Global warming and climate change lead to extreme agricultural drought events which exceedingly affect agricultural production in many countries. The agricultural drought in TienGiang Province, Viet Nam, which can be induced by inconsistent rainfall, a decrease in the freshwater resource, causes the decline in the yield of crops and affects the local economy. Hence, the detection and assessment of agricultural drought are urgent to mitigate its negative effect. Tien Giang Province, the study area, which falls in the lower section of the Mekong River Delta of Viet Nam, has a relatively gentle topography surface at low altitude and is consequently considered a drought-prone and salinity intrusion region. The objective of the study is to analyze the vegetation stress to detect agricultural drought in the province with the calculation of Normalized Difference Vegetation Index (NDVI), Vegetation Condition Index (VCI), and Land Surface Temperature (LST). This study used the Landsat 8 OLI/TIRS satellite data, CGLS-LC100 land cover map, ground-based air temperature record in 2015. To map the land cover types, the random forest supervised machine learning classifier was applied in this study. The result shows that the estimated LST was higher than the ground-based air temperature, around 4.40C, with the max difference at 8.40C. The random forest classifier categorizes the land cover into five classes (e.g., aquaculture, build-up, annual vegetation, perennial vegetation, and water) with an overall accuracy of 97.4%. The correlation between LST and NDVI, VCI can be clearly remarked as negative relationships with a correlation coefficient between LST and NDVI by -0.439 (p< 0.01), LST and VCI by -0.483 (p< 0.01) for the vegetation region of the study area.