Figure 1 . Summary of the literature review on 62 studies that
incorporated RS datasets for parameter estimation in hydrological models
(see Table S1 in Supporting Information). (a) Classification of
publications based on the drainage area of study sites (an average value
was considered for publications that used multiple study sites); (b)
distribution of studies based on the calibration variable; (c)
geographical distribution of study sites; (d) number of publications per
year; (e) number of RS products involved in calibration (in black),
number of independent calibration variables (in blue), and number of
model outputs evaluated (in red); (f) classification of models based on
their spatial configuration; (g) model type; and (h) use of RS data
Aims and Contributions of this
paper
Our study addresses major knowledge gaps identified in the previous
literature review in the context of RS-based calibration of hydrological
models. Firstly, most of the studies analyzed two or less variables
(Figure 1e). Here, we used RS observations of a large number of
variables for model calibration, namely soil moisture,
evapotranspiration, terrestrial water storage, flood extent and river
water levels, and thus move beyond the contributions of RS for improving
only discharge estimates. By simultaneously looking at different
variables, we also move towards an improved representation of the water
cycle as a whole, enhancing our ability to identify model limitations
and inconsistencies. Furthermore, most studies to date focused on
European, temperate watersheds (Figure 1c), which largely differ from
tropical basins in terms of hydroclimatic characteristics and
river-wetland interactions. In this context, large-scale, coupled
hydrologic-hydrodynamic models have faced major developments in recent
years (Yamazaki et al 2011, Paiva et al 2013, Fleischmann et al 2020),
but to our knowledge the complementarity of hydrologic (soil moisture,
evapotranspiration, terrestrial water storage) and hydrodynamic (flood
extent and river water level) RS observations for model calibration has
not yet been addressed in the literature. Here we present a study case
in a tropical basin with extensive floodplains in the Amazon with a
state-of-the-art coupled hydrologic-hydrodynamic model, which together
with the previously mentioned advances provide important contributions
to the growing literature of RS-based calibration of hydrological
models.