1. Master in emergency early warning and response space applications
Perugia, Italy, July 2015
Rainfall-Runoff model implementation on the San Antonio
River catchment for flood prediction using soil moisture from
radar satellite images; rainfall and discharge data
Instituto de Altos Estudios Espaciales MARIO GULICH / CONAE
Istituto di Ricerca per la Protezione Idrogeologica (IRPI) - Consiglio Nazionale delle
Ricerche (CNR)
Author: Cristian Gonzalo Guerrero C´ordova
Tutors: Tommaso Moramarco - Romina Solorza
2. SCOPE OF THIS DOCUMENT
This document illustrates the research activities developed in the framework of the stage in Italy,
according to the AEARTE master’s study program, managed between the Comisi´on Nacional de
Actividades Espaciales (CONAE) of Argentina and the Agenzia Spaziale Italiana (ASI).
Tasks at this stage, have been carried out in the Istituto di Ricerca per la Protezione Idrogeologica
(IRPI), under the supervision of Dr. Tommaso Moramarco, from January to July 2015 in Perugia, Italy.
1
5. Introduction
The basin of the Rio San Antonio is characterized specially for converging on a tourist center of
principal importance, the city of Villa Carlos Paz and his surroundings. In the summer season (December
- March) this river has numerous places where thousands of tourists go to swim. It is in the same period
of the year that the rainfalls has major intensity and frequency in the whole basin. This drives to a great
problematics issues regarding to the risk of human lives during and after the storms. Mainly considering
that rainfalls often take place in the high part of the basin producing sudden rises of the river (flash
floods).
To reduce flood losses one strategy can be of developing real-time flood forecasting systems that reduce
flood risk by issuing warnings (with the complementary strategy of the education of the public on the
appropriate response to warnings). Rainfall-runoff (RR) models play a central role being a component
of a real-time flood forecasting system in small to medium catchments.
This report illustrates the scientific background data, models and procedures that have been
considered to cope with the above matters during the term at Irpi.
4
6. Chapter 1
Objectives
1.1 Main Objectives
Rainfall-runoff (RR) model implementation on the San Antonio River catchment for floods forecasting,
using satelite soil moisture products, rainfalls measurements and discharge data.
1.2 Specific Objetives
• Learn of the theorical knowledge necessary to understand basic concepts, on which hydrological
models are based.
• Study of a hydrological model to be applied to the catchment of San Antonio River.
• Collection of hydro-meteorological and analysis of data format selection of satellite soil moisture
product.
• Selecion of hydrological model and application to the catchment study area.
• Analysis of results.
• Conclusions
5
7. Chapter 2
Theorical Background
2.1 Basic Hydrological Concepts
2.1.1 Catchment
A catchment is an extent or an area of land where surface water from rain, melting snow, or ice
converges to a single point at a lower elevation, usually the outlet of the basin, where the waters join
another waterbody, such as a river, lake, reservoir, estuary, wetland, sea, or ocean.
Other terms that are used to describe catchment are drainage basins, catchment area, catchment
basin, drainage area, river basin, and water basin
The drainage basin acts as a funnel by collecting the water within the area covered by the basin and
flowing it to a single point. Each drainage basin is separated topographically from adjacent basins by a
boundary, the drainage divide making up a succession of higher geographical features (such as a ridge,
hill or mountains) forming a barrier.
The catchment is a logical unit of focus for studying the movement of water within the hydrological
cycle, because the majority of water that discharges from the basin outlet originated as precipitation
falling into the basin. A portion of the water that enters the groundwater system beneath the drainage
basin may flow towards the outlet of another drainage basin because groundwater flow directions do not
always match those of their overlying drainage network. Measurement of flow discharge of a basin may
be made by using waters at a stream gauge located at the basin’s outlet.
In a catchment take place most of the hydrological processes that compose the hydrological cycle in
a region: rainfall, evaporation, run-off and other phases of the total cycle.
2.1.2 Catchment factors
Based on the response of catchment to rainfall flooding may take place.
The catchment is the most significant factor determining the amount or likelihood of flooding.
Catchment factors are: topography, shape, size, soil type, and land use (paved or roofed areas).
Catchment topography and shape are fundamental to determine the concentration time of catchment ,
while size, soil type, and development affect the amount of water to reach the river.
2.1.2.1 Topography
Generally, topography plays a big part in how fast runoff will reach a river. Rain that falls in steep
mountainous areas will reach the main river in the drainage basin faster than flat or lightly sloping areas
(e.g., > 1% gradient).
2.1.2.2 Shape
Shape will contribute to the velocity with which the runoff reaches a river. A long thin catchment
will drain longer than a circular catchment.
6
8. 2.1.2.3 Size
Size will help determine the amount of water reaching the river, as the larger the catchment the
greater the potential for flooding. It also determined on the basis of length and width of the drainage
basin.
2.1.2.4 Soil type
Soil type is fundamental to understand how much water reaches the river. Certain soil types such
as sandy soils are very free-draining, and rainfall on sandy soil is likely to be absorbed by the ground.
However, soils containing clay can be almost impervious and therefore rainfall on clay soils will run off
and contribute to flood volumes. After prolonged rainfall even free-draining soils can become saturated,
meaning that any further rainfall will reach the river rather than being absorbed by the ground. If the
surface is impervious the precipitation will create runoff surface which will lead to higher risk of flooding;
if the ground is pervious, the precipitation will infiltrate in the soil.
2.1.2.5 Land use
Land use can contribute to the volume of water reaching the river, in a similar way to clay soils. For
example, rainfall on roofs, pavements, and roads will be collected by rivers with almost no absorption
into the groundwater.
2.1.3 Precipitation
In meteorology, precipitation is any product of the condensation of atmospheric water vapour that
falls under gravity. Precipitation occurs when a portion of the atmosphere becomes saturated with water
vapour, so that the water condenses and ”precipitates”.
In hydrology the rainfalls are the major factor of design of hydraulic works. Therefore it is needed
to know characteristics, which are defined by the total rainfall depth accordingly to the spatial and
temporary distribution. Therefore, the following parameters are in use for the classification and study:
• Duration: Time in which the rainfall lasts, expressed generally in hours.
• Intensity: Rainfall for unit of time, expressed generally in millimeters per hours
• Return Period: The probability of rainfall event with takes place on a number of years.
2.1.4 Precipitation Classification
Mechanisms of producing precipitation include convective, stratiform, and orographic rainfall.
• Convective: Convection occurs when the Earth’s surface, mainly in the equatorial region, within a
conditionally unstable, or moist atmosphere, becomes heated more than its surroundings, leading to
significant evaporation. Convective rain, or showery precipitation, occurs from convective clouds,
e.g., cumulonimbus or cumulus congestus. It falls as showers with rapidly changing intensity.
Convective precipitation falls over a certain area for a relatively short time, as convective clouds
have limited horizontal extent.
• Stratiform: Is also caused by frontal systems surrounding extratropical cyclones or lows, which
form when warm and often tropical air meets cooler air. Stratiform precipitation falls out of
nimbostratus clouds. When masses of air with different density (moisture and temperature
characteristics) meet, warmer air overrides colder air. The warmer air is forced to rise and if
conditions are right becomes saturated, causing precipitation. In turn, precipitation can enhance
the temperature and moisture contrast along a frontal boundary. Fronts cause sudden changes in
general temperature, and in the humidity and pressure in the air. Warm fronts occur where the
warm air scours out a previously lodged cold air mass. The warm air ’overrides’ the cooler air and
moves upward. Warm fronts are followed by extended periods of light rain and drizzle, because,
after the warm air rises above the cooler air (which sinks to the ground), it gradually cools due to
the air’s expansion while being lifted, which forms clouds and leads to precipitation. Cold fronts
occur when a mass of cooler air dislodges a mass of warm air. This type of transition is sharper,
since cold air is more dense than warm air. The rain duration is less, and generally more intense,
7
9. than that which occurs ahead of warm fronts. A wide variety of weather can be found along an
occluded front, with thunderstorms possible, but usually their passage is associated with a drying
of the air mass.
• Orographic: Also called relief rainfall is caused when masses of air pushed by wind are forced up
the side of elevated land formations, such as large mountains. The lift of the air up the side of the
mountain results in adiabatic cooling, and ultimately condensation and precipitation.
Rainfall intensity is classified according to the rate of precipitation:
• Light rain: when the precipitation rate is less than 2,5 mm/hours.
• Moderate rain: when the precipitation rate is between 2,5 and 10 mm/hours.
• Heavy rain: when the precipitation rate is between 10 and 25 mm/hours.
• Violent rain: when the precipitation rate is greater than 100 mm/hours.
2.1.5 Obtaining average rainfalls on a catchment
Rain gauge data is used to measure total precipitation over a drainage basin, and there are different
ways to analyze the data. If there are gauges evenly distributed over an area of uniform precipitation,
using the arithmetic mean method will give good results
2.1.5.1 Arithmetic mean method
This technique calculates areal precipitation using the arithmetic mean of all the point or areal
measurements considered in the analysis
It is a method of rapid execution and that takes a very relative degree of precision, which depends:
on the number of stations of rain gauge, and the spatial distribution of the observed rainfall. It is the
only method that does not need of a previous knowledge of the location of every station. The looked
value is calculated by means of the equation 2.1:
Pm =
n
i=1
Pi
n
(2.1)
where Pm will be the average rainfall over the catchment, Pi is the gauged precipitation by the station
i and n is the amount of station of rain gauge.
2.1.5.2 Thiessen polygon method
In the Thiessen polygon method, the drainage basin is divided into polygons with the rain gauge in
the middle of each polygon assumed to be representative for the rainfall on the area of land included
in its polygon. These polygons are made by drawing lines between gauges, then making perpendicular
bisectors of those lines forming the polygons Fig. 2.1.
Then, to calculate average rainfall, equation 2.4 is used:
Pm =
n
i=1
PiAi
A
=
n
i=1
Pi
Ai
A
(2.2)
where Pm will be the average rainfall over the catchment, Pi gauged rainfall at pluviometric station
i, Ai polygon area corresponding to active rainfall station i, A total area of the catchment and n number
of active rain gauge station over the catchment.
8
10. Figure 2.1: Example of Thiessen’s polygons determination. Stations: P1...P7. Catchment:
Colorized area
2.1.5.3 Inverse distance weighting
Inverse Distance Weighting (IDW) is a type of deterministic method for multivariate interpolation
with a known scattered set of points. The assigned values to unknown points are calculated with a
weighted average of the values available at the known points.
A general form of finding an interpolated value u at a given point x based on samples ui = u(xi) for
i = 1,2,...,N
u(x) =
N
i=1
wi(x)ui
N
i=1
wi(x)
if d(x, xi) = 0 ∀i
ui if d(x, xi) = 0 for some i
(2.3)
where
wi(x) =
1
d(x, xi)p
(2.4)
is a simple IDW weighting function, x denotes an interpolated (arbitrary) point, xi is an interpolating
(known) point, d is a given distance (metric operator) from the known point xi to the unknown point x,
N is the total number of known points used in interpolation and p is a positive real number, called the
power parameter.
2.1.6 Hydrologic Abstraction
In the hydrological cycle, the rainfall is submitted to a series of losses or abstractions, before it flows
as runoff towards a river or creek. In hydrological design, the losses or abstractions are considered to be
the difference between the total recorded rainfall and the direct runoff at streamgage. The rainfall can
get lost for: interception in the soil, evaporation and evapotranspiration.
2.1.6.1 Interception
Interception refers to precipitation that does not reach the soil, but is instead intercepted by the
canopy of plants and the forest floor. Because of evaporation, interception of liquid water generally leads
to loss of that precipitation for the drainage basin, except for cases such as fog interception.
In the same way, at beginning of the rain, part of this fills the pores of the soil and getting lost then
for evaporation and infiltration.
9
11. These two interceptions (vegetation and soil) have values initially of the rain. If the rain is of small
magnitude, these losses have high influence producing run-off. If the rain is of greater duration and
intensity, the losses are relative and in general they are considered to be insignificant and are named
initial losses. The density of the vegetable coverage is an important factor in assessing interception
values.
2.1.6.2 Infiltration
Infiltration is the process by which water on the ground surface enters the soil. Infiltration rate in
soil science is a measure of the rate at which soil is able to absorb rainfall or irrigation. It is measured in
millimeters per hour. The rate decreases as the soil becomes saturated. If the precipitation rate exceeds
the infiltration rate, runoff will usually occur unless there is some physical barrier. It is related to the
saturated hydraulic conductivity of the near-surface soil. The rate of infiltration can be measured using
an infiltrometer.
During dry periods without crust of soil is free of moisture and for this increases the capacity of
infiltration until saturation condition.
Therefore, during a storm the potential rate of infiltration begins with high values and decreases as
the time goes taking place, a small rate, which is easily overcome by the intensity of the rain and begins
the ponding in the surface of the soil. This logically will be tied always to the relation between the
capacity of infiltration and the intensity of rainfall.
In general, except the particular case of a saturated soil, the velocity with which the water infiltrates,
it is a function of the time and of diverse factors as the initial content of soil moisture.
2.1.6.3 Evapotranspiration
Evaporation is a type of vaporization of a liquid that occurs from the surface of a liquid into a gaseous
phase that is not saturated with the evaporating substance
Evapotranspiration is the sum of evaporation and plant transpiration from the Earth’s land and ocean
surface to the atmosphere. Evaporation accounts for the movement of water to the air from sources such
as the soil, canopy interception, and waterbodies.
Evapotranspiration is a significant water loss from drainage basins. Types of vegetation and land
use significantly affect evapotranspiration, and therefore the amount of water into a drainage basin.
Because water transpired through leaves comes from the roots, plants with deep reaching roots can more
constantly transpire water.
2.1.6.4 Depression Storage
Depression storage accounts for the water that becomes ponded in land surface irregularities.
Depression storage depends on the land use of the watershed and typically values ranging from 0.5
to 8 mm are considered during a single rain event. It is inversely proportional to the watershed’s slope
2.1.7 Infiltration and abstraction calculation methods
Infiltration is a component of the general mass balance hydrologic budget. There are several ways to
estimate the volume and/or the rate of infiltration of water into a soil. Three estimation methods are
constant loss, exponential and SCS method:
• Initial abstraction (Ia) and constant loss
• Exponential
• SCS Method
• Runoff Coefficient
2.1.7.1 Initial abstraction (Ia) and constant loss
The initial value Ia (mm) and the constant rate in mm /hour, allow to consider easily the phenomenon
of interception and infiltration respectively. Both values can be optimized if it exists hidrographs observed.
At the beginning of the rain losses are quantified Ia. From there, if the rain continues and the rate is
less than the rate of infiltration there is run-on until the rainfall intensity is greater than infiltration.
10
12. 2.1.7.2 Exponential
It is an empirical method developed by the U.S. Army Corp. of Engineers that relates the rate of losses
to the intensity of the rain and to the accumulated losses. The accumulated losses are representative of
the capacity of storage of moisture in the soil, by means of a not linear function:
IT = AIE
(2.5)
where IT is the total lost mm/hour, A is the coefficient that represents the combined effect of Ia and
of the infiltration, both depending on the accumulated losses and E is an exponent which is a function
of the intensity of the rainfall (l/t)
2.1.7.3 Runoff curve number
The runoff curve number (also called a curve number or simply CN) is an empirical parameter
used in hydrology for predicting direct runoff or infiltration from rainfall excess. The curve number
method was developed by the USDA Natural Resources Conservation Service, which was formerly called
the Soil Conservation Service or SCS — the number is still popularly known as a ”SCS runoff curve
number” in the literature. The runoff curve number was developed from an empirical analysis of runoff
from small catchments and hillslope plots monitored by the USDA. It is widely used and is an efficient
method for determining the approximate amount of direct runoff from a rainfall event in a particular
area.
The runoff curve number is based on the area’s hydrologic soil group, land use, treatment and
hydrologic condition. References, such as from USDA indicate the runoff curve numbers for characteristic
land cover descriptions and a hydrologic soil group.
The runoff equation is:
Q =
0 for P ≤ Ia
(P −Ia)2
P −Ia+S for P > Ia
(2.6)
where Q is runoff, P is rainfall, S is the potential maximum soil moisture retention after runoff begins
and Ia is the initial abstraction, or the amount of water before runoff, such as infiltration, or rainfall
interception by vegetation; historically, it has generally been assumed that Ia = 0.2S, although a more
recent research has found Ia = 0.05S more appropriate and accurate relationship.
The runoff curve number, CN, is then related
S =
1000
CN
− 10 (2.7)
CN has a range from 30 to 100; lower numbers indicate low runoff potential while larger numbers are
for increasing runoff potential (impervious basins). The lower the curve number, the more permeable the
soil is. As can be seen in the curve number equation, runoff cannot start until the initial abstraction has
been met.
The NRCS curve number is related to soil type, soil infiltration capability, land use, and the depth of
the seasonal high water table. To account for different soils’ ability to infiltrate, NRCS has divided soils
into four hydrologic soil groups (HSGs). They are defined as follows.
• HSG Group A (low runoff potential): Soils with high infiltration rates. These consist chiefly
of deep, well-drained sands and gravels. These soils have a high rate of water transmission (final
infiltration rate greater than 7.5 mm/h).
• HSG Group B Soils with moderate infiltration rates. These consist chiefly of soils that
are moderately deep to deep, moderately well drained to well drained with moderately fine to
moderately course textures. These soils have a moderate rate of water transmission (final infiltration
rate of 4.0 to 7.5 mm/h).
• HSG Group C Soils with slow infiltration rates. These consist chiefly of soils with a layer that
impedes downward movement of water or soils with moderately fine to fine textures. These soils
have a slow rate of water transmission (final infiltration rate 1.3 to 4.0 mm/h).
11
13. • HSG Group D (high runoff potential): Soils with very slow infiltration rates. These consist chiefly
of clay soils with a high swelling potential, soils with a permanent high water table, soils with a
claypan or clay layer at or near the surface, and shallow soils over nearly impervious materials.
These soils have a very slow rate of water transmission (final infiltration rate less than 0.0 to 1.3
mm/h).
Selection of a hydrologic soil group should be done based on measured infiltration rates, soil survey
(such as the NRCS Web Soil Survey), or judgement from a qualified soil science or geotechnical
professional. The table below presents curve numbers for antecedent soil moisture condition II (average
moisture condition). To alter the curve number based on moisture condition or other parameters, see the
CN adjustment section. To define this number are available predefined tables that can be consulted.
2.1.7.4 Runoff Coefficient
From average rainfall and discharge data, runoff coefficient can be calculated:
C =
VT
RA
(2.8)
VT = Qi × 3600 (2.9)
RA = Ri × A × 1000 (2.10)
where C is the runoff coefficient, VT is the direct volumen, RA is the rainfall volume and multiplicative
factors are due to complying with units of measure.
2.1.8 Surface runoff
Surface runoff (also known as overland flow) is the water flow that occurs when excess stormwater,
meltwater, or other sources flows over the earth’s surface. This might occur because soil is saturated,
because rain arrives more quickly than soil can absorb it, or because impervious areas (roofs and
pavement) send their runoff to surrounding soil that cannot be infiltrated. Surface runoff is a major
component of the water cycle. It is the primary agent in soil erosion by water as well.
Runoff that occurs on the ground surface before reaching a channel is also called a nonpoint source.
If a nonpoint source contains man-made contaminants, or natural forms of pollution (such as rotting
leaves) the runoff is called nonpoint source pollution. A land area which produces runoff that drains to
a common point is called a drainage basin. In addition to causing water erosion and pollution, surface
runoff in urban areas is a primary cause of urban flooding which can result in property damage, damp
and mold in basements, and street flooding.
Surface runoff can be generated either by rainfall,snowfall or by the melting of snow, or glaciers.
In areas where there is no snow, runoff will come from rainfall. However, not all rainfall will produce
runoff because storage from soils can absorb light showers.
• Infiltration excess overland flow: This occurs when the rate of rainfall on a surface exceeds the
rate at which water can infiltrate the ground, and any depression storage has already been filled.
This is called flooding excess overland flow, Hortonian overland flow (after Robert E. Horton),
or unsaturated overland flow. This more commonly occurs in arid and semi-arid regions, where
rainfall intensities are high and the soil infiltration capacity is reduced because of surface sealing,
or in paved areas.
• Saturation excess overland flow: When the soil is saturated and the depression storage filled,
and rain continues to fall, the rainfall will immediately produce surface runoff. The level of
antecedent soil moisture is one factor affecting the time until soil becomes saturated. This runoff
is called saturation excess overland flow or saturated overland flow.
• Antecedent soil moisture: Soil retains a degree of moisture after a rainfall. This residual water
moisture affects the soil’s infiltration capacity. During the next rainfall event, the infiltration
capacity will cause the soil to be saturated at a different rate. The higher the level of antecedent
soil moisture, the more quickly the soil becomes saturated. Once the soil is saturated, runoff occurs.
12
14. Figure 2.2: Hydrograph example and its components
• Subsurface return flow: After water infiltrates the soil on hillslope, the water may flow laterally
through the soil, and exfiltrate (flow out of the soil) closer to a channel. This is called subsurface
return flow or throughflow.
As it flows, the amount of runoff may be reduced in a number of possible ways: a small portion of
it may evapotranspire; water may become temporarily stored in microtopographic depressions; and a
portion of it may infiltrate. Any remaining surface water eventually flows into a receiving water body,
lake, estuary or ocean.
2.1.9 Hydrograph
A hydrograph is a graph showing the rate of flow (discharge) versus time at a specific gauged river
site (see Fig. 2.2), or other channel or conduit carrying flow. The rate of flow is typically expressed in
cubic meters or cubic feet per second (cms or cfs).
It can also refer to a graph showing the volume of water reaching a particular outfall, or location in a
sewerage network, graphs are commonly used in the design of sewerage, more specifically, the design of
surface water sewerage systems and combined sewers.
The discharge is measured at a specific point in a river and is typically time variant. Main componentes
of the hydrograhs are:
• Rising limb: The rising limb of hydrograph, also known as concentration curve, reflects a prolonged
increase in discharge from a catchment area, typically in response to a rainfall event.
• Recession (or falling) limb: The recession limb extends from the peak flow rate onward. The
end of stormflow and the return to groundwater-derived flow (base flow) is often taken as the point
of inflection of the recession limb. The recession limb represents the withdrawal of water from the
storage built up in the basin during the earlier phases of the hydrograph.
• Peak discharge: the highest value of the hydrograph.
• Lag time: the time interval from the center of mass of rainfall excess to the peak of the resulting
hydrograph
• Time to peak: time interval from the start of the resulting hydrograph until peak discharge is
reached.
• Discharge: the rate of flow (volume per unit time) passing a specific location in a river or other
channel
13
15. 2.1.9.1 Baseflow separation
A stream hydrograph is commonly conceptualized to include a baseflow component and a direct
runoff component. The former represents the relatively steady contribution to stream discharge from
groundwater return flow, while the latter represents the additional streamflow contributed by surface
runoff.
The separation of baseflow from direct runoff in a hydrograph is often of interest to hydrologists,
planners, and engineers, as it is of support in determining the influence of different hydrologic processes
on discharge from the subject catchment. Because the timing, magnitude, and duration of groundwater
return flow differs so greatly from that of direct runoff, separating and understanding the influence of
these distinct processes is the key to analyzing and simulating the likely hydrologic effects of various land
use, water use, weather, and climate conditions and changes.
2.1.9.2 Unit hydrograph
A unit hydrograph (UH) is the hypothetical unit response of a watershed (in terms of runoff volume
and timing) to a unit input of rainfall. It can be defined as the direct runoff hydrograph (DRH) resulting
from one unit (e.g., one cm or one inch) of effective rainfall occurring uniformly over that watershed at
a uniform rate over a unit period of time. As UH is applicable only to the direct runoff component of a
hydrograph (i.e., surface runoff), a separate determination of the baseflow component is required.
UH is specific of a particular watershed, and specific for a particular duration corresponding to the
one of the effective rainfall. That is, the UH is specified as being the 1-hour, 6-hour, or 24-hour UH,
or any other length of time up to the time of concentration of direct runoff at the watershed outlet.
Thus, for a given watershed, there can be many unit hydrographs, each one corresponding to a different
duration of effective rainfall.
The UH technique provides a practical and relatively easy-to-apply tool for quantifying the effect of a
unit of rainfall on the corresponding runoff from a particular drainage basin. UH theory assumes that a
watershed’s runoff response is linear and time-invariant, and that the effective rainfall occurs uniformly
over the watershed. Actually, none of these assumptions are strictly true. Nevertheless, application of
UH methods typically yields a reasonable approximation of the flood response of natural watersheds.
The linear assumptions underlying UH theory allows for the variation in storm intensity over time (i.e.,
the storm hyetograph) to be simulated by applying the principles of superposition and proportionality.
This allows for a relatively straightforward calculation of the hydrograph response to any arbitrary rain
event.
An instantaneous unit hydrograph is a further refinement of the concept; for an IUH, the input
rainfall is assumed to all take place at a discrete point in time (obviously, this isn’t the case for actual
rainstorms). This assumption can greatly simplify the analysis involved in constructing a unit hydrograph,
and it is necessary for the creation of a geomorphologic instantaneous unit hydrograph.
2.1.10 Rating Curve
Rating curve is a graph of discharge versus stage for a given point on a stream, usually at gaging
stations, where the stream discharge is measured across the stream channel with a flow meter. Numerous
measurements of stream discharge are made over a range of stream stages. The rating curve is usually
plotted as discharge on x-axis versus stage (surface elevation) on y-axis. Fig. 2.3
The development of a rating curve involves two steps. In the first step the relationship between stage
and discharge is established by measuring the stage and corresponding discharge in the river. And in the
second part, stage of river is measured and discharge is calculated by using the relationship established
in the first part. Stage is measured by reading a gage installed in the river. If the stage-discharge
relationship doesn’t change with time, it is called permanent control. If the relationship does change, it
is called shifting control. Shifting control is usually due to erosion or deposition of sediment at the stage
measurement site. Bedrock-bottomed parts of rivers or concrete/metal weirs or structures are often,
though not always, permanent controls.
14
16. Figure 2.3: Example of rating curve
2.1.11 Modeling and simulation
Modeling and simulation (M&S) is getting information about how processes will behave without
testing them in real context. More generally, M&S is uses models, including emulators, prototypes and
stimulators, either statically or over time, to develop data as a basis for making managerial or technical
decisions.
2.1.11.1 Hydrological Model
A hydrological model is a mathematical model describing the rainfall–runoff relations of a rainfall
catchment area, drainage basin or watershed. More precisely, it produces the surface runoff hydrograph
as a response to a rainfall hydrograph as input. In other words, the model calculates the conversion of
rainfall into runoff.
A well known runoff model is the linear reservoir, but in practice it has limited applicability.
The runoff model with a non-linear reservoir is more universally applicable, but still it holds only for
catchments whose surface area is limited by the condition that the rainfall can be considered more or less
uniformly distributed over the area. The maximum size of the watershed then depends on the rainfall
characteristics of the region. When the study area is too large, it can be divided into sub-catchments
and the various runoff hydrographs may be combined using flood routing techniques.
Hydrological models, can be divided in different categories according to his characteristics, principally
in:
• Continuos or Discretes (Based on Events): A system is continuos when the phenomena are
continuos in the time, and discrete when the change of condition is given in discretes intervals,
particularly for models whose aim forecasting of sudden floods, based on events of rainfalls and
runoff.
• Lumped or Distribuited: A model is lumped when it does not consider the spatial variability.
In general the lumped models use only the time as independent variable. A model is distributed
when the variables and parameters of the model depend on the space and the time. For this is that
the area of study can be subdivided in subareas.
• Conceptual or Empirical: A model is conceptual when the functions used in his production have
in consideration the physical processes. The empirical models are those in which there adjust the
values calculated to the observed information, across functions that do not have any relation with
the physical wrapped processes.
15
17. Figure 2.4: Location of catchment of San Antonio River
2.2 Study Area
2.2.1 Location
The area of study is the catchment of San Antonio River, one of the principal tributaries to the
Reservoir San Roque, in the Province of Cordoba, Argentina.
The catchment of San Antonio River, is located completely inside of the department of Punilla, in
the Cordoba Province 2.4.
2.2.2 Catchment Description
It is one of the principal tributaries of the reservoir San Roque, which collects the waters from the
river Cosqu´ın, as well as Las Mojarras and Los Chorrillos streams, and the contributions not channelled
of the Perilago. These tributaries constitute the top catchtment of the Suqu´ıa river, which nowadays it
begins in the reservoir, a few kilometres of its round eastward it crosses the city of Cordoba, the provincial
capital, and flows into Mar Chiquita great lagoon.
The catchment of the San Antonio river has a surface of 496 km2
. It is located on the mountain range
of Sierras Chicas in province of Cordoba by an average height of 1431 mslm.
It can divided into three zones according to the height over sea level:
• Top: Over 2000 mslm, area 108 km2
• Middle: From 1200 mslm to 2000 mslm, area 205 km2
• Bottom: Up to 1200 mslm, area 873 km2
San Antonio River is the product of the confluence of four important watercourses: the Cajon River,
the Malambo River, the Icho Cruz River and the San Antonio stream.
2.2.2.1 Geology
Catchment is constituted principally by granite and metamorphic rocks of the type Gneis that cover
83 % of the area of the basin. Whereas the rest, principally the valleys, are formed by sediments of the
type loessico (see Fig.2.4).
2.2.2.2 Climate
The region is characterized by a climate of height, with low temperatures, irregular rainfalls mainly
in summer and occasional snowfalls in the winter period. The annual average temperature is 16o
C,
decreasing to 10o
C from the 2.000 mslm., with average minims of 9o
C and 5o
C respectively.
16
18. The average annual rainfalls are near to 1.000 mm in the riverheads, decreasing eastward up to 750
mm. 80 % of the rainfalls take place between October and May, with records that overcome 50 mm
monthly. There exists a great range of the monthly averages, which go from 10 mm in dry period, to
240 mm in the rainy one. On the other hand, a decrease marked with the rainfalls with the decrease of
height takes place. On the stations located to major height the records overcome 230 mm per month,
whereas at minor height the rainfalls reach 130 mm.
2.2.2.3 Telemetric Network INA-CIRSA
The Centro de Investigaciones de la Regi´on Semi´arida (CIRSA) of the Instituto Nacional del Agua
(INA) of Argentina, involved in issues related to sudden flood, installed in 1986 a telemetric network of
compilation and transmission of information in the catchment of the San Antonio River. The network has
13 automatic remote gauge stations. All the stations transmit information in real time to the head office
of processing located in the headquarters of the CIRSA in the city of Villa Carlos Paz. This telemetric
network has the aptitude to transmit information, either corresponding to rainfalls (Fig. 2.6) or river
water levels (Fig. 2.7), every fixed interval of time determined by controllers (see Fig.2.5). The above
mentioned information is stored and is what they have been used.
Figure 2.5: Station distribution in the catchment of San Antonio River
17
19. Figure 2.6: Distribution of rain gauge stations in San Antonio catchment .
Figure 2.7: Distribution of level measures stations in San Antonio catchment.
2.3 Reference Models
Models that will be mentioned below were developed by the Research Institute for
Geo-Hydrological Protection (Istituto di Ricerca per la Protezione Idrogeologica (IRPI)).
IRPI is an institute of the Department of Earth System Sciences and Technology for the Environment
(DTA), of the Italian National Research Council (Consiglio Nazionale delle Ricerche, CNR). The mission
of the institute is to design and execute scientific research and technological development in the fields of
natural hazards, environmental protection, and the sustainable use of geo-resources. IRPI carries out its
18
20. mission by operating at different geographical and temporal scales, and in different climatic, physiographic
and geological zones.
The hydrology group consists of engineers and technicians that investigate meteo - hydrological
processes at all scales, from the local to the global scale. The group studies the processes of the formation
of the floods, develops and tests tools and models for hydro - meteorological monitoring, the estimation
of the soil water content, and for flood hazard assessment and risk evaluation. Several models developed
by the hydrology group are operational in civil protection centres.
2.3.1 MISDc
The continuous semi-distributed rainfall-runoff (RR) model, named MISDc (‘Modello Idrologico
Semi-Distribuito in continuo’) (Brocca et. al 2011 [2]) is a parsimonious and reliable continuous RR
model operating at an hourly (or smaller) time scale. Specifically, in order to determine the catchment
hydrological response, the important role of the antecedent wetness conditions (AWC) is emphasized.
The application of MISDc both for design flood estimation and for flood forecasting proved its reliability
and also its computational efficiency.
In particular, the MISDc has been implemented in the framework of Civil Protection activities for
the Upper Tiber River basin.
The code is written in the MATLAB programming language and is fully commented; an executable
version of the model is also available
2.3.1.1 Description
The methodology used for MISDc is based on the following steps:
• Soil Water Balance (SWB) model to simulate the soil moisture temporal pattern,
• A semi-distributed event-based RR model (MISD) (Melone et al. 2001 [5], Moramarco et al. 2005
[8]) implemented and tested over a large number of flood events in several catchments. In the
model, the parameter linked to AWC was set equal to the value optimising the direct runoff volume
reproduction.
• Coupling of the SWB and the MISD models by exploiting the relationship between the parameter
representing the AWC in the MISD model and soil moisture measurements.
Through this relationship the soil moisture simulated by the SWB model and the AWC to be used in
the MISD model were linked, thus providing the MISDc structure. (Fig. 2.8)
2.3.1.2 Soil water balance model
The surface soil layer is assumed as a spatially lumped system for which the following water content
balance equation holds:
dW(t)/dt = f(t) − e(t) − g(t) for W(t) ≤ Wmax
W(t) = Wmax for otherwise
(2.11)
where t is time, W(t) is the amount of water in the investigated soil layer, f(t) is the fraction of the
precipitation infiltrating into the soil, e(t) is the evapotranspiration rate, g(t) is the drainage rate due to
the interflow and/or the deep percolation, and Wmax is the maximum water capacity of the soil layer.
The ratio W(t)/Wmax represents the degree of saturation. to see with more details how to calculate
values for f(t), e(t), g(t) refers to Brocca et al. (2011)
What is important to stress is that SWB model is linked to evet-based RR model by a linear
relationship between the maximum potential retention of soil S and the degree of saturation θe = W/Wmax
before the event (Antecedent Wetness condition, AWC), as:
The SM simulated by the SWB is used to calculate the parameter S of the SCS method by means of
an experimentally derived relationship between S and SM
S = (1 − θe)a (2.12)
where θe is the modelled relative soil moisture at the beginning of the event and a is a parameter to
be estimated. Once the S parameter is estimated, MISD is used for simulating the flood hydrograph.
19
21. Figure 2.8: Schematic diagram of MISDc structure
Therefore, to adress this validation, it is necessary for each rainfall event, to have the value θe provided
and corresponding S value. The value S can be obtained from different ways.
The SCS-CN method for abstraction is employed to estimate the AWC at the catchment scale for the
three investigated catchments. The choice of SCS-CN method, besides the simplicity (only one parameter
has to be estimated), is also due to its wide use since the 1980s in a number of continuous simulation
models. Therefore, the improvement for the AWC estimation in the SCS-CN method is a very important
issue for its significance in the runoff determination. Looking at SCS-CN formulation, the partitioning
of rainfall into runoff for a storm as a whole relies on the following empirical equation:
Q =
(P − Fa)2
P − Fa + S
for P ≥ Fa (2.13)
where Fa is the initial abstraction, S is the potential maximum retention, Q is the direct runoff depth,
and P is the rainfall depth. The quantity Fa is assumed as a fraction of S by
Fa = λS (2.14)
with λ = 0.2 as the SCS standard value. The potential maximum retention, S, is estimated in the
classical procedure considering the dimensionless curve number (CN) assessed as a function of land use,
hydrological soil group and the AWC. For a catchment, assuming that the soil /land use characteristics
are constant in time, the different values of S from storm to storm are only linked to variations in the
AWC. Therefore, to assess the AWC, the potential maximum retention was determined using observed
rainfall and direct runoff depth and it is denoted henceforth as ‘observed’ potential maximum retention,
Sobs. For this aim, Eq. 2.13 has been written in terms of S = Sobs, yielding
Sobs =
1
2λ2
(2λP − λQ + Q − λ2Q2 − 2λQ2 + 4λQ + Q2) (2.15)
The value of λ parameter was set 0,20 as in the classical SCS-CN method because it does not influence
significantly the results. It has to be pointed out as Sobs does not represent the real average soil moisture
condition of the catchment but, mainly for hydrological purposes, can be considered a good indicator of
the catchment AWC, thus allowing to compute directly the storm runoff depth through 2.13 (Brocca et.
al 2009 [4]).
2.3.1.3 Semi-distributed event-based rainfall-runoff model
On the basis of the drainage network and on the geomorphological, soil/land use characteristics, a
given catchment is divided into Nb elements, each one representing either a subcatchment with outlet
along the main channel or an area draining directly into the main channel. Each element is assumed
homogeneous and hence constitutes a lumped system.
20
22. The version of MISD used for this study employs the Soil Conservation Service-Curve Number method
for abstraction (SCS-CN), the geomorphological Instantaneous Unit Hydrograph (IUH) for routing rainfall
excess of subcatchments and of areas draining directly into the main channel, respectively. Finally, the
routing along the main channel is estimated through a diffusive linear approach. (See Brocca et al. (2011)
for more details [2]).
2.3.2 SCRRM
2.3.2.1 Flood Modelling
SCRRM (Massari et. al 2013)[3], might be viewed as an evolution of the MISDc. Because the temporal
evolution of the soil wetness conditions of the catchment is not modelled from rainfall and temperature
data like in MISDc but it is integrated directly into the model by exploiting SM observations (i.e. SWB
is replaced in SCRRM by SM observations).
Based on that, SCRRM uses SM indicators provided by external sources to infer the value of S
parameter for runoff determination. Like in MISDc, the model leverages the observed linear behaviour
between the wetness state of the soil and the parameter S of the SCS method.
SCRRM uses the SM and the event rainfall data as sole inputs to simulate hourly flood hydrographs
as shown in Fig. 2.9. The calibration of the model involves the following three parameters: the coefficient
of initial abstractions λ, the parameter a of the S − θe relationship 2.17, and the parameter η of the lag
time–area relationship.
Figure 2.9: Structure of the Simplified Continuos RR Model (SCRRM)
2.3.2.2 SM product description
RR models applied for operational flood forecasting can be subdivided in two main categories:
continuous and event-based. Continuous RR models simulate the temporal evolution of the soil moisture
(SM) conditions of the catchment, thus being able to model the complex interaction between rainfall
and SM which is necessary to properly predict flood hydrographs. However, the different processes
(infiltration, percolation, evapotranspiration, interception) involved in the simulation of the SM temporal
evolution may require a large number of parameters to be identified.
The major limitations of event-based models lie in the definition of the initial SM conditions that
could be very different from one storm event to another. SM information can be obtained from in situ
and satellite sensors or from land surface models. Several satellite SM products are globally and freely
available from active and passive microwave sensors, e.g. the Advanced SCATterometer (ASCAT)
The accuracy and maturity of these satellite products have contributed to the implementation of a
fully operational nearreal- time (NRT) SM processing chain for ASCAT. All these SM data sets, which are
globally available, might be potentially used for the initialization of event-based RR models in different
catchments and regions worldwide, even for poorly gauged areas.
Simplified Continuous RR Model (SCRRM) exploites SM provided by satellite. This new approach
offers the advantages of continuous models, with the difference that the temporal evolution of SM over
a long-term period is assessed by using SM directly from external sources, thus avoiding of simulating
processes such as evapotranspiration, evaporation and groundwater flow.
Remotely sensed products provide knowledge of soil moisture for a very thin surface layer (ca. 0–5
cm), however this is not sufficient for hydrological applications concerning RR transformation. Indeed,
21
23. root-zone SM data are the main control parameters on the catchment response to a given storm event. To
obtain the root-zone SM product (SWI; soil water index) from the satellite-based surface observations,
the semi-empirical approach was adopted The recursive formulation of the method relies on
SWI(tn) = SWI(tn−1) + Kn[ms(tn) − SWI(tn−1)] (2.16)
whre ms(tn) is the surface SM observed by the satellite sensor, SWI(tn) is the soil wetness index
representing the profileaveraged saturation degree and time tn is the acquisition time of ms(tn). The
gain Kn at time tn is given by (in a recursive form):
Kn =
Kn−1
Kn−1 + e−(
tn−tn−1
T )
(2.17)
where T is the characteristic time length and represents the timescale of SM variation to obtain the
SWI. For the initialization of this filter, K1 and SWI1 were set to 1 and ms(t1), respectively. The
approach is also known as exponential filter.
2.3.2.3 S − θ relationship
Since the SM in SCRRM is provided by an external indicator, the S − θe relationship becomes a
model relation embedded in the model structure and it is used to estimate the value of S for the analysed
events.
22
24. Chapter 3
Methods
The first steps of this work began with the reading of the paper Distributed rainfall-runoff modelling
for flood frequency estimation and flood forecasting (Brocca et. al 2011) [2], the paper explains the
development of the hydrological model MISDc, section 2.3.1. This led to the learn of basic knowledge of
hydrological concepts for the correct comprehension of the document. These concepts are summarized in
section 2.1 of the Theorical Background of this report. This material was chosen as starting point due to
the fact that it is the implementation of a hydrological model which is nowadays operative at IRPI and
is considered to be suitable for the catchment of San Antonio River.
Immediately afterwards, it was considered the paper: Using globally available soil moisture indicators
for flood modelling (Masssari et. al 2013 [3]). This work explains the development of another hydrological
model called SCRRM, (Section 2.3.2). This model, on the one hand is an evolution of MISDc, due to
the fact that it incorporates directly information of SM satellite products in face of soil water balance
model and on the other hand it is a simplification of the previous one, since it is used as a lumped model
instead of a semidistributed model.
For this, SCRRM is a good option as starting point for the first elaboration with information of the
catchment of San Antonio River. SCRRM is implemented in the language of programming MATLAB,
therefore it is necessary to have available this program for the computation.
3.1 SCRRM input data
Considering the aim of this first stage is to perform a simulation of the model, there has be known
inputs data and their format.
Input data necessary for the correct simulation of this model are:
3.1.1 Separate Events
In this case, previously selected events can be provided as input to the model:
• One text file for each event to be simulated, containing two columns separated by blank space where
each row has to correspond to the same date-time:
– 1st Column: Hourly average rainfall of the catchment.
– 2nd Column: Hourly outlet discharge in cubic meters per second.
• One text file containing soil moisture observed for each event. The most recent value of measurement
of soil moisture, previous to the event, must be selected.
3.1.2 Continuous data series
In this case continuos series data can be provided to the model and the program will select
automatically rainfall events to work.
• Rainfall, Discharge and SM data: Text file containing four columns separated by blank space
where each row has to correspond to the same date-time:
– 1st Column: Timestamp, Date in Matlab format.
23
25. – 2nd Column: Hourly average rainfall of the catchment.
– 3rd Column: Hourly outlet discharge in cubic meters per second.
– 4th Column: Hourly SM product.
3.2 Available data survey
The next step was necessary to made a survey of the available data of rainfall and discharge of the
basin as well as knowing important characteristics of the basin, (gauge stations, spatial distribution of
these stations, catchment size, etc.)
3.2.1 Separate Events
INA has provided data of 9 storms events ocurred for the period 2006-2007. For each storm, which
from now we will all events, it is available:
• Rainfall Data: An Excel table where in every page one finds the information about the millimeters
accumulated of 15 previous minutes recorded at each raingauge. (Fig. 3.1)
• Discharge Data: An Excel table containing data of volume discharge recorded at the outlet of
the basin. (Fig. 3.2)
Figure 3.1: Example of rainfall data file provided by INA corresponding to storm event
ocurred 09-Nov-2007
24
26. Figure 3.2: Example of Discharge data file provided by INA corresponding to storm event
ocurred 09-Nov-2007
3.2.2 Continuous data series
INA also has provided continuous data series of rainfall and level measure station, from 1994 to 2015
in different formats depending on the period of time.
3.2.2.1 1994-2009 Pre-processed data in simple columns
• Rainfall Data: A text file for each station (about 10 stations Fig. 2.6), each one of the text file
from rain gauge station contains:
– Header: Information about the gauge station (location, elevation, etc) and about the data
series (format, interpolation method, etc)
– 1st Column: Date in format yyyy-MM-dd
– 2nd Column: Timestamp in format HH:mm
– 3rd Column: Accumulated millimeters recorded in the last hour.
– 4th Column: Percentage of available values.
25
27. Figure 3.3: Example of an extract of text file with pre-processed rainfall data in simple
columns.
• Discharge Data: Text file with values recorded in the station 604, located in the outlet of the
catchment Fig. 2.7 , described as follows:
– Header: Brief information about the gauge station.
– 1st Column: Date in format yyyy-dd-MM
– 2nd Column: Timestamp in format HH:mm
– 3rd Column: Volume in cubic meters per second with step time of 1 hour.
– 4th Column: Percentage of available values.
Figure 3.4: Example of an extract of text file with pre-processed discharge data in simple
columns.
3.2.2.2 2006-2014 Raw data in simple columns
• Rainfall Data: A text file for each station (about 10 stations Fig. 2.6), each one of the text file
from rain gauge station contains:
– Header: Information about the gauge station (location, elevation, etc) and about the data
series (format, interpolation method, etc)
26
28. – 1st Column: Date in format yyyy-MM-dd
– 2nd Column: Timestamp in format HH:mm:ss
– 3rd Column: Accumulated millimeters recorded.
Each line added to this file is determined by:
– If the station has been registered 1 mm of precipitacion, then one record is added to this file
withe the value of accumulated millimeters incremented by one.
– If after of 12 hours and 12 minutes no one millimeter has been registered, then the record is
added with the same previous value of registered millimeters.
Figure 3.5: Example of an extract of text file in simple columns containing raw data of
precipitations at station 706.
• Level measures data: Text file for each of the 4 level measure stations Fig. 2.7, with level
measure registered with step time of 18 minutes and 13 seconds is described as follows:
– Header: Brief information about the gauge station.
– 1st Column: Date in format yyyy-MM-dd
– 2nd Column: Timestamp in format HH:mm:ss
– 3rd Column: Level measure of the river in meters.
Figure 3.6: Example of an extract of text file in simple columns containing raw data of
level measures at station 706.
3.2.2.3 2008-2015 Tabular format for pre-processed rainfall and raw level measures data
• Rainfall Data: A text file containing information about all rain gauge Fig.2.6:
– Header: Information about the provider of the product and date of created.
27
29. – 1st Column: Information is organized in descending order, from newest to oldest with one
hour step time. This columns indicates the date and time corresponding to the values of the
others columns. Date in format yyyy-MM-dd indicates a shift day, after this day change, the
column only will show time in HHmm format.
– Following columns: Each of the remaining columns has the value of precipitacion ocurred
in the last hour. Each line of the column match with the time of the first column. First of
these lines has the ID of the rain gauge for each column.
Figure 3.7: Example of an extract of text file in tabular format containing precipitations
registered on 2012.
• Raw level measures data: A text file containing information about 4 level measure stations
Fig.2.7:
– Header: Information about the provider of the product and date of created.
– 1st Column: Information is organized in descending order, from newest to oldest with one
hour step time. This columns indicates the date and time corresponding to the values of the
others columns. Date in format yyyy-MM-dd indicates a shift day, after this day change, the
column only will show time in HHmm format.
– Following Columns: Each pair of the remaining columns has the value of level measures
and the time in wich the level measures has reached its maximum. Each line of the column
match with the time of the first column. First of these lines has the ID of the level measure
station for each column. In the case of some value has the word “ ” have to be considered
as a NaN value.
28
30. Figure 3.8: Example of an extract of text file in tabular format containing level measures
in 4 stations registered on 2015.
3.2.2.4 SM Data
IRPI provided SM Products in a text file in ascending order where each line represents a SM product,
and columns are described:
• 1st Column: Timestamp, Date in Matlab format
• 2nd Column: Timestamp, Date and time in format dd-MM-yy HH:mm
• 3rd Column: Percentage of available values.
The frequency and date-time of this data depends on the satellite’s passes over the catchment, that’s
why in some cases it can be found zero, one or two values a day.
3.3 Soil Moisture Data (ASCAT)
To obtain SM products (ASCAT), location (latitude and longitude) of INA stations was used.
Locations are identified in the map, and the position of the central point of soil moisture was determined,
then SM product that had minimal distance with the center of the catchment was chosen.
After this, SM product for the period containing storm events (year 2007) was exported.
File contains four columns, but only the first and the second column will be used, with the information:
• Date-time of measured data, Matlab format.
• Soil Moisture value.
3.4 Tools
3.4.1 Average Rainfall
3.4.1.1 Brief description
This tool was developed by Christian Massari at IRPI. The function of this tool is to get the catchment
average rainfall from separate rainfall of each rain gauge station.
29
31. 3.4.1.2 Inputs
• Rainfall Stations Data: This is a plain text file of several columns and lines. Each column is
separated of another one by a tabular space, and represents each rain gauge station. Each line is
the accumlated rainfall of the las step time.
• Stations Location: This is a plain text file containing in each line latitude and longitude of each
rain gauge station.
• Catchment Centroid: This is a plain text file in with the only information is a single line with
the latitude and longitude of the catchment centroid.
• Catchment Shape File: Shape file of catchment containing location of the stations.
3.4.1.3 Output
• Raifall Stations Data: A plain text file with only one column containing for each step time, the
average rainfall of the catchment in float number format.
3.4.1.4 User Interface and Use
The program was developed in Matlab. To do an execution it should be opened from this framework.
The starting point of this software is the file main.c. To get catchment average rainfall it can use stations
location (latitude and longitude) and centroid, or catchment shape file. To select one way or another
of working the user can comment or uncomment the desired block. As we can se in figure 3.9 and then
complete with the correct text file names as required.
Figure 3.9: Beginning of main module program to get catchment average rainfall. Here,
the user can uncomment or comment to use shape file, also can modify name of input files.
30
32. 3.4.1.5 Diagram Flow
Figure 3.10
1. The program starts reading the inputs of rainfall of each station, stations location and catchment
centroid or shape file.
2. Then, NaN values of rainfall data are completed considering closer values.
3. The weight matrix is created applying IDW method. Section 2.1.5.3
4. Finally, without NaN values of rainfall in data and weight matrix, average rainfall of the catchment
is obtained in a vector after of multiplying this two matrix.
5. The resultant vector is written in a plain text file where each line is average rainfall calculated for
one step time.
3.4.2 SCRRM
3.4.2.1 Brief description
This tool was developed by Christian Massari at IRPI. The objective of this tools is to implement
the SCRRM model. This model uses some important functions as extracting events of rainfall and MISC
model developed by Lucca Brocca [2].
31
33. 3.4.2.2 Inputs
Inputs of the program that implements SCRRM model have to be located in local path
• Average rainfall, Discharge and SM Data: This is a plain text file of several columns and
lines. Each column is separated of another one by a tabular space. Description of each column is
detailed in section 3.1
• Instaneous Unit Hidrograph: This is a plain text file with that contain adimensional IUH
information.
• Fixed Parameters: This is a plain text file with fixed parameters: catchment area (kmsq),
computation time step (h), input data time step (h)
• Calibration Parameters: This is a plain text file with calibration paremeters. The program has
a flag variable to indicate if the calibration is done in the moment of running the execution, in the
case of this flag is set with 0 (false) the calibration is not performed and the model uses parameters
provided by this file.
3.4.2.3 Outputs
• Average rainfall, Discharge and SM Data: This is a plain text file of several columns and
lines. Each column is separated of another one by a tabular space. Description of the content of
this file is detailed in section 3.1
• Extracted Events: This is a set of files with “evs” extension. Each file represents one extracted
event containing two columns, the first one contains hourly rainfall and the second one contains the
volume of discharge corresponding with the hour of the first column.
• Plots of Extracted Events: This is a set of files with “emf” extension. Each file represents a
image plot that match with the “evs” file with the same name. The plot has two curves in different
scales, the blue one represents rainfall and the green the discharge. An example of this plot is shown
in the figure 3.11
Figure 3.11: Plot of rainfall and discharge for the event ocurred in 14-Nov-2008 in San
Antonio catchment.
• List of extracted events files: This is a plain text file called “ele.txt” tha contains only one
column listing names of the files “evs” extracted.
• Calibration Parameters: In the case the calibration of model parameters is done while the
execution of the program, this values are written in this file called Xopt.txt by default. Eacher
parameter is written in a different line of the file.
• Simulated Events: This is a set of text files called PQQSimi.txt (where “i” is the of number
simulated event) by default, each file is one event simulated by the model.
32
34. • Plots of simulated events: This is a set of plots generated as output of matlab program. Each
plot have thres curves that represent observed event (rainfall and discharge) and the simulated
event. An example of four simulated events for Tiber River basin are shown in the figure 3.12
Figure 3.12: Plots of a set of simulated events as function of time. Bars represents rainfall,
continuos line represents observed discharge and dotted line, simulated discharge
3.4.2.4 User Interface and Use
The program was developed in Matlab. To do an execution it should be opened in this framework.
The starting point of this software is the file run SWI.m. The function extr evs PQ() perform the
extraction of events. Once the events were extracted, the observed and simulated events take place
using those extracted events. In the case of extracted events have to be ignored, because inconsistent
data or whatelse, the generated files for this event (evs file and emf file) should to be removed and
the corresponding line in nameID.txt file should to be erased. Then, the call to this function can be
commented to work only with the selected events that have relevance when the model is executed again.
To calibrate the model during the execution of the program, the variable cal should be set to 1 in the
other case (use the file Xopt.txt in disk) this variable has to be the value 0.
Plots of rainfall and discharge observed; and simulated discharge curve can be view once the model
ends the execution so, this way it can be analized in detail.
33
36. Figure 3.14: Diagram Flow of the program that implements SCRRM model (Part 2)
Below the list of actions that take place
1. The program starts reading the inputs of average rainfall, discharge and SM Data of the catchment
as it is described in section 3.1
2. Using data loaded in the previous step, if the extr evs PQ() function is not commented, events of
relevant rainfaill are extracted in different text files. Rainfall and discharge of the period of the
event is written in a text file (evs files). Then, plots of those events are generated and written as
images files (emf files). A file to identify the events is created also.
3. The list evs files is loaded by the program.
4. Using the list of evs files generated, ele.txt file is written on disk for later use.
5. File that list events files is generated (ele.txt).
6. For each event in the ele.txt file (extracted events)...
7. IUH and fixedpair described in 3.4.2.2 are loaded.
8. Event Based Lumped Rainfall Runoff Model is executed for each event, it is a lumped version of
the model described in 2.3.1.3.
9. Due to soil moisture data is not present for every hour, this values are interpolated.
10. Soil moisture interpolated values are normalized (0 < SM < 1)
11. If flag of calibration is set to 1...
12. Xopt.txt file containing calibration parameters is loaded
13. Computing of calibration parameters of model is executed
14. Xopt.txt file is updated with the new parameters of calibration.
15. Xopt.txt updated file is loaded to use in the rest of the program.
16. Soil Water Index (SWI) is computed recursively as was explained using the equation 2.17
17. S parameter of SCS is computed using SWI value as is mentioned in section 2.12
18. For each event in the ele.txt file (extracted events)...
19. Curve Number is computed using S value calculated previously.
20. Simulated events are computed using the lumped version of the model described in 2.3.1.3
21. Mean errors are computed.
22. Plots of observed and simulated events are shown for analysis
35
37. 3.4.3 MISD
3.4.3.1 Brief description
This tool was developed at IRPI. The objective of this tools is to implement the MISD model to
estimate discharge at the outlet of a catchment after of an rainfall event. See Melone et al. 2001 [5]
3.4.3.2 Inputs
Input of the program that implements MISD model have to be located in local path,
• Average rainfall and Catchment area: Average rainfall with one hour step time.
• Dimensionless unit hydrograph: Text file called ”in4” containing fix information that had been
proved in several catchments.
3.4.3.3 Outputs
It is a text file where estimated discharge in cubic meters can be found in one hour step time. It can
be found after the text Tempo e portata oraria calcolata”.
3.4.3.4 User Interface and Use
The model was provided as an executable program called ”Argentina.exe”’. After running this file
the program ask for the name of the average rainfall input file in a interface as a command line. Fig. ??
Figure 3.15: Screenshot of the command line interface belonging to the MISD model
implementation program
After writting the file name of the rainfall input file, it should be written the name of the output
file chosen. Then, the method for the initial abstraction coefficients should be selected from 4 differents
options.
After this, the execution of the model is done and if everything is ok, the program finish closing the
window and generating the output with a file name previously defined.
3.5 SCRRM Model execution
3.5.1 Preprocessing Data
Rainfall, discharge and soil moisture continuos data series, after preprocessing described below, was
placed in a Excel file to ease the way for organizing the information. This way, it is easy check the
36
38. correspondence among the dates of values coming from different files, further it is simpler export data
needed to a text file.
3.5.1.1 Rainfall data preprocessing
As was described in section 3.1, average rainfall data is required to execute SCRRM model. Hourly
rainfall data of each station are available. That’s why, it was necessary to process this data to get hourly
average rainfall.
This calculation could be done using tool described in section 3.4.1. Two input files were needed to
run this tool:
Station hourly rainfall All hourly rainfall data was organized in a text file by matching the same
line of the text file with data of all rain gauge station accordingly the same date. Each column of the file
correspond to one rain gauge station. The columns are separated by a tab space.
Catchment centroid To calculate the catchment we need to calculate the polygon centroid because
the location of rain gauge station determines a polygon.
The centroid of a non-self-intersecting closed polygon defined by n vertices (x0, y0), (x1, y1), ...,
(xn1, yn1) is the point (Cx, Cy), where:
Cx =
1
6A
n−1
i=0
(xi + xi+1)(xiyi+1 − xi+1yi) (3.1)
Cy =
1
6A
n−1
i=0
(yi + yi+1)(xiyi+1 − xi+1yi) (3.2)
and where A is the polygon’s signed area,
A =
1
2
n−1
i=0
(xiyi+1 − xi+1yi) (3.3)
In these formulas, the vertices are assumed to be numbered in order of their occurrence along the
polygon’s perimeter. Furthermore, the vertex ( xn, yn ) is assumed to be the same as ( x0, y0 ), meaning
i + 1 on the last case must loop around to i = 0. Note that if the points are numbered in clockwise order
the area A, computed as above, will have a negative sign; but the centroid coordinates will be correct
even in this case.
In this case catchtment centroid was calculated using QGIS Software by providing latitude and
longitude information of each rain gauge station location Table 3.1
Table 3.1: Rain gauge stations location
The catchment centroid value obtained was: −31.490382825 −64.64707432
3.5.1.2 Discharge data preprocessing
As was described in section 3.1, hourly volume discharge data in cubic meters are required to execute
SCRRM model. River level measure with a period of 18 minutes and 13 seconds are available and, in
some cases, with loss of information for longer periods of time. That’s why, it was necessary to process
this data to get hourly volume discharge average rainfall as follows:
37
39. • Fill missing data: Step time of provided data is 18 minutes, 13 seconds. For those periods without
value, it was necessary fill with NaN value.
• Hourly data interpolation: A linear interpolation was necessary for having availability of values
with an hourly step time. For each hour, previous and next value was considered to do the
interpolation. In the case of next or previuos values were NaN, the next one was considered and
so, until a maximum of four NaN values. In the case of reaching four previous or subsequent NaN
values, the new value for hourly step time was set as NaN.
• H-Q Conversion: H-Q Equation developed by Facundo Alonso [6] was applied to stages values,
this way obtain volumen discharge values in cubic meters.
Q =
19, 86 ∗ (H2
) + 7, 8944 ∗ H + 0, 053 for H < 0, 85
77, 9 ∗ (H2
) − 54, 3206 ∗ H + 11, 0019 for H > 0, 85
(3.4)
To do this format conversion, a Python program was developed. See: A.1
3.5.1.3 Soil Moisture Data preprocessing
SCRRM model needs a hourly SM value, in case of this value is not present have to be complete with
NaN value. Furthermore, SM values available does not match exactly of an hour, that’s why SM value
present will correspond with the next hour close to the timestampo of the SM value.
A program in Python was developed to process SM data and complete with NaN for not existing
values. See A.2
3.5.2 Execution with continuos data series
Volume discharge data was provided by INA in cubic meter. Also, raw data level measure of the
catchment outlet to be processed and obtain this way volume discharge, this is, because in different
period of time, the H-Q Transformation Equation could be different, depending of the location of the
sensor. The discharge data preprocessing, explained in section 3.5.1.2, and the program in Python
(Appendix A.1) was developed to do this processing for the set of data.
Once this information is ready, it is placed in Excel file together with the rest of information and then
exported as a text file.
At the beginning of the execution of the program that contains the implementation of the model,
events of rainfall are identified and then, they are extracted as files in disk.
If some of the extracted events present NaN values in discharge data, they should be removed from
the extracted ones in the first stage, the files with extension emf and evs should be deleted and the line
corresponding with those events should be deleted in evID.txt file. The line in the source code of the
program that perform the events extraction should be commented, so in the next execution the model
will work only with existing files of events in path work.
3.6 MISD Model execution
3.6.1 Events ocurred in 2007
3.6.1.1 Rainfall data preprocessing
For preprocessing data, Excel was used accordingly with INA format and for the first step it ensues
a speditive analysis.
One hour step time data: As available data refer rainfall every 15 minutes, records were aggregated
at hourly time step. The selected records were exported to another page of Excel file.
Coefficient of Thiessen polygons for each station: Considering that rainfall data is available for
each station the average areal rainfall of catchment is computed by the method of Thiessen’s polygons
(Section 2.1.5.2).
The information of Thiessen’s polygons of the basin of the San Antonio River was obtained and shown
in Fig. 3.21
For each event a table was drawn up with the available information Table 3.2, where for each station,
the hourly rainfall is multiplied by the coefficient of the corresponding area and, then, the average areal
rainfall is obtained as:
38
40. Figure 3.16: Map of catchment of San Antonio River divided by using Thiessen polygons
method.
¯Rt =
n
i=1
AiRi
n
i=1
Ai
(3.5)
where t is the hour, n is the amount of station of rain gauge, Ai is the polygon area for station i and
Ri is the hourly rainfall recorded at the station i.
Table 3.2: Thiessen polygons coefficients for each station calculated from area.
Thus the average rainfall is obtained as final result for every hour contained inside the duration of
the event.
3.6.1.2 Discharge data preprocessing
Likewise of rainfall data, the discharge data are every 15 minutes, but these values are instantaneous,
by what, unlike the rainfall, only the corresponding information was chosen at each hour and added to
the table of rainfalls. (Table. 3.3)
39
41. Table 3.3: Average rainfall table, calculated from the Thiessen polygons coefficients and
Discharge Data
3.6.1.3 Analysis rainfall-discharge data
Once obtained these two sets of data with time step of 1 hour and aligned according to the occurrence,
graphs were done:
• X axis: time step (hourly)
• Y axis: rainfall and discharge values.
Hereby it is possible to observe if the information makes some of sense. (Fig. 3.17)
40
42. Figure 3.17: Storm events provided by INA, selected to run the SCRRM model.
For storms events ocurred in 2007 in San Antonio River catchment the runonff coefficients were
calculated using Equation 2.8. Overmore, the total rainfall of the catchment, total volume, the lag time
and the soil moisture value corresponding to the most recent value previous to the storm event are shown.
This values are in Table 3.4
Table 3.4: Runoff coefficients, total rainfaill, total volume, lag time and soil moisture values
for storm events ocurred in 2007 in San Antonio River catchment.
After this, the content of the events was exported in separate text files as it was described in section
3.1.1
41
43. 3.6.1.4 Execution
In the case of separate events, MISD model was used to do the simulation. This is the model included
in the SCRRM model implementation to perform the simulation of discarge in the outlet of the catchment
as was described in Section 2.3.2
Unlike the case in wich continuous data series are used and selection of events is done by the program
automatically, in this case, events should be preselected. That is why input data for the model must be
prepared as was described in Section 3.6.1.
Once the preparation of the input data is done, the execution of the MISD model was done to perform
the simulation of catchment discharge.
This results are shown in Figure 3.18
Figure 3.18: Plots of observed and simulated discharge of events storm occurred in 2007
in San Antonio catchment. Simulations were performed by MISD model.
After running the model, results were plotted to compare observed and predicted data, Figure 3.18.
For most of cases a significant shift in the horizontal axis can be observed, indicating a difference of about
15 hours between observed and predicted data. This is an relevant difference having in mind that the
purpose of the simulation is to prevent in advance the hour in which the peak of flushflood reach the
outlet of the catchment so, these results are not considered completely satisfactory.
To check consistency data two events with very different lag time were compared. Event occurred in
21-11-2007, which the simulation seems to be good Figure 3.18, and another with the shift mentioned
above, (event occurred 10-11-2007). As we can see these events have a very differente lag time (table 3.4)
42
44. 3 hours against 18 hours. Distribution of rainfall in the catchment was reviewed also:
Figure 3.19: Distribution of rainfall in the catchment for storm event occurred on
21-11-2007. For each station, the amount of registered milimeters is indicated in parenthesis.
Lag time of this event: 3 hours
Figure 3.20: Distribution of rainfall in the catchment for storm event occurred on
10-11-2007. For each station, the amount of registered milimeters is indicated in parenthesis.
Lag time of this event: 18 hours
After reviewing distibution of rainfall in the catchment for mentioned events, it would be expected
that the intensity of rain is very high for the event with the lag time smallest (event 21-11-2007, 3 hours
of lag time) or closer to the outlet than the other event. But, the event of 10-11-2007 has an uniform
distribution of rainfall in all the catchment and with high registered values also.
These results suggest reviewing the input data to verify reliability of this information.
43
45. 3.6.2 Events ocurred in 2011-2013
After reviewing results obtained from events of rainfall and discharge values for the period 2007-2008,
INA suggested to work with the data in 2011-2013 period, which should provide more reliable results.
The tabular format provided (Section 3.2.2.3) contains the required data for this period and in the
correct data type because the rainfall is accumulated per hour.
In this period discharge is available as raw data, i.e. the river stages instead of the discharge
volume. To perform the conversion from stages to volume of discharge is necessary to have available
the transformation equation H-Q but, in this case, this conversion equation is not precisely defined, due
there are several equations defined (See: Alonso 2011 [6], Guillen et. al 2015[7]). Therefore, for the second
execution model it was decided to select one of the level measure stations of river located upstream in
the basin, to compare the results generated by the model.
3.6.2.1 Rainfall data preprocessing
Because the tabular format provided is not a convenient format to export to Excel format, it was
necessary to develop a program to perform this format conversion. The program was developed in Python
and it can be found in Appendix A.3. After processing the data in a format compatible with Excel, this
information is organized into tables for having a correspondence between rainfall data of different rain
gauge stations.
Considering that the area of the catchment in this instance is different from the total area of the basin
and, the station 600 is not present in the subcatchment, the calculation of the average rainfall is affected
by a change in the coefficients of the Thiessen polygons for each station.
As result, the Polygons Thiessen for the subcatchment is the next one:
Figure 3.21: Thiessen map for subcatchment with station 700 as outlet
A slightly different map is considered in case no data are available from the station 300
Then, new values of thiessen coefficient computed are:
44
46. Figure 3.22: Thiessen coefficient for subcatchment with station 700 as outlet
3.6.2.2 Stages data preprocessing
The same way as for rainfall data, a program was developed in Python for data preprocessing stages
in tabular format. Appendix A.4. This information is also added to the Excel tables containing data of
rainfall to have a correspondence between rainfall data and stages data from different stations and the
same date and time.
3.6.2.3 Execution and analysis
After preprocessing of rainfall and stages data and organizing in tables, rainfall events were identified
in the 2011-2013 period. Rainfall was observed and analised in plots of different stages stations. Station
704 was chosen as the outlet of the subcatchment to work with the model and compare estimated discharge
obtained.
The level measures station 704 is upstream of station 600 and includes almost 60 % of the total San
Antonio catchment. This station was chosen as catchement outlet, area calculation of the subasin was
done. The subcatchment area is 311 km2
and comprises stations 100, 200, 300, 400, 500, 700, 900, 1000,
1200 and 1800. Then, events to execute the model were chosen:
45
47. Figure 3.23: Plots of rainfall and observed river stages of events storm occurred in 2011
and 2013 in San Antonio catchment. Yellow lines represent stage river for station 704, which
we will have in mind for study
Values of rainfall of these events were exported in a text file for each event, the same for stage values
for each event.
After running the MISD model as was described in Section 3.4.3 for each of the above events, results
were compared with the stages data from the station 704 in the outlet of the catchment.
Considering that it is not available the transformation equation H-Q for this level measures station,
comparison of discharge curves was made to carrying values to units without dimension, i.e. values were
normalized by performing the following calculation:
x =
x − min(x)
max(x) − min(x)
(3.6)
where x is an original value, x is the normalized value.
Results are shown as plotted lines:
46
48. Figure 3.24: Results obtained after running MISD model for six events ocurred between
2011-2013
As can be observed, estimations of discharge obtained in this case is much approaching to the stages
observed, although these results are much more satisfactory compared to those obtained on executions
on 2007 period (Section 3.6.1), which have an important shift lag time, we must not forget that the
comparison is being made between dimensionless values because the absence of transformation HQ
equation (Section 2.1.10).
This suggests estimation of a curve of transformation HQ by performing a regression curve. In the
next plot, stages river values were plotted on the x axis and estimated discharge were plotted in the y
axis:
47
49. Figure 3.25: Rating curve estimated plotting stages in X axis and estimated discharge in
Y axis with a confidence bound of 95 %
The equation for this estimated rating curve is a polinomial of 4th degree with:
R2
= 0.8984 (3.7)
It is very strange that for some values of stages, different values of discharge was obtained.
3.7 Conclusions
From the point of view of training, this period of time working at IRPI was an excellent opportunity
to see up close and understand operation of hydrological models and processes involved in the execution
of them.
It would have been interesting to get satisfactory results after the execution SCRRM model,
considering that is a model which is currently operating successfully in IRPI and that takes as input
soil moisture product from ASCAT satellite, which is an interesting point related with the topic of the
master. A correct estimation of the peak of the flood would be a successful result, but this could not be
achieved due to some inconsistencies in the input data, lack of preprocessing or correct preparation of
the information input and inaccuracy in the H-Q transformation equations.
For executions performed by MISD model in events ocurred in the period 2011-2013 better
approximations of estimates compared with river levels were obtained, in this case using as outlet a
station upstream, that station only had stages river information, and for which no transformation curve
H-Q is available at this moment.
For future studies of this basin in which you need to work with rainfall and stages or discharge data
it is recommended:
• To check the quality of rainfall data in all raingauges of basins and filter rainfall data so remove
some inconsistencies. Fig. 3.26
48
50. Figure 3.26: Example of rainfall data errors (over 1000 mm value in one hour in station
100)
• To check H-Q transformation equation for the station 600, outlet of San Antonio catchment. This
is because there are different equationes defined for diferent periods, but it is a little confused.
• To propose calculating H-Q transformation equations for other stages stations within the basin, this
could enable the implementation of a distributed or semidistribuited model, dividing the catchment
into sub-catchments.
• To check and filter data of discharge, it could present inconsistencies. Example: Station 704 values
are usually between 0.3-2.0 meters, but in 2015 the values are about 5.0 meters for a continuos
period of 3 days.
• To have an uniform format for input data, at this moment there are different formats and it was
necessary to develop program to process each of these formats.
• To write documentation containing information of reliability of data, definition of H-Q
transformation equations specifically defined for each period, periods where it is recommended
not to use data to avoid mistakes, etc. To have all this information easily accessible, probably via
web.
49
52. #Author: Cristian Guerrero Cordova
#Date: April-2015
#IRPI-Perugia-Italy
import datetime
import calendar
import math
def interpolator():
LevelFile=open("C:Python342007-2009_Raw.txt")
datesArray=[]
reversedDatesArray=[]
LevelArray = []
reverseLevelArray = []
valuesQ=[]
tempRecord=[]
tempDateTime = datetime.datetime.now()
tempDateSTR=""
#Step time input values: 18 minutes, 13 Seconds
stepTime=datetime.datetime.strptime("00:18:13","%H:%M:%S")
#Read the file parsing the lines separated by tab, ignores lines that not respect the format.
for linea in LevelFile:
tempRecord=linea.split('t')
try:
tempDateTime=datetime.datetime.strptime(tempRecord[0], "%m/%d/%y %H:%M:%S")
datesArray.append(tempDateTime)
LevelArray.append(tempRecord[1])
except:
continue
#Array reverse for data in descending order
reversedDatesArray=datesArray[::-1]
reverseLevelArray=LevelArray[::-1]
#In case data come in ascending order
"""
reversedDatesArray=datesArray
reverseLevelArray=LevelArray
"""
#First of all, fill empty data with NaN,
#Esto seria, si entre un registro y otro la fecha es mayor al paso de tiempo, completar con
NaN.
#O lo que es lo mismo o parecido, si no existe el siguiente paso de tiempo, crearlo con NaN.
#Si en cambio el siguiente registro tiene un intervalo menor, dejarlo como esta y tomarlo como
nuevo registro.
newDateFilled=[]
newLevelFilled=[]
firstDate=reversedDatesArray[0]
# Append first date to the new array
newDateFilled.append(firstDate)
# The same to level water value
newLevelFilled.append(reverseLevelArray[0])
# Add 18 minutes and 13 seconds
firstPlusStepTime=firstDate+ datetime.timedelta(seconds=1093);
# Getting the second records to compare with the step time
secondDate=reversedDatesArray[1]
i=0
lengthArray=len(reversedDatesArray)
while(i<lengthArray-2):
firstPlusStepTime=firstDate+ datetime.timedelta(seconds=1093);
if firstPlusStepTime<secondDate:
# The new FirsDate will be the previous one plus one step time.
firstDate=firstPlusStepTime
#Append this value to the date values
newDateFilled.append(firstDate)
#Append the new record with value NaN
newLevelFilled.append("NaNn")
# The new SecondDate will be the same before.
secondDate=secondDate
else:
# The new firsdate will be the previous SecondDate, because all it is OK.
firstDate=secondDate
#Append this value to de dates array
newDateFilled.append(firstDate)
#Append the corresponding level value to the levels array
newLevelFilled.append(reverseLevelArray[i])
#Adding one the index
i=i+1
#The new SecondDate will be the record next to firstDate actual
53. secondDate=reversedDatesArray[i+1]
#beggining date
startDate=datetime.datetime.strptime("01/01/07 00:00:00", "%d/%m/%y %H:%M:%S")
#final date
endDate=datetime.datetime.strptime("31/12/09 23:00:00", "%d/%m/%y %H:%M:%S")
iterDate=startDate
#Index will be iterating in the array dates to found dates after and before of another one.
index=0
firstDate=newDateFilled[index]
#To save interpolated values.
valoresInterpolados=[]
#To save dates with hourly step time
datesByHour=[]
indexInterpInf=0
indexInterpSup=0
while (iterDate<endDate and index < len(newDateFilled)-3):
if (iterDate<firstDate):
iterDate = iterDate + datetime.timedelta(hours=1)
elif (iterDate==firstDate):
dDate=firstDate
y=float(newLevelFilled[index])
Q=converseHQ(y)
valuesQ.append(Q)
valoresInterpolados.append(y)
datesByHour.append(dDate)
iterDate = iterDate + datetime.timedelta(hours=1)
elif (firstDate<iterDate):
while(firstDate<=iterDate):
index=index+1
firstDate=newDateFilled[index]
#Setting index of interpolation, searching for not NaN values, for the previous
values. Until 4 NaN values acepted.
if(math.isnan(float(newLevelFilled[index-1]))):
if(math.isnan(float(newLevelFilled[index-2]))):
if(math.isnan(float(newLevelFilled[index-3]))):
if(math.isnan(float(newLevelFilled[index-4]))):
y='NaN'
Q=converseHQ(y)
valuesQ.append(Q)
valoresInterpolados.append(y)
datesByHour.append(iterDate)
continue
else:
indexInterpInf=index-4
else:
indexInterpInf=index-3
else:
indexInterpInf=index-2
else:
indexInterpInf=index-1
#Setting index of interpolation, searching for not NaN values, for the next values.
Until 4 NaN values acepted.
if(math.isnan(float(newLevelFilled[index]))):
if(math.isnan(float(newLevelFilled[index-1]))):
if(math.isnan(float(newLevelFilled[index-2]))):
if(math.isnan(float(newLevelFilled[index-3]))):
y='NaN'
Q=converseHQ(y)
valuesQ.append(Q)
valoresInterpolados.append(y)
datesByHour.append(iterDate)
continue
else:
indexInterpSup=index-3
else:
indexInterpSup=index-2
else:
indexInterpSup=index-1
else:
indexInterpSup=index
#Interpolating using values determined by the index
d1Date=newDateFilled[indexInterpInf]
d2Date=firstDate
dDate=iterDate
y1=float(newLevelFilled[indexInterpInf])
y2=float(newLevelFilled[indexInterpSup])
d=(dDate - d1Date).total_seconds()
54. d1=0
d2=(d2Date-newDateFilled[indexInterpInf]).total_seconds()
y= ((d/d2)*(y2-y1)) + y1
#Applying H-Q Equation, can be used another HQ Function here.
Q=converseHQ(y)
#Adding Q value to output array.
valuesQ.append(Q)
#Adding y level measure value to output array.
valoresInterpolados.append(y)
#Adding the corresponding Date to array
datesByHour.append(dDate)
#Writting the outputs
output=open("outputHQInterp.txt", "w")
for i in range(len(datesByHour)):
tempDateSTR=datesByHour[i].strftime('%d/%m/%y %H:%M:%S')
output.write(tempDateSTR+",t"+str(valoresInterpolados[i])+",ttttt"+str(valuesQ[i])
+"n")
output.close()
#H-Q Equation, divided in two parts
def converseHQ(valorH):
Q=0.0
if (valorH == 'NaN'):
Q='NaN'
elif (valorH < 0.85):
Q = 19.86*(valorH**2.0)+7.8944*valorH + 0.053
elif (valorH >= 0.85):
Q = 77.9 *(valorH**2.0)-54.3206*valorH + 11.0019
return Q
#H-Q Equation
def converseHQAfter2010(valorH):
Q=0.0
if (valorH == 'NaN'):
Q='NaN'
else:
Q=47.56*(valorH**(5/3))
return Q
interpolator()
55. A.2 Preprocessing of SM Data
#Author: Cristian Guerrero Cordova
#Date: April-2015
#IRPI-Perugia-Italy
import datetime
def fillSMData():
#Read SM file provided by IRPI
SMFile=open("C:AguaIRPISCRRMtsh_ASCAT_1699820.csv")
datesArray=[]
SMArray = []
tempRecord=[]
tempDateTime = datetime.datetime.now()
#Parse SM file and load relevant data in arrays.
for linea in SMFile:
tempRecord=linea.split('t')
tempDateTime=datetime.datetime.strptime(tempRecord[2], "%d/%m/%y %H:%M")
datesArray.append(tempDateTime)
SMArray.append(tempRecord[3])
#Setting start and end date.
startDate=datetime.datetime.strptime("02/01/07 13:00", "%d/%m/%y %H:%M")
endDate=datetime.datetime.strptime("27/02/13 02:00", "%d/%m/%y %H:%M")
#Date to iterate hourly
iterDate=startDate
#Output File
outputSMComplete=open("outputSMComplete.txt", "w")
j=1
dateAvailSM=datesArray[0]
SMFileComplete=open("SMDataComplete.txt", "w")
tempDateSTR=datetime.datetime.now()
#Loop to complete with NaN value where needed
while (iterDate<endDate):
#If iterator value is lower than next SM Value
if(iterDate<dateAvailSM):
#Create DateTime Value
tempDateSTR=iterDate.strftime('%d/%m/%y %H:%M')
#And write in the file
outputSMComplete.write(tempDateSTR+"t"+"NaN"+"n")
else:
#Create DateTime Value
tempDateSTR=dateAvailSM.strftime('%d/%m/%y %H:%M')
#And write that value of SM
valueSM=SMArray[j]
outputSMComplete.write(tempDateSTR+"t"+valueSM+"n")
dateAvailSM=datesArray[j]
j=j+1
#Changing iterator.
iterDate = iterDate + datetime.timedelta(hours=1)
outputSMComplete.close()
fillSMData()
54
