Research Article

Cartosat-1 Image Segmentation Technique for Shade Tree Crown Density in Tea Gardens of East India in Relation to Terrain Geometry  

Dibyendu Dutta1 , Libeesh Lukose2 , Anju Bajpai1 , Uttam Bhunia1 , Rajkumar Singh1 , Sourav Samanta1
1 National Remote Sensing Centre (ISRO), PO: Balanagar, Hyderabad, Telengana State, India
2 Indian Institute of Technology, Kharagpur, West Bengal, India
Author    Correspondence author
Journal of Tea Science Research, 2018, Vol. 8, No. 1   doi: 10.5376/jtsr.2018.08.0001
Received: 10 May, 2018    Accepted: 04 Jun., 2018    Published: 14 Jun., 2018
© 2018 BioPublisher Publishing Platform
This is an open access article published under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Preferred citation for this article:

Dutta D., Lukose L., Bajpai A., Bhunia U., Singh R., and Samanta S., 2018, Cartosat-1 image segmentation technique for shade tree crown density in tea gardens of east India in relation to terrain geometry, Journal of Tea Science Research, 8(1): 1-9 (doi: 10.5376/jtsr.2018.08.0001)

Abstract

One of the factors determining tea quality is shadow casting by the shade trees. Besides regulating incoming solar radiation shade trees also helps maintaining the moisture in soil and nutrient recycling. However the optimum shade density depends upon the elevation, slope and aspect. In the present study image segmentation technique was employed on Cartosat-1 data to capture the vertical crown density of the shade trees. Significant positive correlations (r2=0.91) were found between observed and measured vertical crown density. Based upon the crown density the tea gardens were classified. Further the relation between crown density and terrain parameters has been analysed. Significant negative correlation was observed with elevation (-0.590) and slope (-0.627) which indicates that to increase in elevation and/or percent slope the shade density decreases.

Keywords
Tea; Cartosat-1; Shade tree; Crown density; Image segmentation; Terrain geometry

Background

Tea (Camellia sinensis L.) is one of the major commercial crops in India and constitutes 21% of global tea area and 28% of world tea production (Majumder, 2010). The Indian states of Assam, West Bengal, Meghalaya, Tripura and Sikkim in the north-east and Tamil Nadu, Karnataka and Kerala in the south contribute significantly to the overall tea production in the country. Tea is a shade loving crop and hence often it is adapted to the shade of understoried trees (Obaga,1984; Carr, 1992). Shading not only limits the direct solar irradiance but also provides a unique microclimatic development leading to good quality leaf production. Shade trees not only have capacity for the beneficial modification of micro-climatic parameters (atmospheric temperature, RH, soil temperature, net radiation, photosynthetically active radiation) (Brunig, 1983; Huxley, 1983), but also enhances complementary resources sharing, functional diversity, carbon sequestration, soil fertility, drought resistance as well as suppression of weed growth and lowering the infestation of pests and diseases (Tscharntke, 2011). Chowdhury et al. (2015) have shown that shade trees play important role in productivity under the of sub-Himalayan regions of West Bengal. Studies regarding the importance of shade pattern and selection of shade trees were carried out by several researchers like (Visser, 1961; Hadfield, 1974; Mohotti, 2004). Mukherjee et al. (2008) have shown that atmospheric and soil temperature were lowered by 2-3°C compared to a non-shaded open condition, whereas relative humidity increased by 3-9% within the shade. However, the shade trees need to be managed properly to maximize their benefits. If the humidity of shaded tea plantations remains unchecked and continues to be high, tea plants become susceptible to pests like mosquito bug and the blister blight disease. As a garden management practice the tree stands are periodically pruned to stabilize the shadow casting on the ground. Normally the shade trees grow as companion trees along with tea bushes in the shaded perennial Agroforestry system. To maintain the export quality of Indian tea there is a need to periodically monitor the micro-environment of the tea bushes created by the shade trees to harvest best quality tea leaves.

 

Satellite remote sensing is a pragmatic approach to monitor the dynamic landcover, growth and condition of tea bushes over a large area by virtue of synoptic, periodic and multi-spectral capability, which is otherwise laborious, time consuming and cost prohibitive. The gardens appear distinct in the satellite images due to their regular shapes, texture (multi-layered canopy of tea and shade trees) and tone. The Indian Remote Sensing satellite Cartosat-1 data have capability to capture the shade tree canopies in the tea gardens under mixed stand by virtue of its pattern, texture and association. Combining the characteristics of the Cartosat-1 image with advanced image processing technique, e.g. image segmentation and objective oriented classification the shade tree canopies can be picked up and represented as the percent canopy cover with respect to the section area which provide valuable information on solar energy inception by the tea bushes. The information acquired through satellite remote sensing can be stored as Geo-database, which can be retrieved, appended, updated and modelled based on the requirement. Remote Sensing data and Geographic Information System (GIS), is considered as a flexible, efficient, speedy, cost-effective and reliable technology for obtaining information on natural resources analysis and modelling. It is an integral part of data management in large number of applications.

 

In the present study Cartosat-1 images of the tea gardens of north Bengal have been segmented based on digital number (DN) thresholding and connectivity using ENVI image processing software. The objectives ware i) to study the feasibility of the Cartosat-1 image segmentation technique for estimation of shade tree canopy density and ii) the interrelation between canopy/crown density with terrain properties.

 

1 Study Site

Darjeeling is the northern most district of West Bengal, lies between 26o27¢N & 27o13¢N latitude and 87o59¢E & 88o53¢E longitude covering an area of 3,149 sq. It shares international boundaries with Bhutan (in the north-east), Bangladesh (in the south-east) and Nepal (in the west). In its north it shares boundaries with Sikkim state, in the east with Jalpaiguri and south by North Dinajpur districts of West Bengal. The district has a temperate climate with wet summers caused by monsoon rains. The mean annual temperature is 24°C in the plains and less than 12°C in the hills. In the hills during summer the temperature reaches 16-17°C and during winter it drops to 5-6°C. The average annual precipitation is 3,092 mm, with an average of 126 rainy days in a year. In general, the rainfall increases from west to east. The orographic factor also causes vertical zonation of temperature and decrease in precipitation. The southern slopes of the ridges get much higher (4,000-5,000 mm) precipitation than the leeward sides (2,000-2,500 mm).

 

The hilly areas of Darjeeling district are located within the Lesser and Sub-Himalayan belts of the Eastern Himalayas. The relief varies from 100 m above sea level to more than 8,000 m. The southern foothill belt is demarcated by a highly dissipated platform of terrace deposits extending along the east west axis. The area is primarily composed of erosional landforms produced by southerly flowing streams, which have exposed a full cross section of different tectonic units. The important rivers include Teesta, Great Rangit, Mechi, Balason, Mahananda, Lish, Gish, Chel, Ramman, Murti and Jaldhaka. The soils are mainly composed of sandstone and conglomerate formations, which are the solidified and upheaved detritus of the great range of Himalayas. However, the soil is often poorly consolidated and is not considered suitable for agriculture. The soils that have developed in the Kalimpong area are predominantly reddish in color. Occasional dark soils are found due to the extensive existence of phyllitic and schists. Soils in the highlands stretching from the west to the east of the district along most of the inter-fluvial areas are mainly of mixed sandy loam and loamy type, while those on the southern slopes of Mirik and Kurseong are mainly clayey loam and reddish in color. Sandy soils are mainly found in the east of the river Teesta. The soils over Darjeeling gneiss is characterized by a high proportion of potassium derived from feldspar and muscovite mica. This soil is poor in lime, magnesium, iron oxides, phosphorous and nitrogen. Based upon the Natural Resources Census (2015), the agricultural land accounts for 25.58% of the district area. The area under different types of forests together accounts for 59.59% of the district area out of which deciduous forests are the dominating types. Plantations/orchards/tea plantations cover 9.68% of the district area, whereas the area under wasteland is 1.19%.

 

Jalpaiguri (erstwhile) is the largest district of North Bengal confined between 26°16'N to 27°00'N latitude and 88°04'E to 89°53'E longitude and covers an area of 6,245 sq km. The district shares international borders with Bhutan and Bangladesh in the north and south-west respectively. It joins the state of Assam in the east, the Darjeeling district in the west and north-west and Cooch Behar district in the south. The district is characterized by monsoon climate and experiences three distinct seasons, viz. summer, monsoon and winter. May is the hottest month with an average maximum temperature of about 32°C. January is the coldest month with minimum temperatures of 11°C. Average annual humidity and rainfall are 82% and 3,160 mm, respectively. The thunderstorm is a common local weather phenomenon during May.

 

Jalpaiguri district is bounded in the north by the hilly ranges of the Himalayas and in the south by the piedmont plains. The upper part of the district mainly consists of Siwalik and older Quaternary formation, which are dominated by thick boulder and conglomerated horizons. The lower portion occurs as a fluvial terrace deposit. The recent sediments, mainly represent thick piles of fluvial, unconsolidated sediments. The entire topography is crisscrossed with rivulets, rivers and hills. The major rivers are Teesta, Torsa, Mahananda, Jaldhaka, Kaljani, Raidak, Korotowa and Sonkos. These are joined by smaller streams like Galandi, Dudua, Tasati, Mujnai and alike. The most of the rivers flow in north-south direction and originate from Sikkim, Bhutan and Darjeeling hills (except Torsa which originates from Tibet). The soils of the Terai regions in the Himalayan foothills are partly developed, mainly formed of young alluvium on the fans of the foothills. Shallow to moderately deep and at places deep with medium to fine texture. These soils are prone to severe runoff and flood hazards associated with low water holding capacity, un-decomposed organic matter, low available nitrogen and phosphorus but high potassium contents. Geographically, this type of soil is found in the northern fringe of the district. The soils of the alluvial plains of Teesta, Torsa and Mahananda rivers are moderately deep to deep, coarse to fine loamy in texture, formed by recent alluvial deposits, have low water holding capacity, low pH, nitrogen and cation exchange capacity. These areas also face problems of water logging and severe flood hazard. Based upon the Natural Census Report (2015) agricultural land accounts for 48.84% of the district area. The area under different types of forests together is 26.14% out of which deciduous forests are the dominating type. Plantations/orchards/tea plantations cover 13.61% of the district area. The area under wasteland, including all categories is 1.61%. The summary table containing the range of geographic extents and area of the tea gardens are given in Table 1. Altogether 90 tea gardens of Darjeeling and 130 gardens of Jalpaiguri district were selected for the study. In Darjeeling Monteviot is the smallest garden of area 0.79 sq km and largest one is Badamtam (12.43 sq km). Whereas in Jalpaiguri district Gopimohan is the smallest garden (0.53 sq km) and Birpara is the largest one (14.91 sq km).

 

 

Table 1 Summary table-geographic extent with area of tea gardens

 

2 Materials and Methods

2.1 Satellite data used

Cartosat-1 data of 2.5 m spatial resolution with a swath width of 30 m and 10 bit radiometry were used for the analysis. Altogether 113 Cartosat-1 scenes were used in the present study. The Cartosat-1 is a circular, polar sun-synchronous satellite orbiting at an altitude of 618 km with a spectral band width of 500-700 nm. The local overpass time is at 10:30 am. To avoid long shadow length which is a function of latitude/longitude and season the images which are of summer was only used. The data are pertaining to the year of 2007-08.

 

2.2 Survey of India topomaps (1:50,000 Scale)

Survey of India Topomaps of 1:50,000 scale was used for locating the Tea gardens on the satellite data and extraction of garden boundaries and basic details in conjunction with the tea garden maps provided by the Tea Board of India.

 

2.3 Tea garden maps

Tea garden maps, of different scales, were procured from Regional offices of Tea Board located at Darjeeling, Jalpaiguri and Kolkata along with section details and basic garden information.

 

2.4 Registration of satellite data

Cartosat-1 data was registered based on the GCP’s collected using DGPS. Well distributed DGPS points across the scene were used to register the data. A second order polynomial model with nearest neighborhood re-sampling technique was used for the purpose. The preferred GCP’s were road intersection and road-rail intersection.

 

2.5 Registration of tea garden maps

Tea garden maps were scanned with a colour scanner at 200 dpi and saved as .tiff format. The maps were Geo-referenced with respect to Cartosat-1 data. Mostly, the intersections of the section boundaries were taken for GCP. About 25 to 30 well distributed points were taken and 2nd order polynomial with the nearest neighborhood technique was used for registration. The map after registration was overlaid on the image for checking registration accuracy.

 

2.6 Digitization of tea gardens and section boundaries

The garden and section boundaries were digitized from the garden maps and refinement was done keeping Cartosat-1 data in the background. After digitization the vectors were checked for errors viz. overshoot and undershoot. Once the vectors are corrected topology was created. After topology creation, section numbers were labelled based on the information provided in the tea garden maps.

 

2.7 Generation of terrain parameters

The mean elevation, slope and aspect angles of the gardens were determined using void filled SRTM 1 arc second digital elevation model (DEM) available in the public domain (http://search.earthdata.nasa.gov/). The DEM was used for generation of slope percent rise and aspect for the entire study area. Further garden wise terrain parameters were extracted using the zonal statistics function of ARC GIS 10.4. A dbase file is generated, containing the details of the gardens with elevation, slope and aspect information.

 

2.8 Image segmentation

Image segmentation refers to the process of partitioning a digital image into multiple segments (sets of pixels, also known as super-pixels). The goal of segmentation is to simplify and/or change the representation of an image into something that is more meaningful and easier to analyze. Each of the pixels in a region is similar with respect to any characteristic or computed property, such as color, intensity, or texture. Adjacent regions are significantly different with respect to the same characteristic(s). Cartosat-1 image segmentation was done based on digital number (DN) and connectedness or neighborhood. The inputs required include a range of DN values represented the shady areas and a number of contiguous pixels to be considered to capture the complete crown. The threshold values are scene specific and was decided interactively. The output of the segmentation was written in 8 bits format and overlaid on satellite data to check the correctness of the segmentation. Any over estimation/under estimation caused due to soil background was thoroughly checked and corrected accordingly. The total number of segmented pixels falling in each section multiplied by the area of each pixel gives the total area of the section covered by the crowns of shade trees. The area covered in a section divided by the area of the section multiplied by 100 gives the percent canopy cover for that section. A routine was developed using VB-6 and Geo-Processing tools of ARC GIS for generation of section wise count of pixels corresponding to shade tree canopy area. The routine uses zonal statistics and other spatial analysis tools of ARC GIS. The inputs required to run the program include Tea garden boundary, Section boundary or any other Reference Grids and segmented Caratosat-1 image. The output is in the form of .dbf file where the pixel count corresponding to the shade tree canopy is returned against the unique codes viz. system ID no. The values of pixel count are appended with the feature data based on unique ID. Further total pixel area is calculated from the pixel number by multiplying the area of each Cartosat-1 pixel. The total pixel area is presented as percent cover with respect to section area or any reference grid area. For convenience, shade tree densities have been classified into 4 categories viz. 0-10%, 10-20%, 20-30% and >30%.

 

3 Results

3.1 Spatial distribution of shade tree density

Considering all the gardens of Darjeeling district the tea garden area under low, medium, high and very high shade density accounts for 88.96%, 8.69%, 12.15% and 6.85% respectively. The spatial extent of shade trees under different canopy density category varies extensively depending upon the physiography. From Table 2 majority of the tea gardens have a low shade cover density with only 7 gardens have ‘very high’ shade density. 90 gardens have a low shade density, 33 medium, 27 high and 7 are very high category. Neither of the gardens have shade densities of more than 75% of the area under ‘medium’, ‘high’, and ‘very high’ category. Some of the gardens where shade cover density is ‘very high’ include Dagapur (1.39%), Girish Chandra (26.69%), Kalama (6.98%), Matidhar (3.43%), ordterai (3.71%), Pahargoomiah (2.15%) and Putinbari (1.72%).

 

 

Table 2 Number of tea gardens under different elevation classes

Note: Figures in the parenthesis indicates the percent of total gardens

 

Similar to the Darjeeling district in Jalpaiguri district the tea garden area under low, medium, high and very high shade density accounts for 48.51%, 31.11%, 15.92% and 7.65% respectively. The spatial extent of shade trees under different canopy density category varies extensively. From Table 3 majority of the tea gardens have low to high shade cover density, spatial extent of which varies from 25 to 50%. Neither of the gardens have shade densities of more than 75% of the area under ‘high’, ‘very high’ density category. Similarly, none of the gardens have shade cover extent between 50-75% under ‘very high’ category. Some of the gardens where shade cover density is between 0-20% only include Gurjangjhora, Jainti, Jalpara, Looksum, Raghu utkarsh, Srinathpur and Uttarsalbarie. In twenty-five gardens ‘very high’ density category is absent. Some of the gardens having a significantly large area under ‘very high’ density category includes Aibheel (46.65%), Moraghat (31.18%), Hope (28.56%), Nowera nuddy (27.47%), Nimtijhora (23.59%), Kohinoor (20.66%) and Mechapara (20.34%). On the other hand, the gardens having a large extent of ‘very low’ shade cover includes Uttarsalbari (100%), Gurjangjhora (99.55%), Raghu utkarsh (96.88%), Bhandiguri (94.89%), Rajpur (90.84%), Jainti (89.43%), Bagracote (86.88%), Baitgoorie (85.91%), Jadabpur (85.56%), Goodhope (84.79%), Bhatpara (84.46%), Baradighi (83.8%), Amarpur (81.78%) and Looksum (81.51%). The spatial distribution of shade tree density of Darjeeling and Jalpaiguri districts is given in Figure 1 and Figure 2 respectively.

 

 

Table 3 The number of tea gardens with varying spatial extents of shade tree crown density

 

 

Figure 1 Spatial distribution of shade tree density of Darjeeling district

 

 

Figure 2 Spatial distribution of shade tree density of Jalpaiguri district

 

3.2 Validation of the results

As a proxy to ground measurement, high resolution multispectral data from IKONOS satellite was used to measure the canopy/crown area with limited ground-based measurements. Altogether 39 sections from different gardens with varying terrain and shade tree density were selected from Dibrugarh, Tinsukia and Jalpaiguri districts. A .kml file was generated using ArcGIS and was loaded in Google Earth. Based upon the crown diameter the shade trees were grouped into 4 categories viz. 0-10 m (low), 10-20 m (medium), 20-30 m (high) and >30 m (very high) for ease of calculating the total area on the ground covered by the shade trees. Total crown cover area divided by the respective section area expressed as percent gives the shade tree cover density.

 

Correlation study was performed between shade tree density generated from Cartosat-1 image segmentation technique and measured area from high resolution satellite data. A Significant positive correlation was obtained (r=0.917**) between observed and modelled cover density (Figure 3) which indicates that the Cartosat-1 data can effectively be used for mapping of temporal dynamics of shade tree density.

 

 

Figure 3 Correlation between measured and modelled shade tree density

 

3.3 Terrain parameters

Different terrain parameters viz. elevation, percent slope and aspect angle were derived based on the SRTM 30 m digital elevation model. All the terrain parameters were grouped into fewer classes for ease of presentation. All the parameters presented here represents the average value of the garden, although marginal variation exists among the sections of each garden. To avoid the marginal variation the parameters were grouped in a few classes as is given in Table 2, Table 4 and Table 5.

 

 

Table 4 Number of tea gardens under different slope classes

Note: Figures in the parenthesis indicates the percent of total gardens

 

 

Table 5 Number of gardens under different aspect classes

Note: Figures in the parenthesis indicates the percent of total gardens

 

Most of the gardens in Darjeeling district (42.22%) are situated within 1,000-2,000 m altitude followed by 21.11% between 500-1,000 m (Table 2). In contrast, most of the gardens of Jalpaiguri district (37.69%) are situated in the lower elevation zone between 100 to 150 m following 23.85% in 200-500 m zone as most of the district is plain except the foothills of Bhutan region.

 

From Table 4 it is apparent that 65.56% tea gardens of Darjeeling district are located at very steep slope of > 35%, followed by 30% in the southern plain areas of the district. At Jalpaiguri district more than 81% tea gardens are located on a gentle slope (1-3%). There is almost no garden above 10% slope except one.

 

The aspect angle plays a major role in the incoming solar radiation. The slopes facing south-east and south-west receives maximum incoming radiation, whereas north-east and north-west facing slope receives less sunshine. In higher latitude sunshine hours is a limiting factor for plant photosynthesis. From Table 5 the slope of most of the tea gardens of Darjeeling district is south-east (21.11%) and east (18.89%) facing whereas in Jalpaiguri district predominantly the aspect angle is in south-west (29.23%) and south-east (28.46%) direction. Very few tea gardens (3 number) in Jalpaiguri district have slope direction in north, north-east and north-west.

 

Correlation study was carried out (Table 6) between shade density and terrain parameters to know whether terrain has any control over shade density. Significant negative correlation was observed with elevation and slope, especially at low crown density level. It reveals that with increase in slope and/or elevation the shade density decreases to a minimum. However, the aspect direction failed to give any correlation except at high crown densities as most of the gardens (76.81%) are oriented between east and west direction only.

 

 

Table 6 Correlation between shade tree density and the terrain parameters (n=220)

Note: *: Significant at 0.001 level

 

4 Discussion

In Cartosat-1, image segmentation technique was utilized to capture the crown density of the shade trees of Darjeeling and Jalpaiguri districts of West Bengal. The spatial variability of shade density differed extensively in both the districts and across the gardens. Significant positive correlations (r2 =0.91) was found between observed and measured crown density. Based upon the modelled crown density the gardens were classified as low, medium, high and very high shade density and presented as a spatial map. The effect of terrain properties viz. elevation, slope and aspect were also analyzed for its relations with shade density. It was observed that except for aspect both elevation and slope have strong correlation with shade density. Significant decrease in the shade density occurs with an increase in elevation and/or present slope. As most of the tea growing areas of the gardens was aligned in an east-south-west direction, no significant correlation was observed between aspect angle and shade density. The limitation of the study is that the segmentation technique depends upon the image radiometry, contrast and soil background. The method gives the best result for perennial shade trees but may not be suitable for annual shade trees. Besides shade tree density is dynamic and changes with garden management practices. Hence, while presenting the shade density the age of the plantation, slope, aspect and management practices also need to be kept in mind.

 

Acknowledgements

The authors express their sincere thanks to Dr V. Jayaraman and Dr V.K. Dadhwal, Former Directors, NRSC for their kind consent for carrying out the study. The time to time guidance rendered by Dr A. Jeyaram, Former General Manager and Project Director of RRSC-East is duly acknowledged. Thanks are also due to Dr K.K. Sarma, Shri T.P. Girish, Shri Sanjoy Choudhury and all the RRSC-East scientists for their valuable suggestions.

 

Authors’ contributions

Dibyendu Dutta – conceptualization, satellite data analysis and documentation. Libeesh Lukose – satellite data analysis, documentation, GIS database creation. Anju Bajpai – coding, customized software development and analysis. Uttam Bhunia – satellite data analysis, GIS layer creation, attribution. Rajkumar Singh - satellite data analysis, GIS layer creation, attribution. Sourav Samanta – Data analysis, GIS layer creation and mapping. All authors read and approved the final manuscript.

 

Reference

Brunig E.F., and Sander N., 1983, Ecosystem structure and functioning, some interactions of relevance to agroforestry, Proc. consultative meeting, 8-15 April, Nairobi, Kenya, International Council for research in Agroforestry

 

Carr M.K.V., and Stephen W., 1992, Climate, weather and the yield of tea, Tea-cultivation to consumption, Chapman and Hall, London, pp.87-135

 

Chowdhury A., Mondal S., and Chowdhury M., 2015, Inventory of shade trees in tea gardens of Sub-Himalayan region of West Bengal, India. Int. J. Sci. Tech., 3(12): 164-168

 

Hadfield W., 1974, Shade in north-east Indian tea plantations, I. The shade pattern, J. Appl. Ecol., 11: 151-178

https://doi.org/10.2307/2402012

 

Huxley P.A., 1983, The role of trees in agroforestry, Plant research in Agroforestry, Proc. consultative meeting, 8-15 April, Nairobi, Kenya, International Council for research in Agroforestry

 

Majumder B.A., Bera B., and Rajan A., 2010, Tea statistics: global scenario, Int. J. Tea Sci., 8(1): 121-124

 

Mohotti A.J., 2004, Shade in Tea: is it beneficial? Sri Lanka J. Tea Sci., 69: 27-39

 

Mukherjee A., Banerjee B., Nanda M.K., and Sarkar S., 2008, Microclimate study under agroforestry system and its impact on performance of tea, J. Agrometeorology, (Special issue – Part I): 99-105

 

Obaga S.O., 1984, Shade trees in tea: a review, Tea, 5: 39-47

 

Tscharntke T., Clough Y., Bhagwat S.A., Buchori D., Faust H., Hertel D., Holscher D., Juhrbandt J., Kessler M., Perfecto I., Scherber C., Schroth G., Veldkamp E., and Wanger T.C., 2011, Multifunctional shade-tree management in tropical agroforestry landscapes – a review, J. Applied Ecol., 48: 619-629

https://doi.org/10.1111/j.1365-2664.2010.01939.x

 

Visser T. 1961, Interplanting in tea 1, Effects of shade trees, weeds and bush crops, Tea Quart., 32: 69-82

 

Journal of Tea Science Research
• Volume 8
View Options
. PDF(0KB)
. FPDF(win)
. HTML
. Online fPDF
Associated material
. Readers' comments
Other articles by authors
. Dibyendu Dutta
. Libeesh Lukose
. Anju Bajpai
. Uttam Bhunia
. Rajkumar Singh
. Sourav Samanta
Related articles
. Tea
. Cartosat-1
. Shade tree
. Crown density
. Image segmentation
. Terrain geometry
Tools
. Email to a friend
. Post a comment