We study the joint problem of fleet sizing and tactical deployment in aerial wildfire suppression under wind uncertainty. The strategic decision selects a heterogeneous fleet of helicopters, amphibious water scoopers, and land-based airtankers, and assigns these airframes across operational bases. The tactical decision routes each airframe over a type-specific time-space-state network and schedules water and retardant drops, forming a contiguous containment line anchored to non-flammable terrain. We formulate the problem as a two-stage stochastic integer program in which the second stage couples a wind-driven cellular-automaton fire model with a multi-commodity routing problem and a discrete control-line construction problem. Wind scenarios are sampled from a regime-switching vector autoregression with Gaussian-process spatial downscaling that captures temporal regime persistence and spatial coherence of the wind field. We solve the program by an exact integer L-shaped decomposition strengthened by monotonicity-based dominance cuts that certify entire down-sets of the binary fleet lattice per scenario evaluation, with each scenario subproblem evaluated as a compact mixed-integer program. Computational experiments on a calibrated multi-type instance show that the cost-minimising fleet pairs a single helicopter as an early-response unit with three land-based airtankers as a perimeter-installation unit, exploiting the complementary roles of temporary water suppression and permanent retardant lines while satisfying airspace non-overlap constraints.