Drought is one of Canada’s costliest natural disasters. Climate change is expected to increase the frequency, extent and intensity of droughts. Canada has an effective monitoring and reporting system for current conditions (known as the Canadian Drought Monitor) that uses a convergence of evidence approach to combine data sets disparate in time and spatial coverage to provide the best estimate of current drought conditions. This system combines multiple drought indicators calculated from meteorological station data and a set of earth observation indicators on soil moisture, groundwater and evapotranspiration to evaluate drought severity on a monthly time scale. There is an increased requirement for data that is updated more frequently and provides a higher spatial accuracy. While numerous objective indicator blends have been evaluated in Canada and elsewhere, these have not been able to provide a robust indicator of drought that is consistent with drought assessment through the Canadian Drought Monitor. A method is under development which uses machine learning, hydrological variables from earth observation data and weather data modelled using the Regional Deterministic Prediction System from the Meteorological Service of Canada. The model was trained using historical Canadian Drought Monitor Data from 2003 to 2022 and a leave one out cross validation analysis was used to evaluate model performance. Early results indicated this approach shows promise for providing updates on drought conditions between assessment periods. Earth observation data sets provide strong prediction of drought severity in different regions and over different time periods.