For evapotranspiration (ETa) estimation, the solar radiation measurement equipment is expensive, it is hardly to deploy a large number of measurements, and the unit regression algorithm has low accuracy and poor generalization performance. An integrated algorithm based on illuminance was proposed to estimate ETa. Firstly, the illuminance instead of solar radiation was used as the input of the model, and a sunny index based on illuminance was proposed to improve the estimation effect. Secondly, an integrated algorithm that fused extreme gradient boosting model (XGBoost), light gradient boosting machine (LightGBM), random forest regression (RFR), support vector regression (SVR) was used to estimate the farmland actual evapotranspiration. The results showed that the illuminance could replace the solar radiation in the estimation of the actual evapotranspiration of farmland. The unit model and the integrated model were used to compare the ETa estimation results based on the illuminance and solar radiation, respectively. The maximum difference of root mean square error (RMSE) between the two methods was 0.031mm/h. The maximum difference of determination coefficient (R2) was 0.053. The sunny index helped the model better learn the distribution characteristics of evapotranspiration data under different weather conditions. Compared with the estimation result of the integrated model without adding sunny index, the RMSE was reduced by 0.028mm/h, and R2 was increased by 0.03. The performance of the integrated algorithm was significantly improved than that of the unit model algorithm. The optimal RMSE was 0.037mm/h and R2 was 0.985. The research explored the data sources, characteristic quantities and estimation algorithms required for estimating evapotranspiration, and provided a new idea for estimating farmland evapotranspiration.