Somewhere between reality and the unknown, science fiction has always flourished. The best sci-fi authors rigidly adhere to one principle: Make it as real as possible, given what's known. Now, as if lifting a chapter from an Isaac Asimov novel, NASA plans to create hundreds of "synthetic planets" that might represent real worlds orbiting faraway stars.

The cyber planets will be modeled on solid science, but the results are likely to be decidedly fantastic. No one knows what sort of worlds will be generated, and that's the whole point.

To be fair, the scientists do have a little to go on, but if they launched the spacecraft tomorrow, it could only be designed and programmed to look for signs of life that are known on Earth. Biological activity elsewhere could be markedly different, producing unexpected atmospheric signals that would be missed if a spacecraft weren't looking for them.

"We will be able to model the composition and appearance of planets very different to our own Earth, with parent stars very different to our Sun," says Vikki Meadows, who oversees the VPL from NASA's Jet Propulsion Laboratory.

Planets will be seeded with life forms, heated and tweaked so environments develop in myriad ways over eons.

One approach will start with a world that resembles Earth billions of years ago, stock it with known microorganisms that for whatever reason do not dominate today's biosphere, and let the critters run amok.

The virtual lab will generate a catalogue of potential light signatures, called spectra, that various combinations of biology (or lack thereof) would generate in a planet's atmosphere. The results will be compared with data collected on Earth and the other planets (and even some moons) in our own solar system, as a way to check the accuracy of the models and to continue refining them.

One important premise for the work is the simple fact that even around our own Sun, a basic set of ingredients produced dramatically different results. Earth and Venus, for example, are thought to have been initially sprinkled with roughly the same mix of water and organic material called volatiles.

Yet Earth became an "ocean-covered oasis" and Venus a "desert hell," Meadows points out.

Other scientists have speculated that Venus, nonetheless, might harbor microbes high in its clouds. And Mars, Jupiter's moon Europa and Saturn's moon Titan are all place that scientists are eager to look for possible life. But positive signatures of biological activity have so far eluded scientists. All this knowledge . along with the lack of it . will feed the VPL.

SPACE.com recently asked Meadows to explain the project in detail:

Vikki Meadows: When we create a planet, we will be using ingredients that we already know about, things like carbon, nitrogen, oxygen, silicon, but we will be looking at variations in the relative amounts of these ingredients, and how planetary processes distribute those ingredients in its core, surface and atmosphere.

Looking at our own solar system, you can see that the four earth-sized planets, Mercury, Venus, Earth and Mars (and Titan, as an honorary member of this group) all have basically the same building blocks, but apportioned in different ways, with spectacularly different characteristics as a result. As an even more specific example, Venus and Earth are believed to have been endowed with the same inventory of volatiles in the early history of the solar system. However, Earth locked up most of its carbon dioxide in sediments and in its crust, while Venus ended up with huge amounts of this greenhouse gas in its atmosphere.

Consequently one world is an ocean-covered oasis for life, and the other is a desert hell with 730 degrees Kelvin [854 degrees Fahrenheit] surface temperatures, and a poisonous, crushing atmosphere. Same ingredients, very different planet.

With VPL, we want to expand our understanding of the potential types of earth-sized planets, especially those with atmospheres, beyond the five examples we have in our own solar system. This tool will also allow us to study the same planet with and without an active biosphere, so that we can gain insight into signatures of living processes that might be detected with an astronomical observation.

VM: What we expect to be able to do is identify what combinations of observable planetary characteristics are likely to indicate a habitable world. That is, if you know the planet's size, distance from the star, orbital path, and something about its atmospheric or surface composition, we want to be able to provide a probability that that planet is habitable, and ultimately, be able to tell whether a planet is already inhabited.

A more specific thing that we are hoping to see from this modeling effort is a better understanding of what our own Earth looked like during its history.

This modeling effort will also help us to understand what the Earth might have looked like early in its history, when its environment was truly alien, but still supported abundant life. Today, the Earth has an atmosphere dominated by nitrogen and oxygen, but before 2.3 billion years ago, when the Earth was about half its current age, there was very little oxygen in the atmosphere. In spite of this, life apparently originated and thrived in that environment.

With the VPL, one of the first things we will tackle is modeling this "family of Earths", many of which are habitable worlds, but unlike the modern Earth. We will use information and constraints available from geology, chemistry and biology to create plausible early atmospheres for Earth, and understand what these might have looked like from space.

There may be extrasolar planets out there that are "early-Earth-like," and we need to understand how to recognize that.

VM: The first and best one we can think of for this will be the early Earth as described above. We'll need to gather information from many existing fields of research on the geological, chemical and biological processes working at that time, such as tectonic and volcanic activity, likely life forms, and any hints we have on surface temperature, atmospheric composition. We can put these best available constraints into the experiment design, and create a suite of possible worlds that are consistent with those constraints.

For the extrasolar planets, obviously we don't have any extra information from the geological record to help us constrain the atmosphere, so we will create a broad range of planets, and determine which ones could support liquid water on their surfaces, and have atmospheres which might be hospitable to any of the myriad of different life forms we already have on our own Earth.

VM: We won't be attempting to model "life as we don't know it." Because we want to understand which planets could support liquid water on their surfaces, we will model for life that requires both water and sunlight (although not necessarily the light from a G2 dwarf star like our Sun).

However, even with those constraints there's still a lot of "weird life" we do know about. When we populate one of our planets with "different" life, it will be with life forms that we know about, but perhaps in a different proportion to that which we have on the present day Earth, depending on the environment we are going to put it in.

An oxygen-dominated atmosphere supports a different suite of life when compared to an atmosphere rich in carbon dioxide or methane, for example.

SPACE.com: How many planets do you expect to create during the life of the project?

VM: The accurate answer is "as many as we can!" The less accurate answer is "hundreds."

The planets we create will be specified by running through a set of different planetary characteristics, such as size, distance from the parent star, orbital path around the star, what the planetary atmosphere is made of, the presence or absence of plate tectonics and other planetary processes, the presence or absence of life, and the nature of that life.

SPACE.com: Will you generate many planet types at once, or does each evolve from things you learn with each creation?

VM: We will initially develop a "grid" of model runs that samples a range of planetary environments that we expect might be habitable. As we run through these, I expect that we will learn a great deal about which of the model planets are habitable or not.

The next round of modeling will then concentrate around the characteristics of the best habitable candidates, and will work more on variations on that theme, including the addition of life.