The progress in hyperspectral imaging systems has opened up avenues for high resolution improved phenotyping in large scale field based breeding programs, providing a platform for dissection of complex traits linked to plant growth and adaptation. Studies at the International Maize and Wheat Improvement Center (CIMMYT) have shown the potential of spectral indices as secondary traits in improving prediction accuracy for grain yield. The aim of this study was to identify genomic regions associated with spectral indices that contribute to adaptation and grain yield. Aerial hyperspectral imaging of CIMMYT spring wheat breeding yield trials was conducted across 4 breeding cycles from 2014-2017 at Ciudad Obregon, Mexico. The yield trials consisting of 1092 lines were sown in three differentially managed treatments, bed sowing with full irrigation (BFI), flat sowing with drought (FDI) and late sowing for heat in beds (LHT). A haplotype based genome wide association mapping was performed on three spectral indices, Green Normalized Vegetation index (GNDVI), Red Normalized Vegetation Index (RNDVI), and Photochemical Reflectance Index (PRI). While GNDVI and RNDVI show significant association with grain yield in all three environments, PRI associated with grain yield in heat stress environment. Preliminary analysis identified genomic regions linked to the three traits that are consistent either in a single environment across years or multiple environments with effects ranging from 2-10%. In the BFI environment, a locus on chromosome 3A associated with GNDVI while the loci on chromosme 7A and 5B linked to both RNDVI and PRI. Under drought, loci on chromosomes 6A and 7B associated with GNDVI and RNDVI respectively. No consistent locus for PRI was present in FDI. In LHT environment, consistent genomic loci were identified on 2A, 3A, 3B, 6A and 7B chromosome for the three traits, of which the loci on 6A were consistent for both GNDVI and RNDVI. The loci on chromosome 3A and 5B also co-localized with the grain yield loci identified across multiple years with effects ranging from 3-6%. The results demonstrate that GWAS of spectral traits can provide additional insights to the genetic architecture of such complex traits.