Effect of Temporal Resolution of NDVI on Potential Evapotranspiration Estimation and Hydrological Model Performance

Key words: 
• NDVI /
• temporal resolution /
• potential evapotranspiration /
• BTOPMC model

Abstract: Normalized difference vegetation index (NDVI) data, obtained from remote sensing information, are essential in the Shuttleworth-Wallace (S-W) model for estimation of evapotranspiration. In order to study the effect of temporal resolution of NDVI on potential evapotranspiration (PET) estimation and hydrological model performance, monthly and 10-day NDVI data set were used to estimate potential evapotranspiration from January 1985 to December 1987 in Huangnizhuang catchment, Anhui Province, China. The differences of the two calculation results were analyzed and used to drive the block-wise use of the TOPMODEL with the Muskingum–Cunge routing (BTOPMC) model to test the effect on model performance. The results show that both annual and monthly PETs estimated by 10-day NDVI are lower than those estimated by monthly NDVI. Annual PET from the vegetation root zone (PET_r) lowers 9.77%–13.64% and monthly PET_r lowers 3.28%–17.44% in the whole basin. PET from the vegetation interception (PET_i) shows the same trend as PET_r. In addition, temporal resolution of NDVI has more effect on PET_r in summer and on PET_i in winter. The correlation between PET_r as estimated by 10-day NDVI and pan measurement (R²= 0.835) is better than that between monthly NDVI and pan measurement (R² = 0.775). The two potential evapotranspiration estimates were used to drive the BTOPMC model and calibrate parameters, and model performance was found to be similar. In summary, the effect of temporal resolution of NDVI on potential evapotranspiration estimation is significant, but trivial on hydrological model performance.