RICE GRAIN YIELD ESTIMATION OVER SOME ASIAN COUNTRIES USING ISRO’S SCATSAT-1 KU-BAND SCATTEROMETER DATA
Rice crop monitoring and yield prediction at country-scale can be effectively done using high-repeat active microwave remote sensors due to its all-weather observation capability of land surface. The Ku-band with a frequency of 13.515 GHz has the ability to interact with the surface layer and hence is useful for providing information on top portion of canopy and hence the grain and awns of rice crop. Also it has the capability to generate information over the whole region of South Asia in one day. Hence the present study was carried out to explore the super resolved Ku band back scattering coefficient from space borne scatterometer (SCATSAT-1) for rice productivity assessment over six Asian countries (India, Pakistan, Nepal, Bangladesh, Myanmar and Sri Lanka). The super resolved sigma-0 in both polarization (H and V) for the kharif rice season of 2017 (May to Mid-Nov) was used for this study. The temporal backscatter was used to generate rice planting date using polynomial fitting. Multiple regression models were developed using the daily SH/SV ratio and the farm-level fresh paddy yield collected through the Crop Cutting Experiment (CCE). The validation of the model was done for India at state level. For other countries national average reported yield was compared with the estimated yield. The rice planting date was found to vary from first week of June to last week of August in different parts of the six countries. Country average yield was found to vary from 3.45 t ha −1 in Sri Lanka to 4.32 t ha −1 in Myanmar. The absolute difference was lowest in India (8 %) followed by Sri Lanka (−11 %) and maximum in Nepal (35 %). In Indian states, the validation results showed a correlation coefficient of 0.95 at state level with a RMSE of 0.28 t ha −1 (11.4 % of mean reported yield). This study showed the possibility of using high frequency and high resolution Ku-band back scattering coefficient for rice grain yield estimation at continental scale such as Asia. The yield estimation can be further improved with the use of country-wise crop cutting data for model development and validation.