mirrored file at http://SaturnianCosmology.Org/ 
For complete access to all the files of this collection
	see http://SaturnianCosmology.org/search.php 
==========================================================
[1]Planetary Science Research Discoveries 

[2][LINK] [3][LINK] [4][LINK] [5][LINK] [6][LINK] [7][LINK] 
_________________________________________________________________

[8]pdf version   [9]PSRD-asteroidHeating.pdf

HOT IDEA posted April 2, 2004
Kernouve

Asteroid Heating: A Shocking View

--- Mineral intergrowths in chondritic meteorites may indicate that
some asteroids were heated by impact.

Written by [10]G. Jeffrey Taylor
Hawai'i Institute of Geophysics and Planetology

Almost all meteorites are chips of asteroids. Their mineralogical and
chemical properties show that they have been heated to varying
amounts, from a little (only 25 ^oC or so) to melting at more than
1100 ^oC. Isotopic dating shows that meteorites formed near the
beginning of the solar system 4.55 billion years ago. What was the
source of energy to heat them? The leading candidate is the decay of
short-lived radioactive isotopes such as aluminum-26 (^26Al). This
isotope has a short half-life, only 700 thousand years. It decays
rapidly, releasing heat as it does so. Studies of chondritic
meteorites prove that ^26Al was present when the solar system formed,
so it is the logical source of heat to warm up and even melt
asteroids.

Parts of asteroids have also been heated by the impact of other
asteroids. Alan Rubin (UCLA) has been studying the effects of shock in
chondrites for a long time. He has documented what this generally
messy process does to chondrites and other stony meteorites whose ages
show that they were affected by impacts long after ^26Al had decayed
completely away. He now reports that some of the features common to
shocked chondrites (those clearly affected by impact) are also
abundant in apparently unshocked chondrites. This leads him to suggest
that impacts played an important role in heating chondrites. If
correct, this conclusion has great implications for the early history
of the asteroid belt. The idea is controversial, however, and will be
debated by meteoriticists.

References:
Rubin, Alan E. (2003) Chromite-plagioclase assemblages as a new shock
indicator; implications for the shock and thermal histories of
ordinary chondrites. [11]Geochimica et Cosmochimica Acta, v. 67,
p. 2695-2709.

Rubin, Alan E. (2004) Postshock annealing and postannealing shock in
equilibrated ordinary chondrites: Implications for the thermal and
shock histories of chondritic asteroids. [12]Geochimica et
Cosmochimica Acta, v. 68, p. 673-689.
_________________________________________________________________

The Heating and Cooling of Ordinary Chondrites

Ordinary chondrites never melted, although some of their constituents
did. Chondrites are named for the millimeter-sized spherules in them,
chondrules, which formed as molten droplets in the gas and dust cloud
surrounding the still-forming Sun. Nobody really knows how chondrules
were made, but the consensus is that they formed before the asteroids
and planets. Chondrules aggregated along with unmelted dust and
metallic particles to form the parent asteroids of chondrites. The
compositions and intergrowths of minerals in chondrites show that the
amount of heating varied considerably inside the chondritic asteroids.
Some were barely heated. Others were heated to 900 ^oC, causing
mineral compositions to become uniform throughout and the boundaries
between chondrules to become vague. Studies of thousands of chondrites
show that there is a sequence of thermal effects, from unheated to
strongly heated, forming a metamorphic sequence from type 3
chondrites, through types 4 and 5, to type 6, the most severely
modified. Mineral compositions indicate that the highest temperature
reached during metamorphism ranged from about 400-500 ^oC (for type 4)
to 900 ^oC (type 6).

Semarkona St Severin

Photomicrographs of ordinary chondrites, showing an unmetamorphosed
type 3 chondrite (Semarkona) on the left and a thoroughly
metamorphosed type 6 chondrite (St. Severin) on the right. Chondrules
(round objects clearly visible in the photograph on the left) become
progressively less visible with metamorphic grade.

The metallic minerals in ordinary chondrites reliably record the rates
at which chondrites cooled after they were heated. The cooling rate is
recorded in variations in the concentrations of nickel and iron in
metallic iron-nickel minerals. Cooling rates are quantified by
knowledge of the rate of movement of iron and nickel in solid metal
and of the way the compositions of two iron-nickel minerals (kamacite
and taenite) vary with temperature. Numerous measurements indicate
that most chondrites cooled at rates between 1 and 100 ^oC per million
years. These rates are generally consistent with rates derived from
precise determinations of the isotopic ages of chondrites: the older
the chondrite, the faster the cooling rate. Rates between 1 and 100
^oC/million years imply burial beneath several tens of kilometers of
rock.

Adelie Land

Photomicrograph of a polished slab of a chondrite (Adelie Land)
showing kamacite and taenite grains. The slab was polished and then
lightly etched in a solution of 1% nitric acid in ethanol, revealing
microscopic properties. The color variation in taenite reflects
variation in the concentration of nickel, which allows us to quantify
the cooling rate of the chondrite.
_________________________________________________________________

Aluminum-26 and Asteroid Heating

^26Al was present when meteorites were forming (see PSRD article
[13]Using Aluminum-26 as a Clock for Early Solar System Events). It is
a radioactive isotope with a half-life of only 700 thousand years, so
its presence means that the solar system formed within a few
half-lives of the formation of ^26Al in an exploding star. It decayed
by emitting a beta particle (an electron), creating ^26Mg
(magnesium-26) and releasing energy. The energy released is
considerable. If ^26Al made up only 5 x 10^-5 (0.005%) of all the
aluminum in a chondrite (most is aluminum-27, which is not
radioactive), it would release enough energy to melt asteroids a few
kilometers across and larger. Lower amounts of ^26Al cause less
melting.

Because aluminum was probably distributed uniformly throughout the
asteroid, the body would have been heated uniformly, except for a
temperature gradient in the regions near the surface where heat
radiated into space. Once ^26Al had decayed for five or ten half-lives
it was not abundant enough to heat an asteroid, so continuous cooling
began.

Other heat sources have been proposed to explain asteroid heating.
Floyd Herbert and Charles Sonnett (University of Arizona) suggested
that huge outflows from the Sun could have caused induction heating of
entire asteroids. It would be like zapping asteroids in a gigantic
microwave oven. Heating would have been heterogeneous inside the body,
dependent on variations in electrical conductivity. However, this heat
source does not leave behind a specific fingerprint, unlike ^26Al
which leaves behind extra ^26Mg, so it is hard to prove or disprove.
Conventional wisdom is that asteroids were heated by the decay of
^26Al.
_________________________________________________________________

The Effects of Impact

High-velocity impacts leave impressive fingerprints. Impacts do more
than make craters. They break up, mix, heat, and melt the target
rocks. Individual mineral grains are damaged, leaving telltale signs
visible in an optical microscope. Meteoriticists have been studying
those effects for many years and this work contributed to development
of a scheme for classifying the effects of shock (the rapid, temporary
rise in pressure resulting from a high-velocity impact) in chondrites.
The classification was devised by Dieter Stöffler (Institute of
Mineralogy at the Museum of Natural History, Berlin, Germany), and Ed
Scott and Klaus Keil (University of Hawai'i). It uses the effects seen
in experimentally shocked samples and rocks found in terrestrial
impact craters to classify ordinary chondrites from unshocked (S1) to
heavily shocked (S6). The table below summarizes the classification
scheme and links the shock effects to shock pressure and the increase
in temperature caused by the shock.

Stages of shock effects in ordinary chondrites (from Stöffler et al.,
1991). S1 shows no effects of shock; S6 shows the most. The most
important criteria are listed in bold red. The classification is based
on changes in the properties of the minerals olivine and plagioclase
feldspar as viewed in thin sections using an optical microscope. Shock
pressure is listed in Giga Pascals (Gpa); 1 GPa is equal to 10,000
times the atmospheric pressure at the surface of the Earth.

Shock Stage Effects resulting from general shock pressure Effects
resulting from local P-T excursions Shock Pressure (Gpa) Minimum temp.
increase (^oC)
S1
unshocked Sharp optical extinction as viewed in microscope. Small
number of irregular fractures (cracks). None < 5 10
S2
very weakly shocked Undulatory (wavy) extinction, irregular fractures
None 5-10 20
S3
weakly shocked Olivine: Planar fractures, undulatory extinction,
irregular fractures
Plagioclase: Undulatory extinction Opaque shock veins; melt pockets,
sometimes interconnected 15-20 100
S4
moderately shocked Olivine: Mosaicism (weak), planar fractures
Plagioclase: Undulatory extinction, isotropic in places, planar
deformation features Melt pockets; interconnected melt veins; opaque
shock veins 30-35 300
S5
strongly shocked Olivine: Mosaicism (strong); planar fractures and
planar deformation features
Plagioclase: Maskelynite (isotropic feldspar) Pervasive occurrence of
melt pockets and veins; opaque shock veins 45-55 600
S6
very strongly shocked Olivine: Solid state recrystallization and
staining, presence of ringwoodite, local melting
Plagioclase: Shock melted Same as in stage 5 75-90 1500
Shock melted Entire rock is melted --- > 90 > 1500

The shock effects are visible using an optical microscope (some
effects are visible in hand samples). Geologists have been using
optical microscopy since the middle of the 1800s, especially after H.
C. Sorby began to examine thin slices of rocks in the microscope. Thin
sections are mounted on glass slides and ground to a thickness of only
30 micrometers (about one ten-thousandth of an inch). Most minerals
are transparent when ground so thin, but some are still opaque. In
chondrites metallic iron-nickel, iron sulfide, and assorted oxide
minerals are opaque.

Petrographic microscopes come equipped with polarizing filters, one
below the thin section and another above it. The filters are polarized
in opposite directions, so if no mineral is in the light beam, no
light gets through to the eyepiece-the first filter polarizes the
light in one direction, blocking other directions of polarization, and
the second filter does the opposite, thus blocking all the light.
Minerals also polarize the light, so when a mineral is placed in the
beam of light it changes the polarization direction so the light does
not become completely blocked by the second filter. Instead, an
interference color is produced. The colors are one of the diagnostic
tools used to identify a mineral.

In this era of high-tech, high-cost laboratory equipment, we sometimes
forget how useful a microscope can be. The shock classification
summarized in the table above is based entirely on optical microscopy
of thin sections. It uses several features in chondrites. At the
lowest shock category, olivine and plagioclase crystals have irregular
fractures (cracks). With increasing shock pressure, the grains develop
planar fractures, which are simply cracks lined up in one direction.
There is often more than one direction of fracturing in a single
mineral crystal, and the cracks are oriented along crystallographic
planes.

In polarized light, minerals blank out the light every 90 degrees when
the stage of the microscope is rotated. This is called "extinction."
(Minerals with cubic structures and uncrystallized glass are
isotropic, so do not further polarize the light, thus making them
appear dark at all orientations.) Unshocked olivine and plagioclase
have crisp extinction-each crystal becomes dark quite suddenly and
evenly when rotated. If shocked somewhat, the crystal structure is
deformed and the extinction is wavy, a feature called "undulatory
extinction." At higher shock pressure, the olivine crystal structure
is messed up even further, forming small (only a few micrometers)
domains that differ in their extinction positions. This causes a
characteristic appearance called mosaicism. With increasing shock
pressure, plagioclase shows increasing undulatory extinction, until at
a critical pressure it simply transforms to a disordered, isotropic
glass called maskelynite. Maskelynite is not melted. The shock wave
rearranges the atoms in a crystal so it no longer has a long-range
order to it. It is effectively a glass.

McKinney This shocked olivine crystal in the McKinney chondrite has
straight, parallel cracks called planar fractures. This is caused by
shock to a pressure of at least 15 GPa.
_________________________________________________________________

When an unshocked olivine grain is rotated in polarized light using a
petrographic microscope, it becomes extinct (dark) evenly, like this
one in Kernouvé (see movie, below, on the left). When shocked to a
pressure exceeding about 45 Gpa, however, olivine is severely damaged
and the extinction is spotty and irregular. This is called mosaicism
as in the movie of the chondrite Alfianello, below, on the right. The
field of view in each movie is 0.9 mm. Images by G. J. Taylor.

Unshocked Olivine        Mosaicised Olivine
movie of unshocked olivine movie of shocked olivine
_________________________________________________________________

Ringwoodite In some highly shocked chondrites olivine is converted to
a denser form called ringwoodite, which is blue in this
photomicrograph taken in polarized light of the chondrite Taiban.

A shock wave does not pass uniformly through a rock, so some regions
might be shocked more than others because of reinforcement by
reflected shock waves. This can create local pockets of and veins of
melted rock. Robert Dodd (State University of New York at Stony Brook)
and Eugene Jarosewich (Smithsonian Institution) showed in 1982 that
melt pockets tend to have higher aluminum than average in a chondrite,
reflecting the fact that plagioclase is easier to melt by shock than
other major minerals in chondrites. At high shock pressures large
portions of the rock begin to melt, starting with small veins, then
dike-like structures, and at the highest pressures almost total
melting of the rock occurs.
_________________________________________________________________

Impact Effects in Unshocked Chondrites?

Al Rubin has developed additional criteria to discern shock effects in
ordinary chondrites. The most important one to us here is the presence
of mixtures of the minerals plagioclase feldspar and chromite. In most
cases chromite grains are surrounded by plagioclase. These
chromite-plagioclase assemblages, as Rubin calls them, vary in size
(20 to 300 micrometers across), size of the chromite crystals in them
(0.2 to 20 micrometers), and abundance of chromite (10 to 70 volume
%). Rubin reports that chromite-plagioclase assemblages occur in
almost every chondrite shocked to shock stage S3 or above.

chromite-plagioclase assemblages
Examples of chromite-plagioclase assemblages in chondrites shocked to
stages S3 and higher. These photomicrographs of polished thin section
were taken in reflected light. Chromite is light gray, plagioclase is
dark gray. Olivine and pyroxene grains are medium gray. The size and
abundance of chromite varies in the assemblages. Al Rubin suggests
that the assemblages formed by preferential shock melting of
plagioclase, with the melt incorporating nearby chromite. (Recall that
Dodd and Jarosewich showed that melt pockets are enriched in aluminum,
hence incorporated more of the easily-melted plagioclase.)

Rubin reports that chromite-plagioclase assemblages also occur in
unshocked (stage S1) and lightly shocked (S2) chondrites. In fact,
they occur in all 210 unshocked and lightly shocked petrographic type
4-6 chondrites he examined. Assuming that chromite-plagioclase
assemblages result from shock, Rubin reasons that type 5 and 6
chondrites, even those apparently unshocked, were shocked to stage S3
or higher. In support of this, he notes that the chondrite Kernouvé
contains both chromite-plagioclase assemblages and veins filled with
metallic iron-nickel. It would appear that Kernouvé, a classic
unshocked type 6 chondrite with an old age (4.45 by the Ar-Ar method)
was shocked, heated, and then annealed to remove the effects of shock.
Rubin cites other convincing examples as well, including annealing
followed by a shock event. A particularly interesting case is the
MIL99301 chondrite, which has an age of 4.26 billion years. (The age
has been determined by Eleanor Dixon, Donald Bogard, and D. H.
Garrison at the Johnson Space Center.) Unless the parent asteroid for
ordinary chondrites was very large, it would have been cooled long
before 4.26 billion years ago and the ^26Al would have been long gone.
The heat for annealing MIL99301 must have been provided by impact.
Rubin uses these observations to conclude that shock events provided
the heat needed to metamorphose Kernouvé and other metamorphosed
chondrites.

shock stage S1 metallic iron-nickel vein

Photomicrograph in reflected light of chromite-plagioclase assemblage
in shock stage S1 chondrite Kernouvé (left). Silicate is dark gray,
metallic Fe-Ni is white, and troilite is light gray. Al Rubin argues
that shock Kernouvé was shocked long ago, but then annealed to remove
almost all evidence for the event. As supporting evidence, he shows a
prominent vein of metallic iron-nickel (white) with silicate grains
(dark gray) in Kernouvé (right). Such veins occur in heavily shocked
chondrites.
_________________________________________________________________

Some loose ends-unraveling the story?

Rubin uses his detailed microscopic observations of numerous
chondrites to make a good case for apparently unshocked type 5 and 6
chondrites having been shocked very early in their histories. Such
shock events might have left chondrite parent asteroids hot and thus
caused the metamorphism they clearly experienced. However, there are
some loose ends that need to be tied up before the idea is embraced by
meteoriticists.

One loose end is that Rubin's story requires that the shock heats
portions of an asteroid to the metamorphic temperature experienced by
type 6 chondrites, about 900 ^oC. If the initial temperature was 0
^oC, from the table above we see that such a temperature increase
would require shocking to between stages 5 and 6. This would result in
widespread veins, which are not observed in many unshocked type 5 and
6 chondrites, Rubin's excellent example from Kernouvé not
withstanding. More important, the olivine grains would be almost
totally mosaicized after the shock event. Rubin's interpretation is
that slow cooling after shock heating would have removed the mosaicism
through a process called annealing. Kernouvé and other type 5 and 6
chondrites cooled at rates of around 10 ^oC/million years. This would
seem to be plenty of time to remove the effects of shock. However, it
is not clear that large mosaicized olivine grains would convert back
to large single crystals. Annealing removes damage from crystals by
growing new, smaller, undamaged crystals, but the process is slow and
it has not been shown quantitatively that the process would result in
large, unshocked olivine crystals, even if cooling takes millions of
years.

Another loose end is the physical setting for the metamorphism and
slow cooling after the monumental shock event that Al Rubin envisages.
Most material in an impact is moved around, but not greatly heated or
shocked. Most of the shocked materials occur inside the crater
produced by an impact. Such craters cannot be too large or deep on an
asteroid because the bodies are only a few hundred kilometers across
to begin with. Even if we assume that an impact makes a crater 50
kilometers in diameter, the melted and heated zone on the crater floor
is no more than one kilometer deep. (This can be calculated using
equations in a book by Jay Melosh, University of Arizona). A depth of
1 kilometer sounds like a lot, but cooling rates of 1-10 ^oC/million
years require burial of tens of kilometers. Thus, an open question is
how a large volume of chondritic asteroid could be shocked and heated,
then buried deeply and cooled slowly. Rubin countered in an email to
PSRD that the situation might be different for asteroid heating. If
asteroids are porous rubble piles, as many meteoriticists and
astronomers think, impact energy is distributed differently,
concentrating in the walls and floor of a crater. If there were enough
impacts into such rubble piles, he argues, perhaps a significant
volume of an asteroid could be heated by impact. We clearly need a
better theoretical understanding of the effects of multiple impacts on
asteroids.

There is strong evidence that ^26Al was present in the early solar
system. Surely it must have affected chondrites to some extent. If
^26Al was the heat source, perhaps it could cause melting of
plagioclase when it was close to chromite, causing formation of the
chromite-plagioclase assemblages. Such assemblages form during shock
events, however, as Rubin shows in his paper, so it is not clear that
another mechanism is needed. If ^26Al contributed to heating an
asteroid, how would that heat change the way shock damaged the rock?
It seems that more work is needed before we understand the relative
roles of shock and ^26Al heating in the early solar system.
_________________________________________________________________

Widespread Smashing in the Early Asteroid Belt?

Almost all shocked chondrites are much younger than unshocked
chondrites. Meteoriticists have interpreted this to mean that
chondrites were not shocked severely when they formed or shortly
thereafter, but were affected later on. However, Rubin and other
meteoriticists have shown that some very old meteorites, the enstatite
chondrites, were shocked, even melted by shock, 4.5 billion years ago
(though they cooled very rapidly afterwards). During the Lunar and
Planetary Science Conference in March 2004, my colleague Ed Scott
summarized meteorite evidence for collisions early in the history of
the solar system, showing evidence from 10 different meteorite parent
asteroids. High velocity impacts clearly happened early in solar
system history. In many cases the impacting bodies were already
molten. What was the heat source for this melting? ^26Al? Impact? In
the early solar system, both might play important roles. Thus, Al
Rubin's ideas about impact heating do not focus on a narrow question
concerning the metamorphism of chondrites. It is part of the larger
puzzle of the role of impact and the earliest evolution of the
asteroid belt. We better tie up those loose ends!
_________________________________________________________________

ADDITIONAL RESOURCES 
Dixon, E. T., Bogard, D. D., Garrison, D. H., and Rubin, A. E. (2004)
39Ar-40Ar evidence for early impact events on the LL parent body,
Geochimica et Cosmochimica Acta, in press.

Dodd, R. T. and Jarosewich, E. (1982) The compositions of incipient
shock metls in L6 chondrites. Earth and Planetary Science Letters,
v. 59, p. 355-363.

Melosh, H. J. (1989) Impact Cratering. Oxford University Press, New
York. 245 p.

Rubin, A. E. (2004) Postshock annealing and postannealing shock in
equilibrated ordinary chondrites: Implications for the thermal and
shock histories of chondritic asteroids. [14]Geochimica et
Cosmochimica Acta, v. 68, p. 673-689.

Rubin, A. E. (2003) Chromite-plagioclase assemblages as a new shock
indicator; implications for the shock and thermal histories of
ordinary chondrites. [15]Geochimica et Cosmochimica Acta, v. 67,
p. 2695-2709.

Rubin, A. E. (2003) Northwest Africa 428: Impact-induced annealing of
an L6 chondrite breccia. Meteoritics and Planetary Science, v. 38,
p.1499-1506.

Rubin, A. E. (2002) Post-shock annealing of MIL99301 (LL6):
Implications for impact heating of ordinary chondrites.
[16]Geochimica et Cosmochimica Acta, v. 66, p. 3327-3337.

Scott, E. R. D. (2004) Meteoritic constraints on collision rates in
the primordial asteroid belt and its origin. In Lunar and
Planetary Science Conference XXXV, abstract [17]1990 (pdf file).

Stöffler, D. (2001) [18]Website: Shock classification of ordinary
chondrites, a manual for the determination of shock stages with
descriptions of thin sections of selected samples.

Stöffler, D., Keil, K., and Scott, E. R. D. (1991) Shock metamorphism
of ordinary chondrites. [19]Geochimica et Cosmochimica Acta, vol
55, p. 3845-3867.

[20]Web site about Henry Clifton Sorby, the man who pioneered
"Microscopical Petrography" in 1849.

Zinner, E. (2002) Using Aluminum-26 as a Clock for Early Solar System
Events. Planetary Science Research Discoveries.
[21]http://www.psrd.hawaii.edu/Sept02/Al26clock.html.
_________________________________________________________________

[22]home [ [23]About PSRD | [24]Archive | [25]Search | [26]Subscribe ]

[ [27]Glossary | [28]General Resources | [29]Comments | [30]Top of
page ]
_________________________________________________________________

[31]psrd at higp.hawaii.edu
main URL is [32]http://www.psrd.hawaii.edu/

References

1. http://www.psrd.hawaii.edu/index.html
2. http://www.psrd.hawaii.edu/PSRDabout.html
3. http://www.psrd.hawaii.edu/Archive/Contents.html
4. http://www.psrd.hawaii.edu/PSRDsearch.html
5. http://www.psrd.hawaii.edu/PSRDsubscribe.html
6. http://www.psrd.hawaii.edu/PSRDglossary.html
7. http://www.psrd.hawaii.edu/PSRDcomments.html
8. http://www.psrd.hawaii.edu/April04/PSRD-asteroidHeating.pdf
9. http://www.psrd.hawaii.edu/April04/PSRD-asteroidHeating.pdf
10. http://www.higp.hawaii.edu/%7Egjtaylor
11. http://www.sciencedirect.com/science/journal/00167037
12. http://www.sciencedirect.com/science/journal/00167037
13. http://www.psrd.hawaii.edu/Sept02/Al26clock.html
14. http://www.sciencedirect.com/science/journal/00167037
15. http://www.sciencedirect.com/science/journal/00167037
16. http://www.sciencedirect.com/science/journal/00167037
17. http://www.lpi.usra.edu/meetings/lpsc2004/pdf/1990.pdf
18. http://www.museum.hu-berlin.de/min/forsch/Schockklassifikation%20St%C3%B6ffler/Final%20Version/title.htm
19. http://www.sciencedirect.com/science/journal/00167037
20. http://www.shu.ac.uk/city/community/sorby/hcsorby.shtml
21. http://www.psrd.hawaii.edu/Sept02/Al26clock.html
22. http://www.psrd.hawaii.edu/index.html
23. http://www.psrd.hawaii.edu/PSRDabout.html
24. http://www.psrd.hawaii.edu/Archive/Contents.html
25. http://www.psrd.hawaii.edu/PSRDsearch.html
26. http://www.psrd.hawaii.edu/PSRDsubscribe.html
27. http://www.psrd.hawaii.edu/PSRDglossary.html
28. http://www.psrd.hawaii.edu/index.html#Resources
29. http://www.psrd.hawaii.edu/PSRDcomments.html
30. file://localhost/www/jnocook.net/saturn/files/asteroids/asteroidHeating.html#Top
31. mailto:%20psrd at higp.hawaii.edu
32. http://www.psrd.hawaii.edu/