CERES Earth Radiation Budget

1
CERES Earth Radiation Budget Cloud Radiative
ForcingBruce A. WielickiNASA Langley Research
Center

• ISSAOS LAquila, Italy
• August 28, 2002

2
There are many ways to view the Earth..
Focus on radiative energetics
3
Outline

• Climate change background
• CERES data introduction
• Early TRMM and Terra CERES science
• Advances of CERES over ERBE
• Climate observation challenges
• Reference List

4
What global surface temperature change has
occurred so far?
Departures in temperature (deg C) From the 1961
to 1990 average
Data from thermometers
IPCC 2001
5
Human Influence on Climate
Carbon Dioxide Trends 100yr lifetime
Methane Trends
Sulfate Trends
Global Temperature Trends
From M. Prather University of California at Irvine
6
Radiative Forcing from 1750 to 2000
Anthropogenic Forcings
IPCC, 2001
7
Global Temperature Predictions
Model natural forcing
IPCC, 2001
8
Global Temperature Predictions
Model natural anthropogenic forcing
IPCC, 2001
9
Global Temperature Predictions
Uncertainty in climate sensitivity
Uncertainty in future emissions
IPCC, 2001
10
Predicted Sea Level rise from 1990 to 2100
Uncertainty in climate sensitivity
Uncertainty in future emissions
IPCC, 2001
11
What is Climate?

• Climate is the long term average of weather.
• 14-day weather prediction limit but no known
  limit to climate prediction.
• Weather data accuracy is 1 degree, but climate
  accuracy is 0.1 degree a factor of 10 tougher
  measurement.

12
How does the Earth Respond?
Earth System Response
Forces Acting On the Earth System
IMPACTS
Feedback
Of the total forcing of the climate system, 40
is due to the direct effect of greenhouse gases
and aerosols, and 60 is from feedback effects,
such as increasing concentrations of water vapor
as temperature rises.
13
Major Climate System Elements
Water Energy Cycle
Carbon Cycle

Coupled Chaotic Nonlinear
Atmosphere and Ocean Dynamics
Atmospheric Chemistry
14
How can we use observations to test and improve
climate models?

15
Climate System Energy Balance
16
CERES Instrument
TRMM Jan-Aug 98 and Mar-Apr 2000 overlap with
Terra
Terra Mar 00 - present planned life 2006
Aqua July 02 start Now in checkout Planned life
to 2008
NPOESS TBD gap or overlap? 2008 to 2011 launch
17
CERES Data Processing Flow
CERES Data
6 Months
6 Months
6 Months
6 Months
CERES Calibration/ Location
ERBE Inversion
ERBE Averaging
ERBE-Like Products
Cloud Imager Data
18 Mo.
30 Mo.
Cloud Identification TOA/Surface Fluxes
Angular Distribution Models
24 Mo.
Atmospheric Structure
Diurnal Models
36 Mo.
36 Mo.
Surface and Atmospheric Fluxes
CERES Surface Products
Geostationary Data
42 Mo.
Time/Space Averaging
Algorithm Theoretical Basis Documents
http//asd-www.larc.nasa.gov/ATBD/ATBD.html Valida
tion Plans http//asd-www.larc.nasa.gov/valid/v
alid.html
42 Mo.
CERES Time Averaged Cloud/Radiation TOA, SFC,
Atmos
18
Terra/CERES Flight Model 1 Lifetime Radiometric
Stability Determined with the Internal
Calibration Module
Normalized to Ground Calibration Data
19
Unprecedented Accuracy of new EOS Radiation Data
20
CERES Terra 14 day Running Average for TOA LW
Flux March 2000 to May 2001
T. Wong, NASA LaRC and Data Visualization Group,
NASA GSFC
21
Early TRMM and Terra Satellite Results On the
Role of Clouds in Climate

• Focus on the Tropics
• What about the recent Iris hypothesis?
• Was the 1997/98 El Nino really different?
• Is there evidence for decadal change?

22
Global Atmospheric Circulation
23
The Iris Cloud Feedback Concept
Normal Sea Surface Temperature
Warmer Sea Surface Temperature
Thermal Emission Increase (Cooling)
Solar
Solar Absorption Unchanged
More efficient precipitation decreases upper
cloud anvil area
Thermal
Thermal
24
The Iris New Observations Reject
Normal Sea Surface Temperature
Warmer Sea Surface Temperature
Solar
Thermal Emission Increase (Cooling)
Cloud Reflection 0.5 (Iris assumed 0.35)
Solar Increase (Warming)
More efficient precipitation decreases upper
cloud anvil area
Re-analysis did not confirm anvil
area temperature relation
Thermal
Thermal
25
The IRIS strong negative feedback?
CERES
Lindzen et al.
26
The dramatic 1997/98 El Nino

• Rivaled only by the 1983 El Nino during the last
  century.
• First useful climate prediction using ocean and
  atmosphere observing systems
• Can we use it as a test of short term climate as
  well as the effects of clouds on long-term
  climate change?

27
Cloud Radiative Forcing Definitions

• Top of Atmosphere SW and LW Flux
• SW CRF Rclr - R
• Where R is TOA all-sky reflected SW flux
• Where Rclr is TOA clear-sky reflected SW flux
• LW CRF Fclr - F
• Where F is TOA LW all-sky, Fclr is clear-sky
• Net CRF SW CRF LW CRF
• Positive values warm the earth system
• N - (SW CRF) / (LW CRF)
• For N gt 1, cooling dominates
• For N 1, Net CRF 0
• N is independent of cloud fractional coverage
• See Cess et al., GRL, Dec 2001

28
Cloud Radiative Energy Anomalies in the 98 El Nino
West Pacific
East Pacific
Cess et al. 2001 GRL
N ratio of shortwave cloud radiative forcing
(cooling) to longwave cloud radiative forcing
(warming). 1.0 is balanced Changes in 1998
unprecedented cloud height or thickness?
29
Fu-Liou Radiative Model Calculations
N is independent of cloud fraction
N decreases with cloud height
N increases with cloud optical depth
30
Cloud Forcing Ratio 2.5 deg monthly mean regions
Jan-Apr 1989 Normal
Jan-Apr 1998 El-Nino
Western Pacific
Eastern Pacific
31
Cloud Forcing Ratio 2.5 deg monthly mean regions
Jan-Apr 1985 Normal
Jan-Apr 1987 El Nino
Western Pacific
Eastern Pacific
32
SAGE II Cloud Heights Confirm Surprising
Radiative Anomalies Cause is Cloud Height
West Pacific Warm Pool
Tropical East Pacific
Cloud Altitude (km)
Large Cloud Ht Increase 1998 El Nino Moderate
Increase 1987 El Nino
Large Cloud Ht Decrease 1998 El Nino Small
change 1987 El Nino
Cess et al, 2001
Cloud Top Cumulative Frequency
33
Normal Year (85,86,89 average) Strong Walker Cell
Pressure/Latitudecross-sections of 5S-5N zonal
wind (m/s) for JFMA
1987 Weak El Nino Weakened Walker Cell
100 mb
1998 Strong El Nino No Walker Cell
900 mb
80E Longitude 280E
34
Jan/Feb 98 El Nino Thermal Flux Anomalies
NASA CERES Radiation Observations
NOAA GFDL Standard Climate Model
NOAA GFDL Experimental Prediction Model
35
(No Transcript)
36
An overlapping Earth radiation climate record 22
years from Nimbus 7 to Terra.
37
What about the latest HIRS/AVHRR Pathfinder OLR
Records? Differences are as large as the signal
HIRS and AVHRR include estimated corrections
for calibrationchanges between instruments as
well as NOAA orbit diurnal drifts.
AVHRR data set from Jacobowitz
HIRS data set from Susskind with
Robertsoncalibration/diurnal corrections.
Figure from Wielicki et al., Science Feb 02, 2002
38
Trenberth Letter to Science/Response

• Trenberth concerns
• a) 3 month ERBS gap caused 3 W/m2 shift in LW
• b) Diurnal aliasing causing seasonal cycle change
  in 90s versus 80s
• Response (both accepted by Science)
• No offset changes expected for cavities and no
  change relative to HIRS/AVHRR. Also fails to
  explain SW changes.
• Yes 36-day precession cycle means instead of 30
  day months removed 2/3 of seasonal cycle change
  in SW, emphasized decadal signal, some
  variability remains.
• Both need stronger emphasis on redundant high
  accuracy calibrated overlapped climate data sets
  e.g. broadband and spectral LW flux.

39
Comparison of Observed Decadal Tropical Radiation
Variation with Current Climate Models
LW Emitted Thermal Fluxes SW Reflected Solar
Fluxes Net Net Radiative Fluxes

• Models less variable
• than the observations
• missing feedbacks?
• missing forcings?
• clouds physics?

40
For future climate analysis and cloud model
testing Think Cloud Objects, Not only grid boxes
41
NCEP 500hPa Omega
ISCCP Cloud Amount
Bates HIRS UTH
Chen, Del Genio, Carlson, NASA GISS
42
Rossow/Carlson ISCCP Rad Model gt TOA flux
anomalies
43
SAGE II Cloud Top Frequency Distribution Reduced
Cloud Ht
Wang et al, GRL, 2002
44
SAGE II Cloud Height Changes Radiative
Parameterization Only Explain 1/3 of ERBS LW
flux anomaly other 2/3 inferred to be Cloud
amount and/or emissivity changes (all 3 shown
below)
Wang et al., GRL 2002
45
Summary Decadal Variability

• No calibration problem yet explains decadal ERBS
  changes. Bob Lee et al. continuing to examine
  ERBS calibration.
• Seasonal changes in 90s tropical SW fluxes
  primarily aliasingof ERBS diurnal cycle into
  monthly means. Some increased 90svariability
  remains for both 36-day and 72-day precession
  cycle averages. Decadal changes now clearer.
• ISCCP, NCEP omega, Bates HIRS UTH roughly
  consistent but phasing, peaks, differ do not
  expect simple linear relationsbetween these
  variables. Further analysis needed.
• SAGE II cloud height decreases explain 1/3 of 3
  W/m2 LW change
• Percentile tests show most changes in high LW
  flux, not low downward branch of Hadley/Walker
  cells consistent with strengthened cells. Will
  redo this analysis to test gain changes.
• Surface observer clouds in 80s/90s show mixed
  bag reduced warm pool cloud (high), and varying
  low cloud amounts

46
Why are clouds so tough?

• Aerosols lt0.1micron, cloud systems gt1000 km
• Cloud particles grow in seconds climate is
  centuries
• Cloud growth can be explosive 1
  thunderstorm packs the energy of an H-bomb.
• Cloud properties can vary a factor of 1000 in
  hours.
• Few percent cloud changes drive climate
  sensitivity
• Best current climate models are 250km scale
• Cloud updrafts are a 100m to a few km.
• A climate model resolving all cloud physics down
  to aerosol scale would require 1038
  supercomputers 190 years of current Moores Law
  rate of advance.

47
How can we improve in the future?
48
CERES
ERBE
100 km global
0.1 micron aerosol
50 km column
100 km global
100m - 1km cloud cell
Range of Cloud/Aerosol/Radiation Model Tests
49
ERBE Error Analysis CERES goals

• Instantaneous TOA Flux error dominated by
• Angle Sampling Error (new adms factor 2-3)
• Monthly mean regional TOA flux errors dominated
  by (CERES improvement)
• Absolute calibration (factor of 2)
• Angle Sampling Error (new adms factor 2-5)
• Time Sampling Error (add geo factor of 2-3)
• Interannual/Decadal errors dominated by
• Calibration stability (lt 1 Wm-2, goal 0.25 Wm-2)

50
TOA Flux Validation

• Regional Direct Integration (DI) Checks
• DI uses 8 months of regional radiances from
  CERES rotating azimuth plane data at all viewing
  angles and solar zeniths no ADMs.
• ADM-DI flux difference (1s) is 0.5 W m-2,
  tropical bias 0.1 W/m2.

51
Iris Climate Sensitivity Varying Cloud Types
Cloud Temperature Tested -15C to -30C
TRMM CERES/VIRS SSF Values
Cloud Particle Phase Tested 50 to 99 ice
Iris Values
Chambers et al. (accepted, J.Climate) Still
conclude that the Iris effect is a small or
slightly positive feedback not a large negative
feedback.
52
Large Ensemble Cloud Object Tests

• Unlike ERBE, new CERES SSF radiative fluxes are
  accurate for individual cloud types when
  ensembled over the range of viewing angles.
• CERES radiative fluxes are also matched with
  imager cloud
• Instead of large ensembles of grid boxes,
  consider large ensembles of cloud type
  unscramble the climate fruit salad and sort
  clouds by cloud type for CERES fovs (15 km avg
  for CERES on TRMM, 30km on Terra).
• Deep Convection (Zgt10km, tau gt 10, Cgt0.99) for
  CERES fovs
• Trade Cumulus (Z lt 3km, 0.1 lt C lt 0.4) in CERES
  fovs
• Transition Scu (Z lt 3 km, 0.4 lt C lt 1.0) in CERES
  fovs
• Stratocumulus (Zlt 3 km, C gt 1.0)
• Use ECMWF matched in space/time for
  T,q,winds,tendencies
• Compare to ECMWF and Cloud Resolving Model Cloud

53
Satellite data analysis method

• Define a cloud system as
• a contiguous region of the
• Earth with a single dominant
• cloud type (e.g. stratocumulus,
• stratus, and deep convection)
• Determine the shapes and
• sizes of the cloud systems by
• the satellite data and by the
• cloud property selection criteria
• (Wielicki and Welch 1986)

54
Satellite data mapped onto ECMWF 0.5 deg grid
55
First Results of CERES/ECMWF/CRM Comparisons

• Deep Convective Clouds with effective diameter gt
  300km
• 29 cases in March, 1998 Warm Pool region
• 0.5 degree ECMWF data down-scaled to CERES TRMM
  fov size of 15 km by using random/maximum overlap
  assumption as in Klein and Jacob (1999) with
  ECMWF cloud fraction, LWP, IWP in layers plus
  Fu-Liou radiation code.
• CRM runs using 2 km 2-D simulations (Xu model)
  forced by ECMWF tendencies and atmospheric state.
  

56
Deep convection, D(object)gt300km Water plus ice
path (excludes graupel, snow, rain)
57
Deep convection, D(object)gt300km Water plus ice
path (excludes graupel, snow, rain)
58
Deep convection, D(object)gt300km Water plus ice
path (excludes graupel, snow, rain)
59
Next steps for CERES cloud objects

• Add boundary layer cloud cases
• Provide matched CERES cloud/radiation and ECMWF
  data to other GCM/cloud modeling groups
• Extend to compare 1998 ENSO clouds to 2000 mild
  La Nina phase
• Extend to larger ensembles
• Improve microphysics parameterization in CRM
  (even cloud height sensitive to this)
• Improve spatial scale consistency of satellite
  and model comparisons.
• Examine as a function of cloud object size and
  frequency of occurrence of cloud type

60
New CERES Surface FluxesImproved Time
InterpolationReleased for TRMMAug, Sept, 2002

• All CERES data products are delivered with a Data
  Quality Summary (web access) for quicker access
  to validation results than the journals.

61
CERES TRMM SSF Surface Flux ValidationBSRN(4),
ARM(21), CMDL(6) Sites, 40N-40S
SSF Surface Flux Algorithms Simpler surface flux
onlyparameterizations CERES TOA VIRS cloud
ECMWF

• Component Samples Bias Sigma
• LW Surface Down (1 min averages)
• Clear-sky 2000 -6 Wm-2 20 Wm-2
• All-sky 6000 -4 Wm-2 21 Wm-2algorithm Gupta
  (clr, all-sky), Ramanathan (clr)
• SW Surface Down (24 hr mean bias, instantaneous
  s)
• Clear-sky (1min) 1000 -7 Wm-2 25 Wm-2
• All-sky (60 min) 3000 6 Wm-2 65
  Wm-2algorithm Staylor (clr, all-sky)

62
CERES CRS TRMM Surface Flux ValidationARM
Central Facility
CRS Surface Flux Algorithms Fu-Liou radiative
modelusing VIRS cloud ECMWF, constrained to
CERES TOA

• Component Samples Bias (CRS-OBS) Sigma
• Longwave
• SFC Down All-sky 450 -1 Wm-2 18 Wm-2
• TOA Up All-sky 450 0 Wm-2 4 Wm-2
• Shortwave (24 hr mean for bias, instantaneous s,
  all qo)
• SFC Down Clr 94 9 Wm-2 26 Wm-2
• SFC Down Clr/Pyr 18 5 Wm-2 14 Wm-2
• SFC Down All-sky 260 9 Wm-2 58 Wm-2
• TOA Up All-sky 260 -2 Wm-2 10 Wm-2

63
Why Is There More Than One CERES Monthly Mean
Product?

• ERBE-like
• Consistent with ERBE processing
• Useful for comparisons with ERBE climatology
• 2.5 grid
• TOA fluxes
• Limited cloud information

• SRBAVG
• Takes advantage of improved CERES fluxes
• Uses improved temporal interpolation to remove
  sampling effects
• 1.0 grid
• TOA and surface fluxes
• Detailed cloud properties
• Product contains GEO and nonGEO monthly means

64
Using Geostationary Data for Temporal
Interpolation of TOA Fluxes

• 3-hourly imager data from geostationary
  satellites is used to define diurnal variations
  between CERES observations
• Calibration of geostationary data is tied to
  VIRS/MODIS
• Cloud retrieval is a subset of CERES VIRS/MODIS
  algorithm
• Narrowband GEO data converted to flux using NB-BB
  relationship CERES ADMs
• Daily regional fluxes are normalized to CERES
  observations
• 5 errors in geo calibration (visible or
  infrared) are reduced to lt1 Wm-2 SW, lt 0.3 Wm-2
  LW when normalized to CERES observations.
• Final validation tests of new time interpolation
  accuracy on daily through monthly time scales
  will be against GERB data (2003).

65
Temporal Interpolation of TOA LW FluxJanuary
1998 E. Sahara 24.5N 20.5E
66
Summary of CERES Advances

• Calibration Offsets, active cavity calib.,
  spectral char.
• Angle Sampling Hemispheric scans, merge with
  imager matched surface and cloud
  properties new class of angular, directional
  models
• Time Sampling CERES calibration 3-hourly geo
  samples new 3-hourly and daily mean fluxes
• Clear-sky Fluxes Imager cloud mask, 10-20km FOV
• Surface/Atm Fluxes Constrain to CERES TOA,
  Fu-Liou, ECMWF imager cloud, aerosol, surface
  properties
• Cloud Properties Same 5-channel algorithm on
  VIRS,MODIS
• night-time thin cirrus, check cal vs CERES
• Tests of Models Take beyond monthly mean TOA
  fluxes to a range of scales, variables, pdfs
• ISCCP/SRB/ERBE overlap to improve tie to 80s/90s
  data.
• CALIPSO/Cloudsat Merge in 2004 with vertical
  aerosol/cloud
• Move toward unscrambling climate system energy
  components

67
What CERES Products are Available?

• All radiance and ERBE-Like data with about a 2
  month lag for Terra. 7 month checkout for Aqua.
• New generation products
• SSF combined cloud/radiation/new ADMs for all
  TRMM.
• SSF Edition 1 starts this summer for Terra but
  initially uses TRMM ADMs until 2 years of Terra
  processed to provide new ADMs in spring, 2003.
• SRBAVG monthly mean combined CERES/geo surface
  and TOA fluxes reach validated TRMM Edition 1 in
  Aug 2002. Terra awaits new Terra ADMs and
  directional models but is critical to correct
  morning sampling bias
• CRS uses SSF plus ECMWF profiles and FuLiou
  model calculations constrained to TOA fluxes to
  improve surface and atmosphere fluxes best
  integrated fov data validated and available for
  all TRMM data July, 2002.
• AVG combines CRS with geo 3-hourly time sampling

68
(No Transcript)
69
A-Train Formation for Aerosol and Cloud
Vertical Profiles Atmospheric State gt
Aerosol/Cloud gt Radiative Heating
A-Train Launch 2004
70
Calipso, Cloudsat and Aqua in Formation Testing
Global Cloud Models
Aqua
Predict Solar and Thermal Infrared Fluxes
CALIPSO Lidar and Cloudsat Radar aerosol and
cloud vertical profiles
CERES energy fluxes, MODIS cloud optics
Aqua
Aqua
Heat or Cool Surface Atmosphere
Predict Layers of Water Ice Clouds
AIRS/AMSU/MHB Temperature, Humidity, Winds
71
Future Climate Observing Challenges

• No Climate Observing System has been designed or
  exists.
• NPOESS not funded to accept climate requirements
• NASA and international research focused.
• Climate observations need research quality
  combined with operational continuity
• Weather and research data systems currently made
  to fit.
• Use well sampled weather data (but often lacks
  accuracy for climate, and misses many variables)
• Use poorly sampled research data (often good
  accuracy but gaps or poor overlap)
• Climate observing different than weather
• Typically factor of 10 higher accuracy/stability
• Rigorous overlap requirements example no overlap
  planned for broadband radiation. We hope to luck
  out like ERBS lasting for 15 years.
• Independent analysis and observation strategies
  critical to confirm surprises.

72
Future Climate Observing Challenges

• Need critical work linking new to old climate
  data
• Leverage new developments like EOS, ENVISAT,
  ADEOS
• Link improvements, knowledge back to ISCCP
  decades
• Link forward to next generation international
  observations.
• A daunting task at climate accuracy
• We await the climate epiphany (at least in the
  U.S.)
• Observing cost could more than double a weather
  system
• Calibration/stability not resolution is highest
  priority
• But no apriori guarantees of success.

73
Future Climate Observing Challenges

• What would we do with climate prediction
  certainty if we had it and climate change is
  predicted to be large?
• Renewable energy development.
• Energy conservation/efficiency.
• Decadal plans for energy system transitions, land
  use change patterns, sea-level rise mitigation.
• Vary response with regional changes.
• Is human society capable of coordinated and
  planned action on global decade time scales?
• For ozone, the answer was yes, but
• Climate has much larger economic stakes.

74
CERES Reference List

• CERES General Background
• CERES Brochure (on the CERES home page)
• Role of Clouds and Radiation in Climate, Wielicki
  et al., BAMS,76, 853-868, 1995.
• CERES Experiment Overview Wielicki et al., BAMS,
  96, 853-868, 1996.
• CERES Instrument Calibration Priestley et al.,
  J. Appl. Met, 39, 2249-2258, 2000.
• CERES Data Products and Algorithms
• CERES Algorithm Theoretical Basis Documents
  (ATBDs) NASA Reference Publication 1376, Volumes
  1 through 4, Dec. 1995. ATBD overview published
  inWielicki et al., IEEE Trans Geoscience Rem
  Sens, 36, 1127-1141, 1998.
• CERES Data Products Catalog summary of data
  products
• CERES Data Collection Guides one per data
  product defines formats/variables.
• CERES Data Quality Summaries one per data
  product summarizes current estimates of the
  accuracy of variables in each validated archived
  CERES product.
• The above can be found at http//asd-www.larc.nas
  a.gov/ceres/docs.html
• Tropical decadal variability
• Wielicki et al., Science, Vol 295, Feb 1, 2002,
  p841-844. (decadal radiation changes)
• Chen et al., Science, Vol 295, Feb 1, 2002
  p838-841. (hadley/walker hypothesis)
• Trenberth, Science 295 (5576) U1-U2 Jun 21 2002
  (letter to science)
• Wielicki et al., Science 295 (5576) U2-U3 Jun 21
  2002 (response)
• Allan et al., J. Climate 15 (14) 1979-1986 Jul
  2002 (UKMO runs)
• Wang et al., GRL, 29, No. 10, 2002. (SAGE II
  cirrus height changes)

75
CERES Reference List, cont

• 1998 El Nino Radiative Anomalies
• Cloud Forcing Ratio Anomaly Cess et al., J.
  Climate, 14, 2129-2137, 2001.
• Cloud Forcing Ratio Anomaly/SAGE II cloud height
  anomalies Cess et al., GRL, 28, 4547-4550, Dec
  15, 2001
• Iris tropical cloud negative feedback hypothesis
• The Iris Hypothesis Lindzen et al., BAMS, 82,
  417-432, 2001.
• Cloud amount/SST relation Hartmann and
  Michelson, BAMS, 83, 249-254, 2002.
• Cloud radiative properties Lin et al., J
  Climate, 15, 3-7, 2002.
• Cloud radiative properties Fu et al., Atm Chem
  Phys, 2, 31-37, 2002.
• Improved cloud radiative properties using new
  CERES merged cloud/radiation data products (TRMM
  SSF) Chambers et al., J Climate, in press (for a
  pdf copy, contact l.m.chambers_at_larc.nasa.gov)

76
Where do I go for CERES data and documentation?

• CERES Documentation/Home Page at
• http//asd-www.larc.nasa.gov/ceres/docs.html
• CERES Data Orders at
• http//eosweb.larc.nasa.gov/latisweb

77
Nature is a mutable cloud which is always and
never the same. - Ralph Waldo Emerson
(1803-1882)
Man masters nature not by force, but by
understanding. - Jacob Bronowski, 1956