What we don't know is the shape of the hull, so the intention is to arrive at an educated guess  by testing the drifting performance with various hull shapes. The midsection is assumed to be relatively rectangular with a possible bilge radius, so the biggest factor is the shape of the bow and stern.
Noah's Ark had the proportions of a true ship. The ratio given in Genesis 6:15 can't be faulted; 300 x 50 x 30 cubits. See Korean research on Noah's Ark. 
There are dozens of shape parameters that can be modified. The most fundamental factor is the "blocky-ness" of the hull, termed the Block Coefficient Cb. This is simply the ratio between the hull's volume and it's bounding rectangular box, L x B x D. A more streamlined hull will typically have a lower Cb. Some example Block Coefficients: Pure Box Cb=1, Cargo Ship Cb=0.56-0.85 , Queen Mary 2 Cb= 0.6, a typical canoe Cb= 0.4
A high block coefficient allows more room for cargo, but tends to give a rougher ride. For example, a very blunt bow will slam into a wave rather than pierce through it. If the wave is big (compared to the boat) this may be very uncomfortable or even dangerous. A higher Cb also increases hull loads like wave bending moment and slamming.
Cut some foam (rigid polyurethane). It is actually a reject block of foam - a bit of porosity and non-uniform density, but it will do...
Glue some marine ply to the end (8mm 7 ply). (No we didn't glue CJ's hand.)
Stern waterline at weatherdeck. The block is attached to the midsection using 1/4" bolts. (x4)
Marking out. The 1 degree stern-down trim was a bit more work, so had to set up waterline #6 as a "horizontal" datum.
Using body lines to trim the hull. This is station 20 and 21. The plans were plotted at the model scale (It would be 1:76 according to the common 18" cubit). These lines were developed by Dr Allen Magnuson using Vacanti. The model is about one foot (300mm) wide.
Stern profile at centreline. So there really is a use for tracing paper after all - and I thought I'd never use it again.
Stern waterlines all met at around 45 degrees. Makes it easy.
Sanded one side.
Foam done. The block is reduced approx 1-2mm all round to account for fiber-glass thickness.
Applying woven fiber-glass using polyester resin.
Prior to glassing the top, a hole was made for provision of weights. Although the desired draft could be obtained by adding mass in the midsection, the short distance to the center of mass could make the moment of inertia too low in pitch motion. The first hole shown below was too small - I realized there may be up to 1/4 of the hull mass concentrated here - or 10kg (22lb). A larger container was substituted - a cylinder with diam 127x152mm (diam 5" x 6") which gives a volume of 1930 cm3 ( 65.3 oz). For the (unlikely) full 20kg this would require a material of specific gravity around 5 which is midway between steel (7.8) and aluminium (2.5). So steel weights can be used - bolted down of course (planning to merrily capsize the thing eventually).
Repeat but with a different set of body lines.
Some "hollow" lines almost forming a gentle skeg.
Bilge radius inserts (Dark Red Meranti) 
Thread inserts for connection to midsection. (Helicoil 1/4" UNC S/S inserts)
Assembly - minus the bilge radius inserts. These will attached using adhesive tape.
Hope it floats... (I wonder if Noah thought that.)
24 March 2005: Pool Test
Ballast added equally to midsection - 8 x 2.8kg bricks which adds 22kg - no attempt to set correct KG at this stage, nor adjustment of roll or pitch inertia.
The CAD profile illustrating stern down trim equivalent to approx 1 degree. Since in this case the decks are level it is not called "trim" but "drag", so the correct term is "6 ft drag to stern". What does that achieve? Well - a transverse center of water pressure located aft of amidships of course - once we stick the skeg on of course..
With ballast distributed equally, the reduced buoyancy of the stern (right hand side) gives the model almost the exact trim predicted (Water surface matches waterline 6 here). Hey, fluked it.
Recording the gyromouse movement. The model was also fully submerged to check for leaks. Yes it did leak a bit - lack of gasket pressure due to flexure of polycarbonate lid between screw fasteners. No big surprise, PC has E=2.7GPa, Steel is 210GPa, so steel is almost a hundred times as stiff as plastic. Need to add some steel reinforcement since I don't want any more screws.
That's about enough megabytes for this page...
A team of nine Korean researchers, including a leading research naval architect
at KRISO and a professor of structural
engineering, studied the effect of altering the proportions of Noah's Ark. Their
conclusion was that the Ark was near optimal, and that realistic timber sizes
were capable of handling wave loads in extreme seas. "Safety
Investigation of Noah’s Ark in a Seaway" by
S.W. Hong, S.S. Na, B.S. Hyun, S.Y. Hong, D.S. Gong, K.J. Kang, S.H. Suh, K.H.
Lee, and Y.G. Je. First published in: Creation
Ex Nihilo Technical Journal 8(1):26–35, 1994. Available here: http://www.answersingenesis.org/home/area/Magazines/tj/docs/v8n1_ArkSafety.asp
Indexed Version here: http://www.worldwideflood.com/ark/safety_aig/safety_aig.htm
Return to text
2. Some commentators claim that "Ark" means box, therefore Noah's Ark was a box shape. However, the English term "Ark" comes from a Latin translation that followed the Greek Septuagint tradition of using the same word for Noah's Ark and the Ark of the Covenant. The original Hebrew uses completely different words for each. Confused? See Does Ark mean Box?. Return to text
3. Yoshiho Ikeda Prediction Methods of Roll Damping of Ships and their Application to Determine Optimum Stabilization Devices. Available in pdf format at http://web02.webb-institute.edu/stability/technical.php, the Ikeda paper is part 10. Return to text
4. Meranti (known as Pacific Maple in Australia) is usually classified into 2 groups according to density. The heavier timber is Dark Red Meranti. http://www.timber.net.au/species/philippine_mahogony.htm . Unlikely candidate for gopher wood, but OK here. Return to text