It is often said that science is based on observation. But observation is never as straightforward as you might expect.
Now we have nice maps of Venus based on Magellan data. But the data is a result of a very complicated data gathering and transformation techniques developed over a long period.
Venus is covered with a dense cloud -- Optical observations can reveal little about the surface of Venus.
Late 19th century -- early telescopic and spectroscopic observations: Venus's atmosphere is similar to Earth's atmosphere
Percival Lowell -- Canals on Venus (1897)
Svante Arrhenius -- `Steamy jungle' model of Venus's surface (1918)
St. John and Nicholson -- Desert landscape model of Venus's surface based on their failure to detect water on Venus (1922)
Adams and Dunham -- Spectroscopic study of Venus's atmosphere: Venusean atmosphere is dominated by CO2 (1932).
Whipple and Menzel -- `seltzer water' model (1955)
Fred Hoyle -- `Petroleum seas' model (1955)
Estimation of rotation period varied from about 24 hours to 300 days.
These early observations told us very little about Venus.
1946 first successful detection of radar return from the moon (birth of radar astronomy)
At this time, it was simply a detection of an object.
Problems at this point
Need to send a strong signal; need a very effective transmitter
Very low signal-noise ratio
Unknown sources of signal loss
When these problems were solved, radar astronomers started to get meaningful data.
To send out very short pulse -- echoes come back with delay due to the curvature of the object.
Sub radar point
Gating
Circle of constant delay
Echo power distribution
Radar cross section
Quasi-specular component
Diffuse component
Scientists tried to find `laws' to govern echo power distribution
Quasi-specular component -- exponential rather than Gaussean
Diffuse component -- cos 3/2 [[phi]]
Curve fitting
Curve fitting is a two-way process
a. Transformation of the data (different scales etc.) -- until you get something familiar (straight line, parabola etc.)
b. Among familiar curves, pick one which fits the data best.
-- Are they laws in Hempel's sense (universal laws which cover counterfactual cases)?
What kind of information can we expect from such data?
Determination of exact locations of planetary bodies -- bootstrapping procedure
Determination of AU
Long wavelength -- only a small portion around the sub-radar point is bright, and the rest is dark
Short wavelength -- the entire surface looks equally bright
This phenomenon can be explained using the geometric ray model together with certain assumptions.
1. A radar signal is bounced back to the antenna when the facet of the surface the radar wave hits is perpendicular to the direction of the ray. (the geometric ray model)
2. For the geometric ray model to work, the facets of the surface should be at least several times the size of the radar wavelength.
3. Such facets are distributed equally on the surface.
A consequence
A region becomes radar bright when that region has a lot of small facets perpendicular to the angle of the ray. -> Radar brightness is correlated with surface roughness, rather than elevation or something else. Radar brightness is also a function of the angle of the radar.
Thus -- Deviation from the average echo power distribution ("anomaly") means that the surface of the region is particularly rough.
However: the echo power at a certain point is a sum of the echo power of the constant delay circle. This is not very specific.
but scientists could correlate an anomaly in the radar brightness with geological features on the moon. -> Identification of Tycho
Comparison between the Moon and Venus
1958~ attempts to detect radar return from Venus
1961 first unambiguous radar return detection (Richard Goldstein)
Radar astronomers obtained a similar echo power distribution for Venus
Exponential law for the quasi-specular component, and cos 3/2 [[phi]] law for the diffuse component.
The comparisons did not yield robust quantitative results, but they yielded a robust qualitative result about the average slope of two planetary bodies:
Average slope of Venus < average slope of the moon
In the case of the moon, astronomers could use optical observation to identify Tycho in the radar return. But in the case of Venus, pulse radar alone could not specify any such geological feature on Venus.
Well, scientists found a way.
If we send a continuous wave, we lose the power to detect constant delay, but instead we can detect changes in frequency due to the Doppler effect.
If the object is moving toward the antenna -- higher frequency
If the object is moving away from the antenna -- lower frequency
By sorting return radar by frequency, we can draw another kind of echo power distribution graph.
What kind of information can we expect from such data?
Rotation speed of the planet (though the direction of the rotation cannot be determined by this)
Again, an anomalous deviation from the average echo power distribution tells us that there is a rough surface. But, again, the echo power is the sum of the radar returns from regions on the same Doppler shift circle. This is not very specific.
Unlike the moon, Venus is rotating relative to the Earth, and this has helped scientists to identify some of the prominent features on Venus:
1965 Identification of alpha and beta regions (Goldstein)
1966 Identification of more small regions (Carpenter)
But what if we use both pulse radar and continuous wave simultaneously?
Radar astronomers found a way to code pulse into continuous radar. This is called Range-Doppler system.
New radar antennae with range-Doppler equipment
1963 Arecibo (Puerto Rico)
1964 Haystack
1966 Goldstone
Planes of constant delay and planes of constant Doppler shift are orthogonal -- there are two intersections
Thus, by correlating an anomaly on a constant delay circle with another on a constant Doppler-shift circle, we can specify the source of the anomaly as one of two places.
North-south ambiguity
a. Aiming at only southern or northern hemisphere of the object (not applicable to Venus)
b. Use of two telescopes (interferometry)
c. Relative motion of the Earth and Venus could be used to solve the north south ambiguity
Finally radar astronomers started to obtain something like a map. But still they cannot solve north south ambiguity around the equator.
1967 Image of Venus using (b) (Haystack and Westford antennae)
1969 Image of Venus using (c) (Goldstone antenna)
From 17 sets of data, Goldstein and his team could manage to reduce the problem to 20,000 linear equations with 20,000 unknown variables. -- They got only an approximate solution to it.
1972 A low resolution map (80km) for a large part of the Venus surface, and a high resolution map (10-15 km) for a limited region. (Goldstone antenna)
Rumsey et al claimed that they found a large (160 km in diameter) impact crater in the high resolution map.
Other planetologists started to conjecture about other possible craters.
1977 Schaber and Boyce `identified' in the 1972 low-resolution map twelve possible impact basins of more than 600 km in diameter.
What planetologists saw were dark circular regions on the maps. Why did they jump to the conclusion that these are impact craters?
They were anxious to learn about impact craters.
Impact craters are especially useful for learning about the history of the planet, and for learning whether the planet is geologically alive.
Compare the earth and other planetary bodies like the moon, Mars and Mercury. What's the difference between them?
A geologically alive planet like the earth (plate tectonics, volcanism) has very few impact craters. Craters are soon obliterated on such a planet.
Size of craters can tell us more. The number of possible impactors was reduced significantly early in the history of the solar system, and such impactors that can create more than 600 km crater became extremely rare three billion years ago. That is, if we find large impact craters on Venus, we can conclude that the age of that region of the surface of Venus is very old, probably more than three billion years.
Conclusion of Schaber and Boyce paper:
"The fact that the radar data indicate a cratering history on Venus that is very similar to that of the moon, Mars, and Mercury is of significant importance. The existence of numerous craters on Venus also implies that the surface of the planet, like that of the moon, Mars and Mercury, has not been significantly disturbed by plate-tectonics as it operates on Earth. Additionally, preservation of abundant large craters on the Venus surface suggests that atmosphere-related erosion has been orders of magnitude slower than on Earth, or that the dense atmosphere is a comparatively recent feature of the planet."
Do you think their conclusion is warranted, given what you have learned about radar astronomy so far? Do radar-dark circular regions indicate the existence of impact craters?
Here is the problem: from hindsight with Magellan data at hand, we can say that most of the identified circular regions are not impact craters (often there are no circular structures either). From hindsight, we can see how sloppy the planetologists were.
But is this a legitimate way to look at scientists?
Let's think again about normative vs. descriptive approaches in the philosophy of science.
Another related issue: Whig history vs. non-Whig history.
Whig history -- to look at the history from the advantageous point of knowing `correct' answers.
Non-Whig history -- to look at the history ignoring what happened later.
Now, many historians of science take the latter position. Whig history, according to them, distorts the way the history actually went, and cannot do justice to the scientists in the past.
Do you agree?