Drought　is　one　of　the　most　complex　and　harmful　natural　disasters.　A　study　on　the　temporal　and　spatial　patterns　and　the　evolution　of　drought　can　provide　a　scientific　basis　for　predicting　drought　occurrences.　Based　on　a　multi　source　dataset,　we　select　a　suitable　control　drought　indicator　for　improving　the　vegetation　health　index　(VHI),　optimize　its　algorithm　through　Pearson　correlation　analysis,　and　compare　the　VHI　performance　before　and　after　the　improvement　for　various　vegetation　types.　Results　show　that　(1)　the　self-calibrated　Palmer　drought　severity　index　is　more　suitable　than　the　standardized　precipitation　evapotranspiration　index　for　improving　the　VHI;　(2)　the　contribution　of　the　thermal　condition　index　to　the　VHI　in　most　parts　of　the　world　is　higher　than　that　of　the　vegetation　condition　index;　(3)　the　enhanced　VHI　significantly　improves　the　detection　of　vegetation　drought;　and　(4)　vegetation　drought　events　occurring　in　high　latitudes　tend　to　worsen,　and　the　response　of　different　vegetation　types　to　drought　is　significantly　different.　Our　research　presents　a　step　forward　in　improving　the　effectiveness　of　the　VHI　in　detecting　vegetation　drought　and　thus　its　application　prospects.　Furthermore,　the　response　characteristics　of　various　vegetation　types　to　drought　are　identified,　deepening　our　understanding　of　vegetation　drought,　which　may　help　decision-makers　and　authorities　to　develop　better　mitigation　and　adaptation　strategies　to　reduce　losses　caused　by　these　events.