Abstract: In order to establish the estimate model of chlorophyll（a+b）and chlorophyll a of tomato leaf，tomato variety Jinfen 2 was taken as test material，four soil water treatments were conducted in all growth stages in greenhouse of Nanjing University of Information Science & Technology from March to July，2013，and another repeat experiment was conducted from August to December，2013.The experimental data during March to July was used to establish estimate model of chlorophyll（a+b）and chlorophyll a，while another one was used to test the model.The soil water content was divided into four levels:control（75% to 80% of field water holding capacity），light drought treatment（LT，55% to 60% of field water holding capacity），moderate drought treatment（MT，40% to 45% of field water holding capacity），and severe drought treatment（ST，25% to 30% of field water holding capacity）.Tomato canopy spectral reflectance of each growth period under each water treatments were determined by using ASD portable spectroradiometer（US），at the same time，chlorophyll（a+b）and chlorophyll a content of leaf were also determined.Then by fitting a variety of common hyperspectral vegetation index models between canopy reflectance and chlorophyll relationships，four models for chlorophyll（a+b）and chlorophyll a content of tomato were established.The results showed that water stress caused chlorophyll（a+b），chlorophyll a and canopy spectral reflectance significantly decrease，moreover，their decrease trend increased with the increased degree of water stress in all the growth stages.With the advancing of tomato growth stage，both chlorophyll（a+b）and chlorophyll a were increasing，while the canopy spectral reflectance decreased in red and blue band wavelength.By comparing the effect of four models，R670 model was the best for chlorophyll（a+b）and chlorophyll a content estimate.Therefore，chlorophyll（a+b）and chlorophyll a content estimate model were established as follows:Cchl（a+b）=44.83R670+7.36，Cchla=39.92R670+5.12.Both of them passed the 0.01 significance level test，and their root mean square error（RMSE）were 0.45，0.42mg·g-1 respectively.Therefore，the way of using hyperspectral remote sensing data to estimate the chlorophyll content of tomato leaves was feasible.

Key words: Hyperspectral, Chlorophyll, Tomato, Water stress