Two phase sampling for wheat acreage estimation

Abstract

A two-phase Landsat-based sample allocation and wheat proportion estimation method was developed. The technique employs manual, Landsat full frame-based wheat or cultivated land proportion estimates from a large number of segments comprising a first sample phase to optimally allocate a small phase-two sample of computer or manually processed segments. Proportion estimates from each phase are then linked by regression or probability proportional to estimated size estimators to provide wheat proportion estimates and standard errors by reporting unit. Application to the Kansas Southwest CRD (Crop Reporting District) for 1974 produced a wheat acreage estimate for that CRD within 2.42% of the USDA SRS-based estimate using a lower CRD inventory budget than for a simulated reference LACIE (Large Area Crop Inventory Experiment) system.