Step selection functions use animal movement patterns to estimate habitats available at each step along a movement track. These data were generated from four species of waterfowl (Anser albifrons [greater white-fronted goose], Anser caerulescens caerulescens [lesser snow goose], Anas platyrhynchos [mallard], Anas acuta [northern pintail]) using the Central Valley of California 2016 to 2022 that were fit with tracking devices collecting GPS data every hour (or subset to hourly locations). Observed movements were used to estimate a 2-mixture log-normal step length distribution for each species. The shorter mean step length mixture generally reflected step lengths consistent with individual immobility and GPS errors. The larger mixture reflected movement within and across habitats in the landscape. Using these 2-mixture distributions, 100 random movements from the origin of each observed step along an animals track were generated to estimate habitat availability at each location along an animals track. Steps corresponding to the larger "moving" class were retained for use in step selection analyses. Habitat classes were aggregated from existing data sources, primarily annual U.S. Department of Agriculture National Agriculture Statistics Service-Cropland Data Layer (NASS_CDL) data to identify landcover type and the California Protected Areas Database to identify regions potentially providing sanctuary or refuge from hunting activity.