This study explores the linkage of frozen hydrometeors (cloud ice and falling ice/snow) with sea ice and adjacent lands through surface energy budget using model-observation comparisons to quantify the roles of the falling ice radiative effects (FIREs) in determining the extent and thickness of Arctic sea ice and adjacent land surface radiation budget and land surface skin (Ts) and surface (SAT) temperatures. The Coupled Model Intercomparison Project Phase 5 (CMIP5) models without FIREs tend to produce underestimated downward longwave radiation, and overestimated shortwave downward radiation and surface albedo, resulting in too-cold skin temperature (TS) and surface air temperature (SAT) and overestimated sea ice concentration (SIC) and thickness (SIT). By comparing two simulations of late 20th Century climate from CESM1-CAM5 model with inclusion and exclusion of FIREs, it is found that TS, SAT, radiation, SIC, and SIT and their seasonal cycles are improved with the inclusion of FIREs. Exclusion of FIREs results in underestimated net downward longwave radiative flux, which is highly correlated with overestimated surface albedo, colder TS, and SAT with a confidence level at 99% (p < 1%). These biases in CESM1-CAM5 resemble those in CMIP5 models without FIREs. With the inclusion of FIREs, the SIC bias is reduced by 2%-15% in summer, while the SIT is improved up to 90% in winter despite little improvement in SIC. These findings suggest a potential link among the increased downward longwave flux, decreased downward shortwave flux, and decreased surface air and land surface temperatures locally, which then drives SIC melting and SIT thinning when FIREs are included relative to when FIREs are excluded. It is suggested that the roles played by FIREs over the Arctic regions are of the same order of importance as those over the Southern Oceans despite the differences in geography and impact of human activity.