Thumbnail Image

Theses

Browse

20211
Reset filters
Subject: Statistics × Author: Muhammed Jaslam, P K × End date: 2023 × Start date: 2020 × Type: Thesis × Thesis Type: Ph.D ×
Reset filters

Search Results

Now showing 1 - 1 of 1
  • Thumbnail Image
    ThesisItemOpen Access
    Small Area Estimation of Wheat Yield Using Remote Sensing Data in Hisar and Sirsa Districts of Haryana
    (CCSHAU, Hisar, 2021-09) Muhammed Jaslam, P K; Manoj Kumar
    This study focuses on estimating the wheat yield by using the direct and indirect small area estimation techniques at block levels in Hisar and Sirsa districts of Haryana also to develop suitable crop yield models for wheat using satellite spectral data. The simple average of the yield recorded in the villages within the block is the usual estimator for block wheat yield, which is unstable and most of the block level estimates have large CVs. Direct small area estimation techniques such as post stratified and GREG estimation are used to get a precise estimate of wheat yield. Implicit models such as synthetic and composite estimation, as well as explicit models such as unit level and Area level SAE, are used in the indirect small area estimation technique. Furthermore, area level SAE model was developed for a total of 42 blocks in 6 districts of western zone of Haryana. The CV percent value of the block level estimate computed using all small area estimation is lower in comparison to the usual estimate. In the post-stratified direct, synthetic, and composite estimation methods used, the CV values of the composite estimators were found to be less in comparison to post stratified direct and synthetic estimators. In agreement to the basic theory, we obtained good estimation results using the unit level SAE model. Furthermore, using the Robust method of the unit level SAE model to reduce the effect of outliers boosted precision level. This study demonstrated that having a closely related auxiliary variable at the area level (SAE at the area level - Class 3 & 4) can provide a comparable level of precision to a unit level model. Since multicollinearity was detected between the predictor variables for crop yield modelling, we investigated ways in which the simple linear model can be improved by replacing plain least squares fitting with some alternative fitting procedures, such as stepwise regression, ridge regression, LASSO, principal component regression, and partial least square regression. The PLS regression model is found to be the best method (in terms of R2 and RMSE) for predicting block level yields using remote sensing data in western zone of Haryana