Varietal discrimination of Basmati rice in northwest

Size: px

Start display at page:

Download "Varietal discrimination of Basmati rice in northwest"

Brett Long
5 years ago
Views:

1 Varietal discrimination of Basmati rice in northwest India A. N. Singh*, Dharmesh Verma** and M. H. Kalubarme *Global Institute of Land, Water and Environment Management, Lucknow, India ** United Phosphorous Ltd., India *** BISAG, Gandhinagar, India

2 What is Basmati Rice? Basmati is a premium long grained aromatic rice grown in a specific geoenvironment, i.e. NW India and parts of Pakistan for centuries. Documentary evidences show that Basmati has been grown in this area for more than 250 years (Nene, 2003). Its high value stems from its unique eating qualities, which includes aroma in both the raw and cooked state, kernel length 7 mm or more, excellent linear elongation on cooking almost double its kernel length, soft and flaky consistency of cooked rice. Basmati 370, Taraori Basmati, Type3 and Ranbir Basmati are the Traditional Basmati varieties grown in Punjab, Haryana, western U.P., Uttarakhand and J&K. These are tall ( cm plant ht.), d growing period, kernel length varying from 6.9 to 7.3 mm, breadth 1.7 to 1.9 mm, kernel elongation ratio after cooking 1.8 to2.1.

3 Evolved Basmati varieties Pusa Basmati-1, Pusa-1121, CSR-30 and Pusa-1509 are the varieties evolved (recently bred) Basmati having traditional Basmati varieties as one of the parent. These varieties are being grown in a larger area now due to higher yield. Factors favouring aroma in Basmati are cool temp. during flowering and grain development (25 degree C/ 31 degree C night/day temp. during crop maturity), use of organic manures, fertile, light-textured and well-drained soil, direct sowing, etc.

4 Objectives of study The study, commissioned by the Agricultural Products Export Development Agency (APEDA), Govt. of India, had the following objectives: Remote Sensing data based area estimate of traditional and evolved Basmati varieties in Punjab (21 dist.), Haryana (20), western Uttar Pradesh (26), Uttarakhand (4), Jammu & Kashmir (2), and Himachal Pradesh (2). CCE based yield prediction in different districts/regions. Monthly report on Basmati during its cropping season growth, crop condition, biotic and abiotic stresses Annual change in area under basmati varieties.

5 Study Area

6 IRS-P6 AWiFS : 24-SEP-2008 Covering Haryana and Punjab States

7 IRS-P6 AWiFS : 03-Oct-2010 Covering Haryana and Punjab States

REMOTE SENSING OF CROP PHENOLOGY An attempt was made to differentiate High yielding rice and basmati varieties based on their phonological stage differences which in turn have impact on spectral

8 REMOTE SENSING OF CROP PHENOLOGY An attempt was made to differentiate High yielding rice and basmati varieties based on their phonological stage differences which in turn have impact on spectral reflectance on the satellite data. In general, the high yielding rice varieties are sown and transplanted one month in advance of Basmati varieties in most of the study areas. The most important reason for late sowing and transplantation of basmati varieties is that these varieties should mature during the cold night periods during late October or first week of November which helps to produce better aroma in the basmati. Basmati Sharbati

9 Seed to seed duration of commonly grown Basmati varieties Traditional Basmati Basmati-386 / Taraori Basmati Basmati-370 Evolved Basmati CSR-30 Pusa Basmati-1 Pusa d 145 d 140 d 140 d d

10 Methodology Data Used Multi-temporal IRS AWiFS (56m), LISS-III (23m) digital data for the rice growing period and Liss-IV (6m) for selected area. Due to cloud cover, good quality data available were generally after 15 Sept.. AWiFS with 5 d repeat cycle provided more frequent Field Data and GPS Measurements Ground Truth (GT) was collected during fourth week of August to first week of October, which coincided with flowering to grain formation stage of rice crop. Agronomic data like variety, stage/vigor, and height of the crop canopy, soil exposure were recorded. Minimum size of plot considered was 300 * 300 sq. m. to collect data using GPS. Crop Cutting Experiments (CCE) for yield estimation in all the states. For example, in Haryana, CCE were conducted in 190 plots covering 10 districts.

11 Selection of images for varietal study based on crop calender In Punjab, Traditional Basmati varieties (Basmati 386) is in flowering stage in last week of October and harvested in 4 th week November. Evolved Basmati (Pusa Basmati-1 and PB1121) flowers in 2 nd week of October and harvested in 3 rd week of November. HYVs are harvested in 4 th week of September. In Haryana, transplanting of Basmati varieties is done about 15 days earlier than Punjab, and accordingly all crop stages. In Uttar Pradesh, transplanting is similar to Punjab. In Jammu & Kashmir, only Ranbir Basmati is grown, which is of shorter duration. Hence, images of last week of September to 1 st week of October and onwards were selected for analysis.

12 Steps in IRS LISS-III Digital Data Analysis Geo-referencing Administrative boundary superimposition Generation of spatial information in GIS environment Superimposing GPS locations of Basmati and high yielding rice varieties on the registered LISS-III digital data, Identification of basmati and high yielding rice varieties on LISS-III digital data, Supervised classification using MXL classifier with boundary mask approach, Area estimation under different rice varieties Generation of spectral vegetation indices like NDVI

13 DN to Radiance Conversion Calculation of at-sensor spectral radiance is the fundamental step in converting image data from multiple sensors and platforms into a physically meaningful common radiometric scale. In order to obtain radiometrically comparable apparent spectral radiance data suitable for further processing, the integer digital number (DN) of each band of all images was transferred into real numbers using the spectral calibration data. The calibration was done by following expression of satellite spectral radiance Lλ, (Lillesand et.al; 2000) which is, Lrad = {[DN/MAX GRAY] * [Lmax - Lmin]} + Lmin Where, DN = Digital numbers of a pixel, Max grey: Maximum DN possible for a given data. Lmax and Lmin are the maximum and minimum radiance values for band (mwcm -2 Sr -1 µm 1 ).

14 Generation of Training Signatures and Separability Analysis LISS-III and LISS-IV images of selected growth phases of major HYVs and Basmati varieties ( Last week of September onwards)were used. GPS based training sites were collected for different rice varieties and other land-use classes. Five-to-six classes with different developmental stages and percent ground cover having different vigour for each rice variety were identified for training signature generation. The training signatures contain multi-band statistics such as mean, standard deviation, and variance-covariance matrix for each class, which is used in supervised classification. Spectral separability of basmati rice varieties and other HYVs were generated. Before using these signature statistics in the supervised classification, the crop separability was studied by computing the Transformed Divergence for different classes.

15 Spectral Reflecatnce of Rice Varieties IRS LISS-III Spectral Bands Band 2-Green: Band 3-Red: Band 3-NIR : DN Values Basmati-1 Basmati-2 Basmati-3 HYV-1 HYV-2 Sharbati-1 Sharbati Central Wavelength (micro meter)

16 IRS-P6 LISS-III images of two dates showing differentiation of Basmati from HYVs in part of Karnal district, Haryana 31-Aug SEP-2008 Basmati High Yielding Varieties

17 Sharbati 31-August September 2008 Traditional Basmati Evolved Basmati South-Western Part of Karnal District

18 GPS points for Training sites & CCE

19 Confusion Matrix Percent Pixels Classified by Code Class Code No. Pixels River sand Waste Land Habitation Water Water Basmati Basmati Sharbati Sharbati HYV HYV Fallow land waterlogged Other Veg Average accuracy = % Overall accuracy = % KAPPA COEFFICIENT =

20 Signature Separability using Transformed Divergence Separability Measure: Transformed Divergence Average Separability: Minimum Separability: Maximum Separability: Minimum Separability: Sharbati-1 and Sharbati-2 Class River Sand Waste Land Habitation Water-1 Water-2 Basmati-1 Basmati-2 Sharbati-1 Sharbati-2 HYV-1 HYV-2 Fallow Land Waterlogged Stream Waste Land Habitation Water Water Basmati Basmati Sharbati Sharbati HYV HYV Fallow Land Waterlogged Stream Other Veg

22 NDVI Image of IRS LISS-III data of 24-Sep-2008 covering Karnal Kaithal & Kurukshetra KURUKSHETRA KAITHAL KARNAL

23 Basmati and HYV variety Classification using NDVI Thresholding

24 Crop Yield Crop yield data collected from CCE and Agriculture department from high yielding and basmati growing states in India, along with agrometeorological data and Spectral Vegetation Index like Normalized Difference Vegetation Index (NDVI) was analyzed for developing zonal Agromet-Spectral Yield models using multiple regression analysis.

25 Relationship between CCE Yield and NDVI values of Basmati Variety in Karnal & Panipat Districts Yield (q/ha) Pusa Basmati: NDVI Vs. Yield (Panipat, Haryana) YIELD (q/ha) NDVI Linear (YIELD (q/ha)) Yield = 25.27*NDVI R 2 = 0.95 Yield (q/ha) Pusa Basmati: NDVI Vs. Yield (Karnal, Haryana) Yield (q/ha) NDVI Linear (Yield (q/ha)) Yield= 45.15*NDVI R 2 = 0.91

26 Relationship between CCE Yield and NDVI values of Sharbati & Basmati Varieties Yield (q/ha) Sharbati Rice: NDVI Vs. Yield (Karnal, Haryana) Yield (q/ha) NDVI Linear (Yield (q/ha)) Yield = 45.66*NDVI R 2 = 0.84 Yield (q/ha) Pusa-1121: NDVI Vs. Yield (Muzaffarnagar, UP) NDVI Yield (q/ha) Linear (Yield (q/ha)) Yield = 16.15*NDVI R 2 = 0.67

Crop Cutting Experiments (CCE) for yield estimation Crop Cutting Experiments were conducted using standard procedures in the study area for assessment of yield.

27 Crop Cutting Experiments (CCE) for yield estimation Crop Cutting Experiments were conducted using standard procedures in the study area for assessment of yield. The CCE derived yield was averaged for the district and a conversion factor used for offsetting the moisture content of the grain for estimation of district level production. In Haryana State, Crop Cutting Experiments were conducted in 190 plots covering 10 districts. Based on CCE data, the range of productivity of different Basmati varieties computed is given in Table-1.

28 Agro-met-spectral Yield Models Normalized Difference Vegetation Index (NDVI) of Basmati and high yielding rice varieties of a particular administrative district / tehsil for 10 crop seasons were generated Meteorological data like rainfall, Tmax, Tmin, Relative Humidity (RH %), sunshine Hours etc. of previous 10-years have been collected form IMD for a particular Met Station. The Basmati yield data at district/tehsil-level was also collected from the Department of Agriculture of the same periods. Agro-meteorological yield models were generated using this data set and using the current seasons met data. These Agro-meteorological yield models can be used for predicting the basmati yields well in advance of the harvesting period. The crop condition term was also be incorporated into the yield models to take into account the yield reduction due yield reducing factors.

29 An attempt was made to predict the Basmati rice yields using the Agro-meteorogical yield model. The observed and model predicted Evolved basmati yields for Karnal, Panipat, Kurukshetra districts are given in the following figure RiceYields inkarnal District 3200 RICE YIELD (kg/ha) OBSERVED PREDICTED YEAR

30 State-wise area under Basmati and other varieties S. No. State Total Rice Basmati-386, 370, Type-3 Area in '000 ha Evolved Basmati Non-Notified Pusa-1121 PB-1 CSR-30 Sharbati Sugandha 1 Punjab Haryana Uttar Pradesh Uttarakhand Jammu & Kashmir Himachal Pradesh Delhi 1.50 Total

32 10000 Aromatic Indica Basmati Rice Acreage, Production and Export During Last One Decade Aromatic Indica Basmati Rice Acreage, Production and Export During Last One Decade AREA (000, HA) AREA (000, HA) YEAR Area 2010 Production 2011 Export (000, tons) Value (million USD) YEAR Area Production Export (000, tons) Value (million USD)

33 Conclusion Traditional and Evolved Basmati varieties were separable by proper selection of data base on phenology and analysis of IRS Liss-3 data. AWiFS data with 5d repeat cycle used in conjuction with Liss-3 proved helpful wherever data gap existed. Within the two groups of Basmati, ground survey based fraction of diff. varieties was used to arrive at percent area under different varieties. Crop cutting based yield was used to calculate the production of different varieties. NDVI and NDWI based yield model was also developed and used to validate the field data. Abiotic stress, like flooding in Punjab in 2010 was also studied using RS data, which helped in flood damage assessment and it`s effect on final yield. Both RS, detailed ground information and expert knowledge are needed to get information on varietal discrimination and production. Accuracy of production data was compared with market arrivals in different states by AIREA and the user organization. The area under Evolved Basmati is increasing due to higher yield and Traditional varieties decreasing in recent years.

34 SALAMAT Thank You

Module 6 Book A: Principles of Contract Design. Agriculture Risk Management Team Agricultural and Rural Development The World Bank

+ Module 6 Book A: Principles of Contract Design Agriculture Risk Management Team Agricultural and Rural Development The World Bank + Module 6 in the Process of Developing Index Insurance Initial Idea