Description
Downscaling methods are used to derive stream flow at a high temporal resolution from a data series that
has a coarser time resolution. These algorithms are useful for many applications, such as water management
and statistical analysis, because in many cases stream flow time series are available with coarse
temporal steps (monthly), especially when considering historical data; however, in many cases, data that
have a finer temporal resolution are needed (daily).
In this study, we considered a simple but efficient stochastic auto-regressive model that is able to
downscale the available stream flow data from monthly to daily time resolution and applied it to a large
dataset that covered the entire North and Central American continent. Basins with different drainage
areas and different hydro-climatic characteristics were considered, and the results show the general good
ability of the analysed model to downscale monthly stream flows to daily stream flows, especially regarding
the reproduction of the annual maxima. If the performance in terms of the reproduction of hydrographs
and duration curves is considered, better results are obtained for those cases in which the
hydrologic regime is such that the annual maxima stream flow show low or medium variability, which
means that they have a low or medium coefficient of variation; however, when the variability increases,
the performance of the model decreases