ANALYSIS OF REMOTE SENSING-BASED ASSESSMENT OF POTATO STATISTICS AND ITS COMPARISON WITH GOVERNMENT ESTIMATES
Keywords: Remote sensing, Potato, Area, Production, Yield, Accuracy assessment, FASAL, CHAMAN
Abstract. Potato (Solanum tuberosum L.) is a major horticultural crop of India. In the present study an effort has been made to evaluate the forecast (area, production and productivity) of potato being carried out under FASAL/CHAMAN project of DAC&FW. For this purpose, remote sensing-based forecasts were analysed for the period of 6 years (2012–13 to 2017–18). The district level area and production estimates were carried out in 5 major potato growing states of India i.e. Bihar, Gujarat, Punjab, Uttar Pradesh and West Bengal. The area estimation has been carried out using multi-spectral satellite data through supervised/ unsupervised image classification techniques. IRS P6-LISS III (, Sentinel- 2A, and LANDSAT-8 OLI data were used for acreage estimation during October to March every year. District level potato crop yield has been estimated using two different procedures - i) Agro-meteorological stepwise regression models, and ii) Remote sensing index (VCI) based empirical models. The estimates of area, production and productivity were compared with the Government (DES) estimates. Good correlation (r) between the forecasted and DES estimates was observed in case of area (0.37, 0.81, 0.93, 0.68 and 0.73) and production (0.59, 0.72, 0.94, 0.30 and 0.09) for Bihar, Gujarat, Punjab, Uttar Pradesh and West Bengal states, respectively. Similarly, low RMSE (%) between the forecasted and DES estimates was observed in case of area (6.90, 14.6, 16.8, 5.4 and 3.8) and but higher RMSE (%), in case of production (18.6, 21.3, 21.8, 9.8 and 20.9) for Bihar, Gujarat, Punjab, Uttar Pradesh and West Bengal states, respectively. Year wise correlation and RMSE% reveals that the accuracy of remote sensing-based area and production has increased significantly over the last few years may be due to improvement in methodology and availability of higher resolution data along with experience gained over the crop.