Why? Antarctica, or even Wyoming provide far more Mars-like environments than anything Luna could provide, and at far less expense. Mars has an atmosphere, and a nearly 24-hour day. The Moon has no atmosphere and a 672 hour day. While Earth's gravity is about 2.5 times that of Mars, Mars's gravity is 2.5 times that of the Moon. In addition, the types of science you would do on Mars resemble the science which you would do on Earth far more closely than the scientific research able to be carried out on Luna. The Moon has no value as a training base for Mars missions. NEOs have even less.

While it is possible to mine oxygen from lunar rocks for rocket propellant or electrolyze water from the lunar poles to get propellant, it does not help for Mars flights to refuel at Luna. Even if infinite quantities of rocket fuel were sitting on the lunar surface right now - and they aren't - it would be, at best, ridculously illogical to refuel a Mars-bound spacecraft at the Moon. Why? Because before you can refuel at the Moon, you have to get there. Getting from the surface of the Earth to the surface of the Moon actually requires more Delta-V than going from the surface of Earth to the surface of Mars. So, from a propulsive point of view, it is much easier to send a spacecraft directly to Mars than to send it to the Moon first. Refueling your Mars craft at the Moon is like having a flight from London to New York stop over in Moscow for refueling. It just makes no sense.

If one small (Prometheus-class) nuclear reactor is available, a complete human Mars mission becomes available with three Saturn V-class boosters. The book The Case for Mars fully explains this in full, but I will summarize the main concepts of the Mars Direct plan here.

First, a Saturn-V class booster launches a 40-tonne unmanned payload to the surface of Mars on a 258-day Hohmann transfer. This craft, called the Earth Return Vehicle, or ERV, lands at the future landing site of the human Mars mission. It is an unfueled methane/oxygen propellant two stage ascent and Earth return vehicle. It lands with a 100 kWe nuclear reactor, a light truck, a set of compressors, and a chemical processing unit. The truck is telerobotically driven a few hundred meters away, and it deploys the nuclear reactor which will power the chemical processing unit. Six tonnes of hydrogen are brought from Earth, and it is reacted with Martian carbon dioxide to produce methane and water. The water is electrolysed to provide oxygen, both as air for the crew and as propellant, and the hydrogen is recycled into the system. This produces 48 tonnes of methane and 24 tonnes of oxygen. Martian carbon dioxide is disassociated to provide 36 more tonnes of oxygen. 108 tonnes of methane/oxygen propellant is now available, as is nine tonnes of water. 96 tonnes of propellant will be used to fuel the ERV, and the remaining 12 tonnes are available for internal combustion engine rovers. Now, from one Saturn-V class booster, we have a fully fueled Earth Return Vehicle, complete with water and air, waiting for the crew on Mars, one which has survived a landing.

Twenty-six months later, in the next launch window, two more Saturn-V class boosters are sent towards Mars. One carries another ERV for the next landing site, and the other is a "Hab", the crew's vehicle. It carries a crew of four, an internal combustion pressurized rover, two light rovers, and 500 kilograms of scientific equipment (500 more are on the ERV, making for a total of one thousand kilograms of scientific equipment).
Provisions for three years are carried. After launch, a cable is extended between the Hab and the upper stage of the booster, providing artifical gravity. The total mass of the Hab is 28 tonnes and it can be sent on a 180-day trajectory.

This is an extremely safe plan. The crew has a rover with a one-way range of 1,000 kilometers if they do not land right next to the ERV. Even if they land on the other side of the planet, the second ERV can be targeted to land near it. Even if both of them miss, the crew has provisions for thee years and can just wait until another ERV can be sent out. The plant is extremely robust.

The crew stays on the surface for 1 1/2 years, and there is ebough rover fuel available for about 22,000 kilometers of traverses. Thus, each mission can explore about 800,000 square kilometers during their stay. The Mars Direct harware can be easily modified to accomplish lunar missions, also.

So, we have a complete Mars exploration mission with nothing more than three Saturn Vs, some chemical engineering that has been around for over a century, and some present-day technology. No fantastical schemes are required for a Mars mission. We can do this, now. So, in contrast to your statement, while chemical rockets are certainly nowhere near as good as nuclear rockets, they certainly are sufficient, at least initially. The time it would take to convince the public of the safety of NTRs is not worth postponing our first Mars missions for.

This is possibly the statement I disagree most strongly with. The limitations of Mars robotic exploration have to be placed in context. Robots are fine on Mars for photographic surveys, seismology, meterology, and limited geochemical science. But searching for fossils, let alone extant life, requires intelligence and versality of an entirely different type. Fossil hunting requires heavy work - to do it for real, digging trenches and hammering rocks is required. So is fine work- splitting layers of shale to look for life between layers. It also requires complex perception. Sojourner and Athena have no manipulative capablities. Athena cannot travel over 100 meters a day. Both types of rovers, indeed, most any type of robot would be stopped dead by a boulder field or slope that would easy for a five year old.

Rovers and robots of other types are no substitute for real live scientists. You could parachute thousands of MER-type rovers onto Earth, and it is a fair bet that they might not find any fossils, at least not before the arrival of the next ice age, when they would be crushed by the glaciers which they would not be able to outrun.

Looking for extant life has much greater demands. No robot that will available in the next fifty years will be able to find groundwater or do any serious subsurface studies of Mars, where the life or paast life is likely to be found. First, a spot must be chosen with radar. Then a drilling rig must be set up - and the complex operations required to set one up totally rule out the robots which will be available for at least the next 25 years. And even if a robot could do this, it is doubtful that it could then analyze the life and the context of where it was found. But a human can easily do all of these, and more. A trained geologist's eyes are orders of magnitude beyond what a rover could tell us. Take the Mini-TES instrument on MER. It tells us of the composition of rocks. Now, a trained geologist could operate such an instrument (of far more complexity, owing to the 500 kilogram margin for scientific equipment) and tell us exactly what it was we are looking at. No robot could do that, because robots cannot think. Geology done by robots, as well as biology, comes nowhere near the level achievable by humans. Could robots hike through a mountainous rocky area, set up complex instruments, describe what they are seeing, know how that landscape formed, and pick up and examine rocks? No. Could they pick up rocks of special interest? No. Could they distinguish important geological finds from unimportant ones? No. Robots cannot think; scientists can.

As a final example, take an Apollo mission, Apollo 15. The entire ALSEP would probably be too complex for modern robots to set up. Could these same robots than explore the region? It is doubtful that such finds as the Genesis rock or things like removing stuck drills from the surface could be done by robots. On Mars, humans will simply be so much better than robots that comparisons become useless. Even collecting samples and core samples on Mars (hundreds of pounds at least) would be far beyond the ability of robots. And could these robots also explore the region, take thousands of pictures, and generally perform all of the scientific functions that humans can?

In a word, no.

__________________
"Too low they build, who build beneath the stars". - Edward Young, 1745