069MSW405_Devendra Tamrakar_Presentation

EVALUATION OF TRMM PRECIPITATION ALL
OVER NEPAL AND IT’S APPLICATION IN
HYDROLOGIC MODELLING IN NARAYANI BASIN
Presentation:
A thesis in partial fulfillment of the requirement for
the degree of Master of Science in Water Resources
Engineering
Institute of Engineering, Pulchowk Campus
Presenter:
Devendra Tamrakar (069/MSW/405)
Supervisor:
Prof. Dr. Narendra Man Shakya

CONTENTS
 Introduction
 Objectives of study
 Literature Review of radar based precipitation
 Methodology to extract TRMM data’s
 Statistical Comparison(TRMM & Gauge
data’s)
 PINEHBV Model
 Input data preparation in model
 Model simulation and results
 Conclusions and Recommendations

INTRODUCTION
 Precipitation as one of the most important hydrological
model inputs for planning, development and operation of
water related projects.
 The widely used method of rainfall estimation in the
catchment is the method of point measurements from a
network of ground based hydro meteorological stations. So
for accurate estimation of precipitation over the area,
precipitation stations should be placed closed together.
This demands high number of precipitation stations which
is very costly for the developing country like Nepal.
 An alternative to ground based precipitation estimation is
the use of radar based precipitation data.
 In this study, TRMM data’s are taken under consideration.

OBJECTIVES
 Download TRMM raw data’s.
 Python programming techniques to process the TRMM
data’s and statistical comparison with gauge data’s in
Nepal.
 TRMM rainfall data as input in PINEHBV model to
calculate runoff in Narayani basin’s outlet.

LITERATURE REVIEW
 The first meteorological satellite was launched in 1960.
The techniques include use of Visual (VIS) and Infrared
(IR) data to infer precipitation intensity based on the
characteristics of cloud tops and its temperature. These
techniques have relatively low degree of accuracy, sensing
of cloud physics properties and the precipitation reaching
the surface.
 During 80s, passive microwave sensors (PMW) were
developed. These provide more accurate estimates of
rainfall than that of VIS and IR but there was a limitation
that this satellite could provide limited (only 1 or 2)
samples in a day.
 Recently developed remote sensored precipitation
estimation techniques use both active and passive sensors.

LITERATURE REVIEW contd…
Details of Satellite based data’s

TRMM 3B42 V6
 TRMM is a joint mission between the National
Aeronautics and Space Administration (NASA) of the
United States and the National Space Development
Agency (NASDA) of Japan.
 The objectives of TRMM are to measure rainfall and
energy (i.e., latent heat of condensation) exchange of
tropical and subtropical regions of the world. The
satellite was launched on November 27, 1997 from the
Tanegashima Space Center in Tanegashima, Japan.

Components of TRMM

Data characteristics of TRMM 3B42 V6
Temporal Resolution: 1998 to present
Geographic Coverage:
Latitude:50° S to 50°N
Longitude:180°W to 180° E
Temporal Resolution: 3 – Hourly
Spatial Resolution 0.25° x 0.25°
Grid Size 400 x 1440 pixels
Average File Size Compressed: ~285 KB;
Original: ~4.5 MB
Projection : Geographic WGS 1984
File Type: HDF
Precipitation measurement: mm/hr.
Missing value -999.9

Methodologies to extract TRMM data
 Click on this link: http://mirador.gsfc.nasa.gov to
download raw HDF format TRMM rainfall data.
 The precipitation data need data processing before it
could be used to compare with the ground based gauge
precipitation data and to make use in runoff modeling.
 Data is processed by point to pixel comparison.
 The use of python script with GDAL (Geospatial Data
Abstraction Library) library makes conversion possible
between different formats of raster and vector geospatial
data.

Diagrammatic presentation of use of
scripts

Statistical Comparison
(TRMM & Gauge data’s)
 Different visual and statistical methods are used to
compare satellite precipitation with gauge precipitation;
the comparison has been done from 2001 to 2008.
 Gauging stations with no recorded precipitations on
either satellite or gauge or lots of missing data on gauge
are omitted for analysis. The limit for number of
missing data for the gauge station is calculated as,
 If the value of ‘Analysis days’ falls behind 0.8, then the
data of the gauge station is not used for comparison.

Statistical Comparison (TRMM & Gauge
data’s) contd…
 Scatter plot
Monthly comparison plot
Daily comparison plot

(TRMM & Gauge data’s) contd…
 Nash-Sutcliffe Coefficient of Efficiency (R²)
Daily comparison Monthly comparison
R2 No. of Station
1 – 0.5 5
0.5 – 0 85
-1 – 0 143
-3 – 1 10
-30 – -3 3
-40 – -30 2
Total Stations 248
R2 No. of Station
1 – 0.5 171
0.5 – 0 52
-1 – 0 8
-3 – 1 3
-20 – 3 8
Total Stations 242

 Root Mean Squared Difference (RMSD)
RMSD No. of Station
0-1 0
1-5 7
5-10 20
10-20 99
20-40 117
40-60 5
Total stations 248
RMSD No. of Station
0-10 0
10-50 2
50-100 91
100-200 120
200-400 25
400-600 4
Total Stations 242

 Mean Relative Absolute Difference (MRAD)
MRAD No. of Station
0-1 30
1-2 83
2-5 102
5-15 33
Total stations 248
MRAD value No. of Station
0-1 186
1-2 32
2-5 19
5-15 5
Total Stations 242

Spatial Variability within a Single Pixel
Station
ID
Station name District Type of Station
Easting
(deg)
Northing
(deg)
Elevation(m)
1022 Godavari Lalitpur Climatology 27.58 85.4 1400
1029 Khumaltar Lalitpur
Agrometeorolog
y 27.67 85.33 1350
1030
Kathmandu
Airport Kathmandu Aeronautical 27.7 85.37 1337
1039 Panipokhari Kathmandu Climatology 27.73 85.33 1335
1052 Bhaktapur Bhaktapur Precipitation 27.67 85.42 1330
1059
Changu
Narayan Bhaktapur Precipitation 27.7 85.42 1543
1060 Chapa Gaun Lalitpur Precipitation 27.6 85.33 1448
1073 Khokana Lalitpur Climatology 27.63 85.28 1212
1075 Lele Lalitpur Precipitation 27.58 85.28 1590
1080 Tikathali Lalitpur Precipitation 27.65 85.35 1341
1082 Nangkhel Bhaktapur Precipitation 27.65 85.47 1428

TRMM map showing a 0.25° x 0.25° pixel with gauge stations

Comparison of daily rainfall within a same pixel

Comparison of monthly rainfall within a same pixel

PINEHBV Model
The model is
 Linear Model.
 Lumped Model.
 Conceptual Model.
 Deterministic Model.

Data preparation for Model
The data required are
 Precipitation
 Temperature
 Potential Evapotranspiration
 Area Elevation Curve/Hysometric curve

Data preparation for Model- contd…
 Precipitation
Thiessen polygon for 71
stations with TRMM data
are used.
 Temperature
15 stations temperature is
transferred using
temperature lapse rate.
100
HH
0.65xTT nr
rn


where,
= Temperature at new station
= Temperature at recorded station
= Elevation of recorded station
= Elevation of new station
nT
rT
rH
nH

Data preparation for Model- contd…
 Potential Evapotranspiration
The method is based on sound theoretical reasoning and is
obtained by a combination of the energy balance and mass
transfer approach which is given as:
 Hypsometric Curve
Catchment area is divided into 10 elevation levels. At each
zone the model computes air temperature, amount of
precipitation, precipitation type, snow melt based on air
temperature and temperature lapse rate.
γA
γaEAH
PET
n



% Area
less than
0 10 20 30 40 50 60 70 80 90 100
Elevation
(m)
170 750 1051 1420 2044 2905 3798 4441 4966 5446 8148

Model simulation and Results
The goodness of fit is evaluated by a function built using the Nash-
Sutcliffe efficiency (R2) and is given as :
   
 



 2
00
2
0S
2
002
QQ
QQ.QQ
R

Model simulation and Results- contd…
Parameters- Confined and Free
Name Meaning
Value
adopted
Units
Tx Threshold temperature Rain/Snow -1.2 °C
Ts Threshold temperature Snowmelts 0 °C
Cx Degree-day-factor 3 mm/°C*Day
CFR Re-freezing efficiency in snow 0.01
PKORR Precipitation correction -Rainfall 1.15
SKORR Precipitation correction -snowfall 1.15
TTGRAD Temperature lapse rate for clear days -0.9 °C/100m
TPGRAD Temperature lapse rate for precipitation days 0.65 °C/100m
PGRAD Precipitatin lapse rate 2
FC Field capacity in soil moisture zone 500 mm
LP Threshold value for potET in soil moisture 80% % of FC

Model simulation and Results- contd…
Parameters- Confined and Free
Name Meaning
Value
adopted
Units
β Parameter in soil moisture routine 1.5
UZ1 Threshold level for quick runoff in upper zone 40 mm
KUZ1 Recession constant in upper zone 0.1 5 1/day
KUZ Recession constant in upper zone 0.11 1/day
PERC Percolation from upper to lower zone 3 mm/day
KLZ Recession constant in lower zone 0.045 1/day
UZ2 Threshold level for quick runoff in uppermost
zone
300 mm
KUZ2 Recession constant in uppermost zone 0.2 mm/day

Model simulation and Results – contd…
 Changing KLZ will change the base flow (slow runoff) and hence the
simulated flow in the model.
 By changing PERC, the soil moisture in the upper zone will change and
hence the runoff from the upper zone. If PERC is decreased, runoff will
increase and vice-versa.
 If PKORR is reduced, the peak runoff will be reduced drastically as it is
the main component responsible for runoff generation, in the lower zone
and vice-versa.
 If KUZ1 is changed, the runoff from upper tank will change and hence the
total runoff from upper zone. This will change the faster runoff so the plot
will be steeper as the value changes.
 If UZ1 is changed then the runoff response will be increased or decreased
based on whether the threshold (UZ1) is decreased or increased.
 If KUZ2 is increased, then the runoff from the lower tank of the upper
zone is increased and hence the total runoff and vise-versa.
 IF FC is decreased, then the output to the upper zone is increased which
will add significantly in the fast runoff generation and hence vice-versa.

 If BETA is increased, then output to upper zone is decreased and hence
the runoff generation and vice-versa.
 For the given soil moisture content, if LP is decreased then actual
evaporation will be nearly equal to the potential evaporation and hence
the total runoff will not be changed. But, if it is increased, then Ea will
be reduced thus increasing the output to the upper zone.

Calibration Result:
parameter set manually
Calibration Result:
parameter set by model

Validation Result 1
Validation Result 2

CONCLUSION AND RECOMMENDATIONS
 TRMM precipitation product is assessed against the observed gauge
rainfall data for the period from 2001 to 2008 over 264 gauge station of
Nepal.
 Point to pixel comparison method is adopted for the evaluation of
satellite products.
 TRMM data’s seemed to be underestimating which is stated by different
statistical parameters.
 TRMM data without any bias correction is fed as input rainfall data’s to
simulate the runoff at outlet of Narayani Basin.
 The model is run for five years from 2004 to 2008 during calibration
and run for two years from 2009 to 2010 during validation.
 The model simulation showed quiet acceptable result.
Conclusion:

CONCLUSION AND RECOMMENDATIONS
 TRMM satellite product is underestimating precipitation in most of the
regions over Nepal. An appropriate correction factor should be
developed that could take an account of change in elevation. It is wise to
study the reason behind the underestimation of TRMM satellite data.
 The study is based on point to pixel comparison. Study on pixel to pixel
analysis can be carried out as well.
 Since the results showed better monthly results while comparing with
gauge data’s but the model was based on daily analysis. Monthly model
would probably give better simulation results.
 The model proposed is conceptual model. Distributed model which
include detail soil profile, land use etc. would probably meet the exact
scenario.
 Another satellite product such as CMORPH can be proposed which
produces precipitation analysis at a very high spatial and temporal
resolution.
Recommendations:

069MSW405_Devendra Tamrakar_Presentation

069MSW405_Devendra Tamrakar_Presentation

More Related Content

What's hot (20)

Viewers also liked (20)

Similar to 069MSW405_Devendra Tamrakar_Presentation (20)

069MSW405_Devendra Tamrakar_Presentation

Editor's Notes