Chemotaxis is the directed motion towards a chemical attractant. Many bacteria chemotax by swimming repeatedly in a randomly chosen direction and biasing their swim lengths according to whether their environment is improving in the current direction. At a macroscopic level this biased random walk has been modeled by the Keller-Segel (K-S) equations which are conservation laws that have a bacterial flux with a component proportional to the gradient of attractant concentration. The K-S equations predict that bacteria will aggregate at the maxima of the attractant concentration, but this is not always observed. For rapidly spatially-varying concentration gradients, the peak in bacterial concentration is some distance away, lying on a ring in two-dimensions. This is the ”volcano effect”. Our work, starting from a simplified biochemical description of each bacterium and then extracting population level models, shows how to bridge these two regimes (K-S and volcanic). The results are verified against stochastic simulations of virtual bacteria. We shall also discuss applications to the more complex chemotactic process where the bacteria are themselves producing the chemoattractant.