. B( max,T) ~ T**5.   E =   B(,T) d = T4, E =   B(,T) d  = T4, o o c13 c1   T = c2/[ln(______ + 1)] T = c2/[ ln(______ + 1)] B 5 B

Spectral Distribution of Energy Radiated from Blackbodies at Various Temperatures

Normalized black body spectra representative of the sun (left) and earth (right), plotted on a logarithmic wavelength scale. The ordinate is multiplied by wavelength so that the area under the curves is proportional to irradiance.

Temperature Sensitivity of B(λ,T) for typical earth scene temperatures B (λ, T) / B (λ, 273K) 4μm 6.7μm 2 1 10μm 15μm microwave 250 300 Temperature (K)

Telescope Radiative Power Capture proportional to throughput A Spectral Power radiated from A2 to A1 = L() A11 mW/cm-1 Instrument Collection area A1 A2 R 1 = A2 / R2 Radiance from surface = L() mW/m2 sr cm-1 Earth pixel {Note: A1 A2 / R2 = A11 = A22 }

Solar (visible) and Earth emitted (infrared) energy Incoming solar radiation (mostly visible) drives the earth-atmosphere (which emits infrared). Over the annual cycle, the incoming solar energy that makes it to the earth surface (about 50 %) is balanced by the outgoing thermal infrared energy emitted through the atmosphere. The atmosphere transmits, absorbs (by H2O, O2, O3, dust) reflects (by clouds), and scatters (by aerosols) incoming visible; the earth surface absorbs and reflects the transmitted visible. Atmospheric H2O, CO2, and O3 selectively transmit or absorb the outgoing infrared radiation. The outgoing microwave is primarily affected by H2O and O2.

Solar Spectrum The top of the atmosphere incoming solar radiation is characterized by a Planck blackbody of temperature about 6000 K. Of the electromagnetic energy emitted from the sun, approximately 50% lies in wavelengths longer than the visible region, about 40% in the visible region (0.4-0.7 m), and about 10% in wavelengths shorter than the visible region. The radiation sensed at the surface of the earth has been attenuated by atmospheric O3, O2, CO2, and H2O (most of the water vapor sensitive bands occur at wavelengths longer than 0.8 um). The visible remote sensing from geo orbit with GOES has been traditionally covering .5 to .9 um; from leo orbit with AVHRR two spectral bands .58 to .68 um (lower reflection from vegetation) and .72 to 1.00 um (higher vegetation reflection) have been maintained. With the launch of the MODIS, the monitoring in the visible has been expanded to 19 bands.

VIIRS, MODIS, FY-1C, AVHRR CO2 O2 H2O O2 H2O H2O H2O O2 H2O H2O CO2

AVIRIS Movie #2 AVIRIS Image - Porto Nacional, Brazil 20-Aug-1995 224 Spectral Bands: 0.4 - 2.5 mm Pixel: 20m x 20m Scene: 10km x 10km Another sequence from AVIRIS shows the different images as a function of wavelength when viewing a partly cloudy scene. Again how the reflectance changes as one approaches the absorption bands and the surface disappears. The reflection from the vegetated surface increases after 0.72 um and surface features become evident. The heavy higher cloud reveals different details at different wavelengths. It is imperative that the satellite evolution in the next decade introduce this type of remote sensing. The science possibilities are enormous. This image loop is the work of Mike Griffen and his co-workers at MIT/Lincoln Lab. 1 1

AVIRIS Movie #1 AVIRIS Image - Linden CA 20-Aug-1992 224 Spectral Bands: 0.4 - 2.5 mm Pixel: 20m x 20m Scene: 10km x 10km A sequence of slides shows the different images as a function of wavelength. Note how the reflectance changes as one approaches the absorption bands and the surface disappears. The reflection from the vegetated surface increases after 0.72 um and surface features become evident. Above 1.0 um the low level smoke becomes less opaque and more transparent. The heavy higher smoke cloud reveals different details at different wavelengths. The fire on the ground becomes evident at the longer wavelengths as thermal emission supplements solar reflection. This image loop is the work of Mike Griffen and his co-workers at MIT/Lincoln Lab. 1 1

Aerosol Size Distribution There are 3 modes : - « nucleation  »: radius is between 0.002 and 0.05 mm. They result from combustion processes, photo-chemical reactions, etc. - « accumulation »: radius is between 0.05 mm and 0.5 mm. Coagulation processes. - « coarse »: larger than 1 mm. From mechanical processes like aeolian erosion. « fine » particles (nucleation and accumulation) result from anthropogenic activities, coarse particles come from natural processes. 0.01 0.1 1.0 10.0

Aerosols over Ocean Radiance data in 6 bands (550-2130nm). Spectral radiances (LUT) to derive the aerosol size distribution Two modes (accumulation 0.10-0.25µm; coarse1.0-2.5µm); ratio is a free parameter Radiance at 865µm to derive t Normalized to t=0.2 at 865µm Ocean products : • The total Spectral Optical thickness • The effective radius • The optical thickness of small & large modes/ratio between the 2 modes

NDSI = [r0.6-r1.6]/[r0.6+r1.6] is near one in snow in Alps

Large positive values of BT4-BT11 occur where reflected solar contributions are increasing BT4 – clouds and low fog and sunglint on open water are obvious BT4 – BT11

Atmosphere transmits visible and traps infrared Selective Absorption Atmosphere transmits visible and traps infrared Incoming Outgoing IR solar E  (1-al) Ysfc  Ya top of the atmosphere  (1-as) E  Ysfc  Ya earth surface. (2-aS) Ysfc = E = Tsfc4 thus if as<aL then Ysfc > E (2-aL)

MODIS IR Spectral Bands This slide shows an observed infrared spectrum of the earth thermal emission of radiance to space. The earth surface Planck blackbody - like radiation at 295 K is severely attenuated in some spectral regions. Around the absorbing bands of the constituent gases of the atmosphere (CO2 at 4.3 and 15.0 um, H20 at 6.3 um, and O3 at 9.7 um), vertical profiles of atmospheric parameters can be derived. Sampling in the spectral region at the center of the absorption band yields radiation from the upper levels of the atmosphere (e.g. radiation from below has already been absorbed by the atmospheric gas); sampling in spectral regions away from the center of the absorption band yields radiation from successively lower levels of the atmosphere. Away from the absorption band are the windows to the bottom of the atmosphere. Surface temperatures of 296 K are evident in the 11 micron window region of the spectrum and tropopause emissions of 220 K in the 15 micron absorption band. As the spectral region moves toward the center of the CO2 absorption band, the radiation temperature decreases due to the decrease of temperature with altitude in the lower atmosphere. IR remote sensing (e.g. HIRS and GOES Sounder) currently covers the portion of the spectrum that extends from around 3 microns out to about 15 microns. Each measurement from a given field of view (spatial element) has a continuous spectrum that may be used to analyze the earth surface and atmosphere. Until recently, we have used “chunks” of the spectrum (channels over selected wavelengths) for our analysis. In the near future, we will be able to take advantage of the very high spectral resolution information contained within the 3-15 micron portion of the spectrum. From the polar orbiting satellites, horizontal resolutions on the order of 10 kilometers will be available, and depending on the year, we may see views over the same area as frequently as once every 4 hours (assuming 3 polar satellites with interferometers). With future geostationary interferometers, it may be possible to view at 4 kilometer resolution with a repeat frequency of once every 5 minutes to once an hour, depending on the area scanned and spectral resolution and signal to noise required for given applications.

GOES Sounder Spectral Bands: 14.7 to 3.7 um and vis The GOES Sounder spectral bands are indicated along with there sensitivity to a particular atmospheric layer. Blue are the temperature sensitive bands, red are moisture bands, and green are surface bands. As indicated earlier, sampling in the spectral region at the center of the absorption band yields radiation from the upper levels of the atmosphere (e.g. radiation from below has already been absorbed by the atmospheric gas); sampling in spectral regions away from the center of the absorption band yields radiation from successively lower levels of the atmosphere. Away from the absorption band are the windows to the bottom of the atmosphere. Surface temperatures of 296 K are evident in the 11 micron window region of the spectrum and tropopause emissions of 210 K in the 15 micron absorption band. As the spectral region moves toward the center of the CO2 absorption band, the radiation temperature decreases due to the decrease of temperature with altitude in the lower atmosphere.

Radiative Transfer through the Atmosphere The radiance leaving the earth-atmosphere system which can be sensed by a satellite borne radiometer is the sum of radiation emissions from the earth surface and each atmospheric level that are transmitted to the top of the atmosphere. Considering the earth's surface to be a blackbody emitter (emissivity equal to unity), the upwelling radiance intensity, R, for a cloudless atmosphere is given by the indicated expression. The first term is the surface contribution and the second term is the atmospheric contribution to the radiance to space. The fundamental principle of atmospheric sounding with meteorological satellites detecting the earth-atmosphere thermal infrared emission is based on the solution of the radiative transfer equation. In this equation, the upwelling radiance arises from the product of the Planck function, the spectral transmittance, and the implied weighting function. The Planck function consists of temperature information, while the transmittance is associated with the absorption coefficient and density profile of the relevant absorbing gases. Obviously, the observed radiance contains the temperature and gaseous profiles of the atmosphere, and therefore, the information content of the observed radiance from satellites must be physically related to the temperature field and absorbing gaseous concentration.

line broadening with pressure helps to explain weighting functions ABC  High Mid Low A B C  ABC 

CO2 channels see to different levels in the atmosphere 14.2 um 13.9 um 13.6 um 13.3 um

Radiative Transfer Equation When reflection from the earth surface is also considered, the RTE for infrared radiation can be written o I = sfc B(Ts) (ps) +  B(T(p)) F(p) [d(p)/ dp] dp ps where F(p) = { 1 + (1 - ) [(ps) / (p)]2 } The first term is the spectral radiance emitted by the surface and attenuated by the atmosphere, often called the boundary term and the second term is the spectral radiance emitted to space by the atmosphere directly or by reflection from the earth surface. The atmospheric contribution is the weighted sum of the Planck radiance contribution from each layer, where the weighting function is [ d(p) / dp ]. This weighting function is an indication of where in the atmosphere the majority of the radiation for a given spectral band comes from.

Clear sky layers of temperature and moisture on 2 June 2001 MODIS TPW Cross section of temperature (degees Centigrade in top right panel) and mixing ratio (g/kg in bottom right panel) across the south western United States (red line in left panel showing total precipitable water vapor in mm) on 2 June 2001. Clear sky layers of temperature and moisture on 2 June 2001

Global TPW from Seemann

RTE in Cloudy Conditions Iλ = η Icd + (1 - η) Ic where cd = cloud, c = clear, η = cloud fraction λ λ o Ic = Bλ(Ts) λ(ps) +  Bλ(T(p)) dλ . λ ps pc Icd = (1-ελ) Bλ(Ts) λ(ps) + (1-ελ)  Bλ(T(p)) dλ λ ps + ελ Bλ(T(pc)) λ(pc) +  Bλ(T(p)) dλ ελ is emittance of cloud. First two terms are from below cloud, third term is cloud contribution, and fourth term is from above cloud. After rearranging pc dBλ Iλ - Iλc = ηελ  (p) dp . ps dp Techniques for dealing with clouds fall into three categories: (a) searching for cloudless fields of view, (b) specifying cloud top pressure and sounding down to cloud level as in the cloudless case, and (c) employing adjacent fields of view to determine clear sky signal from partly cloudy observations.

Ice clouds are revealed with BT8 Ice clouds are revealed with BT8.6-BT11>0 & water clouds and fog show in r0.65

Cloud Properties RTE for cloudy conditions indicates dependence of cloud forcing (observed minus clear sky radiance) on cloud amount () and cloud top pressure (pc) pc (I - Iclr) =    dB . ps Higher colder cloud or greater cloud amount produces greater cloud forcing; dense low cloud can be confused for high thin cloud. Two unknowns require two equations. pc can be inferred from radiance measurements in two spectral bands where cloud emissivity is the same.  is derived from the infrared window, once pc is known. This is the essence of the CO2 slicing technique.

RCO2 RIRW Cloud Clearing For a single layer of clouds, radiances in one spectral band vary linearly with those of another as cloud amount varies from one field of view (fov) to another Clear radiances can be inferred by extrapolating to cloud free conditions. clear RCO2 x partly cloudy xx x x x x cloudy x N=1 N=0 RIRW

Moisture Moisture attenuation in atmospheric windows varies linearly with optical depth. - k u  = e = 1 - k u For same atmosphere, deviation of brightness temperature from surface temperature is a linear function of absorbing power. Thus moisture corrected SST can inferred by using split window measurements and extrapolating to zero k Moisture content of atmosphere inferred from slope of linear relation.

SST Waves from Legeckis

AIRS data from 28 Aug 2005 Clear Sky vs Opaque High Cloud Spectra

AIRS radiance changes (in deg K) to atm & sfc changes

Silicate (ash cloud) signal at Anatahan, Mariana Is Image is ECMWF bias difference of 1227 cm-1 – 984 cm-1 (double difference) obs Note slope obs - clear sky calc

Cirrus signal at Anatahan Image is ECMWF Tb bias difference of 1227 cm-1 – 781 cm-1 (double difference) obs Note slope obs - clear sky calc

AIRS Spectra from around the Globe 20-July-2002 Ascending LW_Window This shows the ascending data (day-time) for the current AIRS “focus day” of 20 July 2002. The image is of a longwave window channel brightness temperature (using a GOES colormap), along with various sample spectra. We’ve had global observations of high spectral resolution before (IRIS and IMG), but they were not really atmospheric profile sounders. The AIRS data is very clean, and the applications are numerous. This is what many of us have been waiting for, and working towards for many years. There are many un-tapped opportunities: SO2 detection from volcanoes, the blue-spike fire detection, new cloud detection techniques, low level inversion detection for severe weather, new spectroscopy investigations, climate change signatures, …

Barren vs Water/Vegetated Inferring surface properties with AIRS high spectral resolution data Barren region detection if T1086 < T981 T(981 cm-1)-T(1086 cm-1) Barren vs Water/Vegetated This is just one example, out of many, of the information you get from high spectral resolution data. It shows how barren surfaces can be distinguished from water/vegetated (nearly black) surfaces due to the spectral variation of their respective surface emissivities. This example is particularly important because this type of technique (combined with the on-line/off-line techniques, also recently called “MLEV”) will be used to distinguish between surface temperature and surface emissivity, which will ultimately improve the retrieval of low level atmospheric temperature and water vapor over land. T(1086 cm-1) AIRS data from 14 June 2002

Sensitivity of High Spectral Resolution to Boundary Layer Inversions and Surface/atmospheric Temperature differences (from IMG Data, October, December 1996)

Offline-Online in LW IRW showing low level moisture Red changes less

Twisted Ribbon formed by CO2 spectrum: Tropopause inversion causes On-line & off-line patterns to cross 15 m CO2 Spectrum Blue between-line Tb warmer for tropospheric channels, colder for stratospheric channels --tropopause-- Signature not available at low resolution

Cld and clr spectra in CO2 absorption separate when weighting functions sink to cloud level

Cld and clr spectra in CO2 absorption separate when weighting functions sink to cloud level

II II I |I I ATMS Spectral Regions

Radiation is governed by Planck’s Law c2 /T B(,T) = c1 /{  5 [e -1] } In microwave region c2 /λT << 1 so that c2 /T e = 1 + c2 /λT + second order And classical Rayleigh Jeans radiation equation emerges Bλ(T)  [c1 / c2 ] [T / λ4] Radiance is linear function of brightness temperature.

For atmospheric applications, MODIS reflective bands can indicate particle size in clouds. Notice how the spectrum opens with increasing wavelength so that measurements at 2.1 and 0.75 um can be used to discriminate 5 to 30 um particles. Again this is a multispectral application of the MODIS data to reveal cloud and clear sky properties.

On board Terra, MODIS (Moderate Resolution Imaging Spectroradiometer) is beginning to deliver exciting data about the oceans, land , and atmosphere. The following slides explain the spectral channel selection and present a few applications examples in each area. For ocean applications, the MODIS team has selected several spectral bands that are on line and off line absorption features associated with chlorophyll and accessary pigments. The multispectral data will be used to reveal their respective concentrations in the ocean waters.

Or land applications, MODIS has several spectral bands that are above and below step function increases or decreases in the reflectivity of vegetation (increases above 0.72 um) or snow/ice (decreases above 1.4 um). The multispectral data will be used to reveal the extent of vegetation and snow/ice in the various regions of the globe.

It is important to note that the atmospheric windows are not transparent (there is some moisture absorption of radiation in the 8 to 12 um region) and that the earth surface does not exhibit blackbody behavior. These characteristics must be accounted for when using remote sensing data over land to infer temperature and moisture profiles. They are also important for mapping land surface properties. MODIS is well situated to make multispectral determinations of land cover type and temperature.

The brightness temperatures of the outgoing radiance observed by an airborne interferometer are shown along with the MODIS infrared spectral bands. The brightness temperatures generally decrease as the center of an absorption band is approached. This decrease is associated with the decrease of tropospheric temperature with altitude. Near about 690 cm-1, the temperature shows a minimum which is related to the colder tropopause. On the basis of the sounding principle already discussed, the MODIS team selected a set of sounding wave numbers such that a temperature layers in the troposphere can be described. The arrows indicate the selection. The associated weighting functions are also shown revealing that the radiation is coming from broad overlapping layers, helping to confound the inversion problem from multispectral radiance measurements to temperature profile.

MODIS

Ice reflectance

Snow Cover from Hall

Energy Cycle From Levizzani

Water Cycle From Levizzani

Comparison of geostationary (geo) and low earth orbiting (leo) satellite capabilities Geo Leo observes process itself observes effects of process (motion and targets of opportunity) repeat coverage in minutes repeat coverage twice daily (t  30 minutes) (t = 12 hours) full earth disk only global coverage best viewing of tropics best viewing of poles same viewing angle varying viewing angle differing solar illumination same solar illumination visible, IR imager visible, IR imager (1, 4 km resolution) (1, 1 km resolution) one visible band multispectral in visible (veggie index) IR only sounder IR and microwave sounder (8 km resolution) (17, 50 km resolution) filter radiometer filter radiometer, interferometer, and grating spectrometer diffraction more than leo diffraction less than geo

HYperspectral viewer for Development of Research Applications - HYDRA MSG, GOES MODIS, AIRS Freely available software For researchers and educators Computer platform independent Extendable to more sensors and applications Based in VisAD (Visualization for Algorithm Development) Uses Jython (Java implementation of Python) runs on most machines 512MB main memory & 32MB graphics card suggested on-going development effort Developed at CIMSS by Tom Rink Tom Whittaker Kevin Baggett With guidance from Paolo Antonelli Liam Gumley Paul Menzel V1.7.4 Transect: Now supported in the psuedo-channel display. Fixed a bug when the user failed to 'drag out‘ a line.         Generally more robust, will update to change in projection, subset.  Works in instrument projection.          ** Note:  only one transect per session.  If you transect on the MCV, you should close          the transect window before you start one on the psuedo-channel window.  I'll add multiple          independent transects in v1.7.6. Planck function:  I've just really started this, but for MODIS and AIRS you can type 'radToBT(value)'     and 'BTtoRad(value)' at the >>>  in the run window.  It uses the channel currently selected in the MCV for     the wavelength, so only have one MCV up when you     do this.  I'll improve this for v1.7.6.  Things are a     little different on the ADDE side, so this won't work  for MSG or GOES - next rev. http://www.ssec.wisc.edu/hydra/

For hydra http://www.ssec.wisc.edu/hydra/ For data and quick browse images http://rapidfire.sci.gsfc.nasa/realtime For MODIS amd AIRS data orders http://daac.gsfc.nasa.gov/