The invention discloses a crop biomass inversion method based on a SEBAL-HJ model. The crop biomass inversion method is characterized by comprising the following steps: S1, acquiring HJ-1 CCD, IRS images and elevation data DEM in a research plot, and carrying out geometric fine revision to the images; S2, inversing a land surface albedo, NDVI, SR, land surface emissivity and inversing land surface temperature; S3, calculating net radiation flux according to inversion results in the S2, and further calculating soil heat flux on the basis; S4, calculating sensible heat flux through cyclic reduction; S5, calculating the coefficient of evaporation ratio and obtaining evapotranspiration per day through space-time dimension expansion; S6, inversing fPAR on the basis of vegetative cover indexes NDVI and SR; S7, inversing NPP according to the inversion results of evapotranspiration per day in the step S5 and the inversion results of fPAR in the step S6; and S8, obtaining crop biomass through the accumulation of NPP after space-time reconstruction. According to the invention, the high-precision inversion of crop biomass is achieved, and the amount of calculation is relatively smaller.