Affected by the global warming, the frequency and intensity of drought events have shown a significant increase in recent years, which has seriously affected crop yields. Therefore, selecting appropriate monitoring indicators and constructing accurate yield estimation models are of great significance to ensure the country’s food security. The Guanzhong Plain in Shaanxi Province was chosen as the study area, and remotely sensed vegetation temperature condition index (VTCI) and leaf area index (LAI) which are closely related to crop growth were selected as the growth monitoring indicators. The principal component analysis (PCA) combined with Copula function were used to construct univariate (VTCI or LAI) and bivariate (VTCI and LAI) winter wheat yield estimation models at the county scale. The results showed that the liner regression model of comprehensive values of LAI and winter wheat yield constructed based on PCA-Copula had the highest accuracy (R2=0.567, P<0.001). However, when the liner regression model of comprehensive values of VTCI and LAI and winter wheat yield (R2=0.524, P<0.001) was used to estimate the yield of winter wheat in each county (district) in the study area from 2012 to 2017, the distribution of the error between the estimated yield and the actual yield was more concentrated, and the maximum error value was also smaller, which was more reliable than the results of the winter wheat yield estimation model based on a single variable. These results indicated that the bivariate yield estimation model constructed by PCA-Copula was more suitable for largescale winter wheat yield estimation.