Browsing by Author "Umesh"
Now showing 1 - 3 of 3
Results Per Page
Sort Options
ThesisItem Open Access DEVELOPMENT OF YIELD PREDICTION MODELS FOR OILSEEDS BASED ON WEATHER PARAMETERS IN INDIA(Indira Gandhi Krishi Vishwavidyalaya, Raipur, 2019) SURENDRA, A.R.; Pandey, K.K.; Shukla, S.; Verma, Praveen; UmeshOilseeds cultivation is undertaken across the country in about 26.2 million hectares, of which 72% is confined to rainfed farming. The diverse agro-ecological conditions in the country are favourable for growing nine annual oilseed crops, which include seven edible oilseeds (groundnut, rapeseed & mustard, soybean, sunflower, sesame, safflower and niger) and two non-edible oilseeds (castor and linseed). Any changes in weather parameters might affect the oilseeds yield. So, the crop yield prediction based on weather parameters will help farmers, policy makers and administrators to manage activities. The present study examines the development of yield prediction models for oilseeds based on weather parameters in india. The study was undertaken based on secondary data, the data on yield of total oilseeds in india for 48 years data from 1970-71 to 2017-18 has been collected from www.indiastat.com. Weather data for five weather parameters (Max & Min Temperature, Rainfall, RH1 and RH2) were collected from India Meteorological Department (IMD), Pune and www.indiawaterportal.org. Four models were developed for oilseeds yield namely LASSO Stepwise Regression, Models based on weather indices (MWI), Models using composite weather variables (CWV) and Artificial Neural Networks (ANN). A comparison study of the results obtained from LASSO, MWI, CWVand ANNis also performed. The results were compared using the R2, RMSE statistic and percentage prediction error. The R2& RMSE of the models varied between 0.72&66.66, 0.69 &79.39, 0.74 &62.45 and 0.81 &56.99 respectively. The average percentage prediction error for LASSO, MWI, CWV and ANN models on the test set were found to be 5.99, 7.20, 5.55and 5.18 respectively. The results indicate that the ANN model had a higher R2 value, lower RMSE value and a lower percentage prediction error when compared to other models. The ranking of the model reveals that Artificial Neural Networks (ANN) was the best performing model followed by models using composite weather variables,LASSO Stepwise Regression model and models based on weather indices. The reason behind the good performance of ANN model is that, in a real sense, neural networks are one of the best solutions in search for a few agriculture problems, especially when it comes to predict crop yield. A conclusion has been made that ANN has better explained yield variability rather than other methods. Undeniably, the application of ANN to precision agriculture plays a crucial role in future evaluation of the concept of precision agriculture as a sustainable means of meeting crop yield demands. However, further research about the ANN impacts towards crop yield production must be conducted to ensure sustainability of future needs.ThesisItem Open Access DEVELOPMENT OF YIELD PREDICTION MODELS FOR OILSEEDS BASED ON WEATHER PARAMETERS IN INDIA(Indira Gandhi Krishi Vishwavidyalaya, Raipur, 2019) Surendra, A.R.; Pandey, K.K.; Alivelu, K.; Shukla, S.; Verma, Praveen; UmeshOilseeds cultivation is undertaken across the country in about 26.2 million hectares, of which 72% is confined to rainfed farming. The diverse agro-ecological conditions in the country are favourable for growing nine annual oilseed crops, which include seven edible oilseeds (groundnut, rapeseed & mustard, soybean, sunflower, sesame, safflower and niger) and two non-edible oilseeds (castor and linseed). Any changes in weather parameters might affect the oilseeds yield. So, the crop yield prediction based on weather parameters will help farmers, policy makers and administrators to manage activities. The present study examines the development of yield prediction models for oilseeds based on weather parameters in india. The study was undertaken based on secondary data, the data on yield of total oilseeds in india for 48 years data from 1970-71 to 2017-18 has been collected from www.indiastat.com. Weather data for five weather parameters (Max & Min Temperature, Rainfall, RH1 and RH2) were collected from India Meteorological Department (IMD), Pune and www.indiawaterportal.org. Four models were developed for oilseeds yield namely LASSO Stepwise Regression, Models based on weather indices (MWI), Models using composite weather variables (CWV) and Artificial Neural Networks (ANN). A comparison study of the results obtained from LASSO, MWI, CWVand ANNis also performed. The results were compared using the R2, RMSE statistic and percentage prediction error. The R2& RMSE of the models varied between 0.72&66.66, 0.69 &79.39, 0.74 &62.45 and 0.81 &56.99 respectively. The average percentage prediction error for LASSO, MWI, CWV and ANN models on the test set were found to be 5.99, 7.20, 5.55and 5.18 respectively. The results indicate that the ANN model had a higher R2 value, lower RMSE value and a lower percentage prediction error when compared to other models. The ranking of the model reveals that Artificial Neural Networks (ANN) was the best performing model followed by models using composite weather variables,LASSO Stepwise Regression model and models based on weather indices. The reason behind the good performance of ANN model is that, in a real sense, neural networks are one of the best solutions in search for a few agriculture problems, especially when it comes to predict crop yield. A conclusion has been made that ANN has better explained yield variability rather than other methods. Undeniably, the application of ANN to precision agriculture plays a crucial role in future evaluation of the concept of precision agriculture as a sustainable means of meeting crop yield demands. However, further research about the ANN impacts towards crop yield production must be conducted to ensure sustainability of future needs.ThesisItem Open Access Seasonal incidence of guava fruit fly, Bactrocera spp. and its management through methyl eugenol traps(CCSHAU, 2018) Umesh; Rajesh KumarThe present study entitled “Seasonal incidence of guava fruit fly, Bactrocera spp. and its management through methyl eugenol traps” were carried out during both rainy and winter season at farmer’s field at village Sunderpur, District Rohtak, Haryana during 2016-17.The maggot population of guava fruit fly, Bactrocera spp. started appearing from 29th SMW (4.8 maggots/infested fruit) reaching its maximum in 31st SMW (32.3 maggots/infested fruit). During winter season, the maggot population was at its peak during 44th SMW (12.1maggots/infested fruit) which went on decreasing till 48th SMW (1.12 maggots/infested fruit). There was no maggot population from 49th to 9th SMW. During termination phase of winter crop season, a low maggot population i.e. 2.29 and 3.31 maggots/infested fruit were recorded in 10th and 11th SMW, respectively. The fruit infestation in rainy season guava due to fruit fly ranged from 60 to 90 per cent on number basis while 53.49 to 87.20 per cent on weight basis, being highest during 31st SMW and lowest during 36th SMW both on number and weight basis. During winter season guava, the highest fruit infestation due to fruit fly i.e. 26.67 and 22.43 per cent on number and weight basis, respectively was recorded during 44th SMW. There was no fruit infestation from 51st to 8th SMW. There was 68.53 per cent reduction in yield of guava in rainy season due to guava fruit fly, Bactrocera spp. as compared to 5.49 per cent in winter crop season with an overall yield reduction of 37.54 per cent. The population of fruit fly started appearing in methyl eugenol traps from 23rd SMW (7.2fruit flies/trap/week) reaching its maximum in 33rd SMW (264.2 fruit flies/trap/week) and then population suddenly declined in 37th SMW which coincided with termination of harvesting of guava crop. The population of guava fruit fly, Bactrocera spp. had significantly negative correlation with maximum temperature and significantly positive correlation with morning and evening relative humidity. Various management strategies i.e. S1 (methyl eugenol traps @ 40 traps/ha), S2 (S1 + 3 sprays of NSKE 5%), S3 (S1 + 2 sprays of NSKE 5%), S4 (S1+ 1 spray of NSKE 5%), S5 (S1 + mulching with black polythene sheet), S6 (S1+ raking under tree canopy twice), S7 (S1 + collection and destruction of dropped fruits on alternate days), S8 (three sprays of NSKE 5%) and S9 ( two sprays of NSKE 5%) were evaluated against guava fruit fly during rainy crop season. The overall reduction in fruit infestation over untreated control ranged from 18.08 to 39.97 and 12.43 to 41.45 per cent on number and weight basis, respectively being the highest in S2 and lowest in S9. In S2, there was an increase of 175.33 per cent in marketable yield of guava over control followed by S3 (125.30%). On the basis of increase in marketable yield of guava over control, the efficacy of the various management strategies in descending order was S2, S3, S7, S4, S5, S6, S1, S8 and S9.The highest total marketable yield (95.76q/ha) and net profit (Rs.65674/ha) was obtained from S2.