Drumlins and their longer cousins, mega-scale glacial lineations (MSGL), are streamlined hills of till shaped beneath moving ice. Field data show that the faster the ice flowed, the more elongated the bedforms (Clark et al. 2009). This lab reproduces that trend with a deliberately simple bed model: till of thickness h(x,y) is dragged downstream by the deforming bed and spreads sideways,
dh/dt + v_d dh/dx = D (d2h/dx2 + d2h/dy2)
v_d = 0.05 * u_ice (till deformation velocity)
bedrock knobs pin h = H0 exp(-r^2 / 2 s^2)
Downstream of a pinned knob the steady solution is a parabolic plume: its Gaussian width grows as sigma_y = sqrt(2 D x / v_d), so faster ice (larger v_d) gives longer, narrower tails; high D gives wide, short ridges. The explicit upwind finite-difference solver is checked against this analytic law on a refined grid.
Honest scope. This is a toy advection-diffusion model, not the full ice-till instability theory (Hindmarsh 1998, Fowler 2009) and it does not produce the steep-stoss/gentle-lee asymmetry of real drumlins, subglacial water or erosion. The 5 % deformation ratio is an illustrative constant. Vertical scale is exaggerated.