Rishiraj Dutta, Dr. N.R. Patel
Indian Institute of Remote Sensing, Dehradun
Prof. Dr. Ir. Alfred Stein
International Institute for Geoinformation Science & Earth Observation
Tea is one of the most important beverage in India. It is the number one foreign exchange earner. India is the largest producer of tea in the world. The Indian states of Assam, Meghalaya, Tripura, North Bengal (Darjeeling) and Sikkim contribute significantly to the overall tea production in the country. Apart from those, South Indian states of Tamil Nadu, Karnataka and Kerala also contribute to the production of tea.
Over a past few years, it was found that the tea industry is loosing it’s ground. This is mainly because of wrong production mix, inability to compete with other tea producing countries due to high cost of production, organization of small holder farmers, poor quality control at the processing level and more significantly from pests and disease infestations.
Remote sensing and GIS technologies have been efficiently used for monitoring several annual crops like rice, wheat, etc. Therefore, developing an approach for monitoring tea plantations using remote sensing and GIS has become a pressing need. The lack of previous studies in monitoring tea using remote sensing provided the idea to develop an approach that can aid in monitoring the growth of plantations and help in taking effective measures when the need arises.
To test whether MODIS derived NDVI is related to LAI, an empirical equation was established which shows that LAI in tea had significant and linear relationship with NDVI (R2=0.36). This study showed that MODIS based NDVI during April, June and August was significantly correlated to tea leaf yield at estate level. However it was found that NDVI observation at different time period alone could not explained much variance in tea leaf yield. This shows that statistical model for tea yield does not seem to be encouraging.
Tea is indigenous to India and is an area where the country can take a lot of pride. This is mainly because of its pre-eminence as a foreign exchange earner and its contributions to the country’s GNP. In all aspects of tea production, consumption and export, India has emerged to be the world leader, mainly because it accounts for 31% of global production. It is perhaps the only industry where India has retained its leadership over the last 150 years. Tea production in India has a very interesting history to it. The range of tea offered by India – from the original Orthodox to CTC and Green Tea, from the aroma and flavour of Darjeeling Tea to the strong Assam and Nilgiri Tea- remains unparalleled in the world. While the tea industry in Assam grew, but with the passage of time it also started facing many problems. The present crisis in the tea industry started in 1999, when unprecedented drought during the early part of the season led to drastic production cuts. Year 2000 saw the marginal improvement in production but there was a sharp drop in price realization. Production in Assam in 2001 was low as compared to the national average. During the year, prices further declined. Export also dropped by 27 million Kgs and Assam could export only 18 million Kgs of tea. It is believed that Assam is losing exports due to wrong production mix, inability to compete with the other tea producing countries due to high cost of production, old age of tea plants, organization of small holder farmers, poor quality control at the processing level and more significantly from pests and disease infestations. The quality of Assam tea has also deteriorated in the past couple of years as planters are paying more stress on quantity over quality. Most of the small holder producers faced the problem in tea sector. There is a falling share of producer price and poor infrastructure for the small growers.
- To examine relations between tea leaf yield and remotely sensed NDVI
Data: MODIS surface reflectance (250m) at 8 day interval from 2000-2005 for the month of April, June and August. The data was downloaded from Earth Observation Science (EOS) Gateway site. The images were obtained in the UTM projection which were further reprojected to polyconic projection.
Study Area: Sonitpur district is spread over an area of 5324 sq. Kms. on north bank of Brahmaputra river. In terms of area Sonitpur is the second largest district of Assam after Karbi Anglong district.
- North: The state of Arunachal Pradesh.
- South: Morigaon, Nagaon, Jorhat and Golaghat districts.
- East: Lakhimpur District.
- West: Darrang District.
(Pachnai river serves as the boundary)
The District lies between 26° 30’N and 27° 01’N latitude and between 92° 16’E and 93° 43’E longitude. Located between mighty Brahmaputra River and Himalayan foothills of Arunachal Pradesh, the district is largely plain with some hills. Brahmaputra River forms the south boundary of the district. A number of rivers which originate in the Himalayan foothills flow southwards and ultimately fall in Brahmaputra River.
Data Collection: Data used for this study can be categorized in the following two types: Earth Observation Data and Field Survey Data.
Field Data: In this study both attribute and spatial data were considered. The attribute data was collected from tea estate records, meteorological records and by measuring field-by-field Leaf Area Index. The spatial data is extracted from satellite images and existing maps. These spatial and attribute data were linked within a GIS database. All existing maps of the gardens were collected like digital coverages of field boundaries, landuse, soil boundaries, road and stream network, slope, elevation and aspects. Further the leaf area index were collected from the gardens using plant canopy analyzer.
Satellite Data: The MODIS images were reprojected to polyconic projection. The MODIS NDVI images were generated.
5.1 Yield Estimation:
For the yield estimation the MODIS LAI image for August and MODIS NDVI image for April, June and August from 2000 – 2004 were used. 1 x 1 and 3 x 3 kernel pixel extraction method was used to extract the NDVI and LAI values from the NDVI and LAI images. Finally the mean NDVI for all the MODIS images were extracted using the area weighted average or zonal attribute.
5.2 LAI and NDVI Relationship:
Similarly, MODIS NDVI image was generated and the tea garden patches were masked out. Using the masked NDVI image, the NDVI values were extracted from the tea masked area using 3 x 3 kernel for extracting the pixels. The average of the actual LAI and NDVI values were calculated and the linear regression analysis was carried out using the LAI and NDVI values. The LAI – NDVI values were plotted and linearly regressed. From the analysis it is observed that there exist a linear relationship between LAI and MODIS based NDVI. The relationship was quite weak but yet significant with moderate R2=0.40 value. Thus it could be inferred that MODIS derived NDVI can approximately provide information on leaf area index for tea. The relationship is shown below.
5.3 Relation Between Tea Leaf Yield and MODIS NDVI:
Correlation analysis was carried out between area weighted averaged NDVI of tea for selected tea estate with their tea leaf yield for different years (2000-2004). The correlation coefficient ‘r’ values between yield and NDVI at critical time periods are shown in the table. Results showed that correlation is positive and significant irrespective of month of the NDVI over the years. During 2000 and 2001, tea leaf yield was found significantly related to NDVI at 95% level of significance while during 2003, correlation is positive at 1% level of significance.
*Significant at 0.05% level & **Significant at 0.01% level
Year wise correlation between tea leaf yield and MODIS based NDVI of tea estates during different months
Variations could be well observed from the table. There were reasons for such variations. During the year 2000, there was outbreak of disease infestation due to heavy showers accompanied by high temperature and high humidity. This condition is highly favourable for Red Spider and Helopeltis attack. Apart from this, the ground LAI collected from the field showed lot of variations due to continuous plucking, infestations and also due to un-plucked areas. The different species also showed variations in the LAI readings. But still a significant correlation could be seen between tea leaf yield and MODIS NDVI of tea estates during different months.
5.3.1 Tea Yield Model
Linear relationships between tea leaf yield and MODIS NDVI were developed. Here five years yield from 2000 – 2004 as well as their corresponding NDVI were used for the yield model. The observations were linearly regressed for three different months as shown in the Table 10. It was observed that there is a significant positive correlation between yield and NDVI. The variance of 0.243 and 0.292 indicate that there is greater variability in the yield during different periods. Though the R2 value of the coefficient of determination was found to be less the relationships were significant.
|1||1106.06 + 1472.55 * NDVIJune||88||0.243||287.09||27.56|
|2||797.29 + 1208.08 * NDVIJune + 890.43 * NDVIAug||88||0.292||279.24||17.52|
Tea Leaf Yield Models Based on MODIS NDVI
The significance (probability) of the F value was set at 0.05 level. The entry probability of F value taken is less than 0.05 for significant result. The F probability result shows for all linear regression equation are less than significance F change. There is a close relationship between yield and NDVI. Therefore the Null Hypothesis1 (H0) which shows that there is a close relationship between yield, LAI and NDVI used for finding the yield variability is accepted while the alternative hypothesis (Ha) is rejected.
- To test whether MODIS derived NDVI is related to LAI, an empirical equation was established.
- Equation shows that LAI in tea had significant and linear relationship with NDVI (R2=0.36).
- NDVI observation at different time period alone could not explained much variance in tea leaf yield.
- Statistical model for tea yield does not seem to be encouraging.
- The performance of the model would have been much better if the weather parameters for the entire state would have been taken into consideration.
- An improved statistical model for tea yield needs to be developed.
- Azhar, H. S. (1993). Vegetation Studies Using Remote Sensing Techniques. Malayasia, Universiti Teknologi.
- Blakemore, M., Ed. (1984). Principles of Geographical Information Systems for Land Resources Assessment, Oxford Science Publication, Clarendon Press, Oxford.
- Bordoloi, P. K. B., B.C. (1994). A Study on The Landuse Pattern And Ground Water Potential Through Remote Sensing Technique. Jorhat, Toklai Experimental Station, Tea Research Association.
- Brisco, B., Brown, R.J., Hirose, T., and Staenz, K. (1998). “Precision Agriculture and Role of Remote Sensing: A Review.” Canadian Journal of Remote Sensing Vol. 24, No.3: 315-327. •
- Cohen, W., B., Maiersperger, Thomas K., Yang,Zhiqiang, Gower, Stith T., Turner, David P., Ritts, William D., Berterretche, Mercedes & Running, Steven W. (6th June, 2003). “Comparisons of land cover and LAI estimates derived from ETM+ and MODIS for four sites in North America: a quality assessment of 2000/2001 provisional MODIS products.” Remote Sensing of The Environment 88(2003) 233-255.
- Dadhwal, V. K. “Crop Growth and Productivity Monitoring and Simulation Using Remote Sensing and GIS.” Satellite Remote Sensing and GIS Applications in Agricultural Meteorology: pp. 263-289.
- Dadhwall, V. K. a. R., S.S. (2000). “Crop Assessment Using Remote Sensing-Part II: Crop Condition and Yield Assessment.” Indian Journal of Agricultural Economics, 2(1947)-4.
- Ghosh, A. (2003). “GIS Anchored Integrated Plantation Management: Tea.”
- Gonzalez, R. C. a. W., R.E. (1993). Digital Image Processing.
- Groten, S. M. E. (1993). “NDVI-Crop Monitoring and Early Yield Assessment of Burkina Faso.” International Journal of Remote Sensing 14(8), 1495-1515.
- Guissard, V., Lucau-Danila, Cozmin & Defournya, Pierre (2005). “Crop Specific LAI Retrieval Using Optical and Radar Satellite Data for Regional Crop Growth Monitoring and Modelling.” Society of Photo-Optical Instrumentation Engineers: 1-12.
- J.G.P.W., C. (1986). Application of Remote Sensing to Agricultural Field Trials, Agricultural University, Wageningen: 86-4.
- Jones, B. P. (15 February, 2002). Determination and Manipulation of Leaf Area Index to Facilitate Site-Specific Management of Double-Crop Soybean in the Mid-Atlantic, U.S.A. Faculty of the Virginia Polytechnic Institute and State University: pp: 1-169.
- Lee, K. S., Park, Y.I., Kim, S.H., Park, J.S., Woo, C.S. & Jang, K.C. “Remote Sensing Estimation of Forest LAI in Close Canopy Situation.”
- Lillesand, T. M. R. W. K. (2000). Remote Sensing and Image Interpretation. New York, Chichester, John Willy & Sons.
- Lydia Serrano (1), J. A. G., 2) and J. Berry (3), “Estimation of leaf area with an integrating sphere, Tree Physiology.” 17:571to576.
- Martin, F. N. “Correlation Between Remote Sensing Data and Strawberry Growth and Yield.” pp: 1-4.
- Megown, R. A., Webster, Mike and Jacobs, Shayne “Using Landsat TM Imagery to Estimate LAI in a Eucalyptus Plantation.”
- Patel, N. R., Bhattacharjee, B., Mohammad, A.J., Tanupriya, B & Saha, S.K. (In the Press). “Remote Sensing of Regional Yield Assessment of Wheat in Haryana, India.” International Journal of Remote Sensing.
- Rajapakse, R. M. S. S., Tripathy, Nitin, K. and Honda, Kiyoshi (2001). “Modelling Tea (Camellia(L)O.Kuntze) Yield Using Satellite Derived LAI, Landuse and Meteorological Data.”
- Rajapakse, R. M. S. S., Tripathy, Nitin, K. and Honda, Kiyoshi (2002). “Spectral Characterization and LAI Modellng For Tea (Camellia sinensis (L) Kuntze) Canopy.” International Journal of Remote Sensing Vol. 23, No. 18, 3569-3577.
- Schlerf, M., Atzberger,Clement, Vohland, Michael , Buddenbaum, Henning , Seeling, Stephan and Hill, Joachim (March, 2003). “Derivation Of Forest Leaf Area Index From Multi- and Hyperspectral Remote Sensing Data.”
- Xavier, A. C. V., Carlos Alberto (May/June, 2004). “Monitoring Leaf Area Index at Watershed Level Through NDVI From LANDSAT-7/ETM+ Data.” Sci. Agric. (Piracicaba, Braz.), v.61, n.3, p.243-252.