ABSTRACT
Since I last reviewed the modeling of harmful algal blooms (HABs) (Franks 1997), great strides have been made. The advances have been the result of three primary factors: 1) the increased power and decreased cost of computers, 2) improvements in sensors and assays and improved fi eld sampling, and 3) the increased training of interdisciplinary scientists. In the last decade modeling of HABs has increasingly incorporated fi eld data —physical, chemical and biological-and has advanced to the point that operational forecasts are available for HABs in persistently affected regions. Two decades ago HAB models were mostly heuristic tools that were used to explore theoretical scenarios in attempts to explain HAB occurrence and patchiness. In the past decade model usage has matured signifi cantly to the point that heuristic models now form a small fraction of the models published and used.
