Geology Reference
In-Depth Information
In semi-distributed models, the whole hydrological system is divided into different
blocks, each represented by a lumped model. Even now, serious debate is taking
place in the research community to establish the success rate of lumped models over
complex distributed models and vice versa. Though, the conceptual and physics
based models have given greater accuracy in terms of hydrograph modelling, there
were still many issues to be further addressed by many researchers. Those dif
culties
include implementation and calibration dif
culty, the vast amount of calibration data
and the need of sophisticated tools etc. [ 29 , 30 , 87 , 103 ].
1.2 Stochastic Modelling Case Studies in This Topic
This topic focuses on four major components in the hydrological cycle: solar
radiation, precipitation (rainfall), evapotranspiration and runoff. These are illus-
trated as four case studies towards the end of this topic. The following section gives
the current modelling status in rainfall-runoff modelling, solar radiation modelling
and evapotranspiration modelling.
1.2.1 Data Driven Rainfall-Runoff Modelling
Rainfall-runoff is a very complicated process due to its nonlinear and multidi-
mensional dynamics. Hence it is dif
cult to model. The ASCE Committee on
Surface-Water Hydrology (1965) introduced a new discipline incorporating statis-
tical consideration into hydrology named
the
manipulation of statistical characteristics of hydrologic variables to solve hydro-
logic problems on the basis of stochastic properties of the variables
Stochastic Hydrology
,de
ning
. This attempt
has made a drastic change in the conventional direction of research and has
encouraged many researchers to explore further the statistical and stochastic
properties of hydrologic time series which have de
nite physical causes. An
extensive review of the several types of stochastic models proposed for operational
hydrology can be found in Lawrance and Kottegoda [ 58 ], Franchini and Todini [ 34 ]
and Bras and Rodriguez-Iturbe [ 17 ]. Most of the black-box models include sto-
chastic components and just relate outputs to inputs through a set of empirical
functions, mathematical expressions or time series equations. The success rate of
these data based stochastic models always encouraged hydrologists to implement
simpler system theoretic models than the troublesome physics based or conceptual
model, which demand more implementation and calibration effort but with the
quality of the results comparable to the early mentioned stochastic models. In the
early days, research concentrated more on Autoregressive (AR) and mixed Auto-
regressive and Moving Average (ARMA) models [ 16 ]. Later, linear time series
models
like ARX (auto-regressive with exogenous
inputs) and ARMAX
 
Search WWH ::




Custom Search