Annual Report of the project

MONITORING THE HEALTH OF MANGROVES OF

MAHARASHTRA STATE USING NEAR REAL TIME
SATELLITE REMOTE SENSING DATA

Submitted to

Marine Biodiversity Conservation and mangrove Cell,

Department of Forest, Government of Maharashtra, Mumbai

March 2019

INDIAN INSTITUTE OF SPACE SCIENCE AND TECHNOLOGY

Department of Space, Government of India
Thiruvananthapuram
MONITORING THE HEALTH OF MANGROVES OF MAHARASHTRA STATE
USING NEAR REAL TIME SATELLITE REMOTE SENSING DATA

1. Introduction:

Mangrove forest is one of the ecologically and economically important ecosystems that thrive in
brackish water environment of tropical countries. Worldwide mangrove forest cover was
declining due to both natural and anthropogenic causes till 2/ 3 decades earlier. However,
recently the importance of mangroves is realised and global and regional level conservation
strategies are being developed for sustainable management of the mangrove vegetation cover and
the ecosystem.

Globally mangroves spread for about 150,000 sq.km with maximum cover found in South East
Asian countries followed by South America and North and Central America (World Mangrove
Atlas, 2010). Indian mangroves contribute 46% of South east Asian mangroves with an extent of
4921 sq.km (SFR, 2017). Though there exists random pattern of the mangrove cover from 1987
to 2015, gradual increase is seen during last five years.Based on the status of Forest Report,
Forest survey of Indian, mangroves of Maharashtra is spread for about 304 sq.km in the 6 coastal
districts namely Mumbai City and Sub-urban, Raigarh, Ratnagiri, Sindhudurg, Thane and
Palghar. About 106 sq.km of mangroves are mapped in Raigarh district followed by Thane
(90sq.km) and Mumbai Sub-urban (60 sq.km).Though FSI reports the district wide mangrove
forest cover, it is mapped at 1:50,000 scale using coarse resolution satellite data with identified
limitations.

Present study is aiming to map and monitor the extent of mangroves of Maharashtra at larger
scale (1: 10,000) using very high resolution satellite data ; for which the study uses mangrove
maps prepared by Maharashtra State Remote Sensing Centre using high resolution Cartosat1
satellite data of 2005/ 2006 as base reference. Second major objective is set to assess the health
of mangrove vegetation using near real time open satellite data by developing an application
tool.

This report gives the progress of the outcome of the project for the period April 2018 – March
2019. Separate expenditure statement of project for the above period is submitted separately.
2. Study area:

The mangroves of Maharashtra coast extending all along the 6 coastal districts: Mumbai &
Mumbai sub-urban, Raigarh, Ratnagiri, Sindhudurg Thane and Palghar. Out of the total coastal
length of 720km, 70% of the coast has patches of mangrove vegetation spreading inwards around
44 creekswith major extent in Thane creek, followed by 10 other major rivers namely Manmad,
Kundalika, Amba, Vaghthane (Vijaydurg Creek), Shastri (Jaigad Creek), Vashishti (Dabhol
Creek),Savitri (Bankot Creek), Ulhas, Vaitarna and Damanganga. Also there are few coastal
region having elevation of 10m or more from msl within a kilometre from coastline. Such coast
of the state is devoid of mangroves vegetation.

Figure 1. Location of the mangroves (given in green) along the coast of Maharashtra.
3. Data used

Present study involved various types of satellite data (open and commercial), base map and field
data. Here we categorise the data into two types namely (i) Primary and (ii) Secondary data.

3.1. Primary data:

Data collected by the study team directly and the satellite digital data are categorised as primary
data and are listed below.

1. World View Satellite data of 2018

2. Open satellite data – Landsat series date from 2009, 2012, 2014 and Sentinel of 2017,
Google earth images.
3. Ground truth data collected from sample locations (Figure 2) on vegetation types,
species, biophysical properties and soil

Table 1. Satellite data used in the study.

SN Year Satellite/ Sensor

1 2009 Landsat5 TM
2 2012 Landsat5 TM and Landsat7 ETM
3 2014 Landsat 8 OLI
4 2016 Sentinel 2
5 2018 WorldView 2/3
Data from 1 – 4 are open data and 5 is the commercial data procured through National Remote
Sensing Centre, Hyderabad, the nodal agency of India to procure satellite remote sensing data
after Ministry of Defence clearance for very high resolution data. As Word View data has very
high resolution, the entire coast of the state was not covered in single time of acquisition but
spanned with the year. However, due care was taken to acquire the data in nearby date.

3.2. Secondary data:

The data collected from secondary source i.e., collected and compiled by third person/
organisation are categorised as secondary data. Such data used in this study are listed below.

1. Survey of India topographic maps at 1:50,000 scale

 2. Mangrove vegetation cover map prepared by MSRAC at 1:10,000 scale
3. Google map for spatial data of the study area
4. Literatures on the mangrove vegetation and its ecology of the study area

5. Methods
5.1. Mapping and Monitoring of mangrove vegetation

Different types of systematic satellite data analysis were implemented in mapping the mangroves
using various resolution satellite data. Following illustrates the procedure involved in preparing
the data for current study.

- More than two scenes of satellite data are required to cover the entire study area. Hence,
the cloud free satellite data covering the study area for the particular year is identified to
fall in nearby dates.
- Such identified scenes are joined and the study region covering the mangroves alone is
extracted from the whole scenes.
- The Landsat and Sentinel images are classified digitally using supervised classification
methods for the vegetation density mapping (Figure 2) and contextually edited to
improve the accuracy. This analysis is carried out for the entire study area.

Forest density classification procedure

Forest Survey of India (FSI) gives following definition to map the forest density classes of
mangroves which are in par with the terrestrial forest classes. However, there is no statistics on
the features of mudflats and tidal flats in the mangrove estimates given by FSI.

Table 2. Forest density classes as given by Forest Survey of India.

SN Density classes Definition
1 Very Dense (VD) forest mangroves with more than 70% canopy cover;
2 Moderately Dense (MD) mangroves with 40 – 70% canopy cover;
3 Open Forest (OF) mangroves with 10 - 40% canopy cover;
Additional classes used in the present study
4 Mudflat/ Tidal flats mangroves with less than 10% canopy cover or
devoid of vegetation inside the mangrove patch
or along the creek and coast.
 5 Salt pan/ Manmade Regular shaped water bodies near to mangrove
ponds forests
6. Other vegetation Other terrestrial vegetation lying associate with
mangroves

- Multispectral World View data having a spatial resolution of 2 m of Mumbai and

Mumbai Sub-urban covering part of Thane creek and Malad, Manori creeks are visually
interpreted and mapped for its density classes at 1:5,000 scale. Minimum mapping unit
adopted in this study is 3 pixels x 3 pixels i.e., the minimum size of the feature which can
be mapped at this scale is 6m x 6m. The density level mapping does not consider the
height of the mangrove vegetation as the criteria but only the canopy cover of mangrove
vegetation.
5.2. Accuracy assessment:

Mangrove wetland maps prepared are validated for the classification accuracy with sample
regions, ground truth data and collateral data collected from Mangrove Foundation, Mumbai.
Sufficient number of samples are collected from satellite data and surveyed during field survey
(Field survey Reports, November 2018 and February 2019: Figure 3). Overall accuracy and
kappa coefficients were estimated to validate the mangrove density maps prepared using
Landsat, Sentinel and World View data.
 Figure 2. Sample mangrove forest area of Thane creek shown in False Colour Composite (NIR , R, G are
displayed using RGB colours) illustrates themethod to map mangrove vegetation density classes. Very
Dense (VD)– mangrove with more than 70% canopy cover; MD – Moderately Dense mangrove with 40 –
70% canopy cover; OF - Open Forest mangrove with 10 - 40% canopy cover ; MF – Mud Flat with less
than 10% canopy cover or devoid of vegetation.

5.3. Development of Geospatial Tool to assess the health of the mangroves using near real
time open satellite data

As illustrated in the proposal development of this tool involves various independent components
of health indicators. The tool is developed in python. There are basically 4 types of indicators
namely Vegetation Health Index, Canopy closure, Fragmentation, Sedimentation and Tidal
inundation. The front end of the tool is developed with one of the indicators - Vegetation Health
Index in standalone mode. Development of other indicators are in progress.

5.3.1. Vegetation Health Index:

This indicator is derived based on vegetation condition (VCI) and temperature condition (TCI)
normalised over a particular interval of time using time series satellite data on which is derived
using widely known Normalised Difference Vegetation Index (NDVI) and Land surface
temperature.

VHI = 0.5 VCI + 0.5TCI eq. 1

 Where VCI = NDVIi – NDVImin/ NDVImax - NDVImin eq. 2

and TCI = LSTmax – LSTi/ LSTmax - LSTmin eq. 3

suffix i represents the time at which the health is assessed, max and min are the maximum and
minimum values over the time period of study.

Figure 3. Map showing the sample locations in which ground truth and vegetation data collected during Novermber
2018 and February 2019.

5.3.2. Development of Application tool

The tools is designed to be user friendly. Current version of the tool takes open satellite data –
Landsat series depending on the user input. The user can identify the satellite data for which the
VHI has to be derived.

6. Results
Following section give the results obtained from satellite data analysis for mapping and
monitoring the mangroves of the study area and few outcomes of the Health Assessment tool.
6.1. Mapping and Monitoring the mangroves (2009, 2014 and 2017).

Multispectral satellite data based mapping of Maharashtra mangroves for the periods 2009, 2014
and 2017 show that there is a gradual increase in overall area of mangrove vegetation from
213.83 sq.km to 363.27 sq.km (against 304 sq.km by FSI).

Table. 3. Area of mangrove forest density classes present in the coastal district of Maharashtra for the
period 2009, 2014 and 2017.

Area (Sq.km)
District 2009 2014 2017
VD MD OF Total VD MD OF Total VD MD OF Total
Palghar 9.37 21.72 13.51 44.60 17.73 19.95 7.86 45.54 13.20 20.21 22.94 56.34
Thane 23.09 19.93 8.74 51.76 29.15 15.20 6.84 51.19 24.46 19.36 11.64 55.46
Mumbai -
Suburban 30.03 20.91 6.08 57.01 36.67 17.93 5.58 60.18 37.63 16.84 9.86 64.33
Raighar 49.75 56.70 34.37 140.83 65.40 40.43 27.63 133.46 61.06 65.95 34.64 161.65
Ratnagiri 25.38 7.80 5.05 38.23 29.51 8.05 7.94 45.50 62.08 4.29 4.08 70.45
Sindhudu
rg 26.60 0.43 0.00 27.04 12.37 6.27 4.10 22.75 30.99 1.85 2.15 35.00
Total 164.22 127.49 67.76 359.46 190.82 107.84 59.94 358.61 229.43 128.50 85.31 443.24

Raighar district has the maximum mangrove cover of about 36% of the total mangrove cover of
the state followed by Ratnagiri and Mumbai & Sub urban districts with 16% and 14.5%
respectively (Figure 4). This result has significant shift from the report of Forest Survey of India
for the mangroves of Maharashtra by additional of more than 100 sq.km. Such deviation could
be due to the following reason: Very Dense mangrove forest is not identified by FSI but in our
study the maximum area is covered by Very Dense forest followed by Moderately Dense and
Open Forest. Hence the additional extent of Open Forest category could have added to this
marked difference.

Figures 5 to 23 show the district wise satellite data and mangrove density classes overlaid on
satellite data for the analysed period 2009, 2014 and 2017; and figures 24 to 29 show the change
in mangrove forest happened between 2009 and 2017.
 District wise distribution of mangrove cover (Sq.km)
180.00
160.00
2009 2014 2017
140.00
120.00
100.00
80.00
60.00
40.00
20.00
0.00
Palghar Thane Mumbai - Raighar Ratnagiri Sindhudurg
Suburban

Figure 4. District wise extent of mangrove forest cover for the year 2009, 2014 and 2017

The mangrove wetland cover change happened between 2009 and 2017 (Table 4) shows
significant changes in mangrove cover within density classes and with other associate land cover
features.

Table 4. Change in the extent of mangrove density classes within and between other associated land
covers (Area in Sq.km)

2017 Total Total mangrove

2009 Moderately 2009 (Sum of VD,
Very Dense Dense Open Forest Others MD and OF)
Very Dense 110.5 31.8 5.8 16.3 164.4 148.1
Moderately Dense 37.0 51.0 14.2 25.2 127.4 102.1
Open Forest 5.3 11.0 12.8 37.9 67.1 29.2
Others 77.7 34.6 52.0 4861.4 5025.7
Total 230.5 128.5 84.8 4940.8 5384.5
Net change in
Total mangrove 2017
(Sum of VD, MD and mangrove =
OF) 152.8 93.8 32.8 +84.93

Within mangrove density classes, change matrix shows no change for about 130 sq.km for all 3
density classes, density reduction for 51.8.sq.km and density increment for 53.3sq.km. About
164.3 sq.km extent of mangrove forests regeneration is noticed by change from other land cover
features to either VD or MD or OF at the same period about 79.4 sq.km of mangrove forests
(VD, MD and OF) have been converted to other land cove features.
Figure 5. Mumbai and Suburban 2009 - (i) False Color Composite and (ii) FCC overlaid with Mangrove and associated features
Figure 23. Change in mangrove cover between 2009 and 2017 – Mumbai and Mumbai SubUrban
Figure 24. Change in mangrove cover between 2009 and 2017 – Raighar
Figure 25. Change in mangrove cover between 2009 and 2017 – Ratnagiri
Figure 26. Change in mangrove cover between 2009 and 2017 – Sindhudurg
Figure 27. Change in mangrove cover between 2009 and 2017 – Palghar
Figure 28. Change in mangrove cover between 2009 and 2017 – Thane
Accuracy assessment:

The classification of satellite images are validated for the accuracy using testing regions. Accuracy of
the classification was estimated to be 82.4%, 83.5% and 85.5% for mangrove density mapping with
2009, 2014 and 2017 satellite data. Classified maps were further improved through contextual editing
based on the field knowledge and ground truth information. Thus the accuracy of final maps achieved
is better than above mentioned accuracy. However there exists about 5% of uncertainly level in the
transitional area of mangroves and terrestrial vegetation which will be addressed while mapping with
high resolution World View data at better scale.

6.2. Mangrove mapping at 1:5000 scale using high resolution satellite data (World View)

Mangroves of Mumbai and Mumbai Sub-urban was visually interpreted for the vegetation density
and associated wetland features using World View 2/ 3 multispectral data. Preliminary estimates of
the extent of Mumbai mangroves mapped show that total mangrove cover is about 59.1 sq.km in
2018 with 50.6 sq.km of dense mangroves, 6.7 sq.km of sparse mangrove and 1.8 sq.km of open
mangrove forest cover (Figure 23). This needs to be verified with more number of sample points and
field knowledge.
Figure 29. Mangrove wetland and associate feature cover map of Mumbai and Sub-urban in 2018
Figure 30. Mangroves of part of Thane Creek (left) Manori and Maalaad Creek (right) in Mumbai Sub-urban region
6.3. Development of Geospatial tool to assess the health of mangroves

As per the proposed application tool, the first level indices directly derived from satellite data are
Normalised Difference Vegetation Index (NDVI) and Normalised Difference Water Index (NDWI).
Based on that, second level health indices are derived to monitor the mangrove and its ecosystem
health namely, Vegetation Health Index, Canopy closure, Fragmentation, Water spread Index and
Patchiness. The progress in the application development is made for the development of first level
indices and Vegetation Health Index using time series satellite data. Some of the screen shots of the
interface of Application tool under development and few of the outcomes of health indices for the
mangroves of Thane creek are given below.

Figure 31. User interface to assess Mangrove Health Index using time series remote sensing data
Figure 32. User interface to assess Structural health indices using time series remote sensing data
Figure 33. Brightness Temperature, NDVI and NDWI of the Thane Creek region
Figure 34. Vegetation Condition Index (VCI), Temperature Condition Index (TCI) and Mangrove Health Index (VHI) for 2014

----- ----- ----- ---- ----- ----- ------ ------ ------ ---
Status of the project activities (grey shade shows the percentage of work done)

Progress of the activity 0...........................50%.......................100%

1. Time series density level mapping of the

mangroves of Maharashtra and change detection
2. Large scale mapping (1:5000)
 Procurement of World View Data
 Mumbai and Mumbai Sub Urban
 Other districts
3. Field Data collection
4. Developing Mangrove Health Assessment application
 Deriving satellite data based vegetation indices
 Static base line health indices
 Coordinating to get Expert input on mangrove
health
 Development of geospatial tool
Annexure:

Field Data survey Report

It is important to relate the ground reality to the image data for better calibration of
remote sensing data and improving the results. As a part of the project “Monitoring the health
of mangroves of Maharashtra state using near real time satellite remote sensing data” the first
field visit and data collection was carried out from 13 to 17 November 2018 and the second
field survey during 7 – 17 February 2019. The first field visit was concentrated mainly
around the Greater Mumbai and Mumbai suburban area and also some part of Raigad and
Thane (Fig 1) and second visit was carried out in the mangroves of Raighar, Ratnagiri and
Sindhudurg. A team of three members

Figure 1. Map showing the sample locations in which ground truth and vegetation data collected during
Novermber 2018 and February 2019.

from IIST and Mangrove Foundation visited various major creeks in the area where most of
the Mangrove vegetation appeared in the satellite imagery and collected various field
measurements from these sites. The major
 Figure 2. Mangrove of Thane creek

areas covered in the field include Thane creek, Karanja creek and Malad creek. These creeks
receive a large amount of sewage and waste water from various drains. Mangrove vegetation
present in the southern districts along Dharmata creek, Savitri River, Vashishti, Kajali Rivers,
Jaigat port, Gad and Naringre Rivers are surveyed during second field visit.

Major Species Found in the area

The major mangrove and associated flora found in the Thane, Karanja and Malad creeks are
listed in the Table 3.

Table 3: Mangrove flora found in the Thane, Karanja and Malad Creek.

SL Species Common name Family Growth Form

No.
1 Aegicerascorniculatum Kajala, Myrsinaceae Shrub/Small
Sugandha Tree
2 Avicennia marina Tiwar Avicenniaceae Shrub/Tree
3 Avicenniamarina var. Tiwar Avicenniaceae Shrub/Tree
accutisima
4 Avicennia officinalis Bhartiya Tiwar Avicenniaceae Tree
5 Bruguieracylindrica Lahanzumbar Rhizophoraceae Tree
6 Excoecariaagallocha Geva, Hura Euphorbiaceae Large shrub/
Small Tree
7 Rhizophora apiculata MothaKandal Rhizophoraceae Tree
7 Rhizophora mucronata Lal Kandal Rhizophoraceae Tree
8 Sonneratiaapetala Chipee Sonneratiaceae Tree
Associated Species
1 Acanthus ilicifolius Kateri, Marandi Acanthaceae Shrub
2 Clerodendruminerme Vanajai Lamiaceae Shrub
3 Salvodorapersica Mesvak Salvadoraceae Shrub/Small
Tree
4 Derris trifoliata Karanjvel Fabaceae Climber
In each plot geographic location in terms of latitude and longitude, species present, average
canopy height, chlorophyll readings, and spectra of species were taken. The growth and well-
being of mangroves are very much affected by their soil and water quality. For testing the soil
and water quality, soil and water samples were also taken (Fig.3).

Figure3. Soil and water samples collected

Chlorophyll concentration of each species may vary with the site condition. Chlorophyll
readings of various species of mangroves in the area were noted (Fig.4) and were shown in
Table 4. The overall chlorophyll concentration in the species ranged from 370 to 800
µmol/m2, but there was significant difference in chlorophyll concentration within species at
different location.

Table 4: Chlorophyll concentration of mangrove species

SL No. Species Chlorophyll conc.(µmol/m2)

1 Aegicerascorniculatum 269.8 to 638.4
2 Avicennia marina 362.7 to 615.6
3 Avicenniamarina var. accutisima 379.7 to 476.52
4 Avicennia officinalis 343.7 to 679.9
5 Bruguieracylindrica 598.7
6 Rhizophora mucronata 377.22 to 800.7
7 Sonneratiaapetala 616.6
Each and every species has its own characteristic spectra. It is important to have a clear idea
on the spectral signature of various species of mangroves for better differentiation and also to
assess their present health status. The spectra were collected using the ASD Fieldspec3 Field
Spectroradiometer (Fig. 5). The sample spectra collected from the study area are shown
below in Figure 6.

Figure 4. Taking Chlorophyll concentration in leaves of Mangrove species

Figure 5. Taking Spectra of canopy reflectance of Mangrove species

 Figure 6. Field canopy spectra of mangroves and associated species

Physical measurements are essential for studying the biomass and productivity of
particular mangrove ecosystem. Measurements like diameter at breast height, height of tree
were taken along with species name in some of the plots in the study area. Forking was a
common feature seen in the mangroves in all plots. The diameter ranged from 3 cm to 40 cm
and average canopy height was in the range of 3 m to 10 m.

Summary

As part of project titled „Monitoring the health of mangroves of Maharashtra state

using near real time satellite remote sensing data‟, first data collection was carried in and
around Mumbai region November 2018 and southern district in February 2019. The field was
mainly planned for the preliminary assessment of the study area. Occurrence of various
mangrove species along with their location, canopy spectra of each species in the area,
chlorophyll concentration in the leaf of each mangrove species, bio physical measurements
for biomass estimation, were collected during the field. To assess the quality of the mangrove
environment, soil and water samples were also collected. Field enabled the collection of basic
information on various aspects of this study; detailed enumeration is required for the
accomplishment of study objectives.