Remote sensing and US crop insurance program integrity: data mining satellite and agricultural data

Abstract

The objective of this investigation is to (1) integrate remote sensing data into an existing data warehouse of the US crop insurance program 1990 to 2007, (2) test remote sensing correlations with crop production, and (3) use remotely sensed time series data to assess variation in crop production. Previously (2000 to 2007) data mining of the data warehouse was based upon probabilistic and algorithmic approaches to identification of possible fraud, waste, or abuse. The value of adding satellite data warehouse to data mining resources is provision of (1) an additional empirical metric, (2) objective data on vegetative health, (3) measurable metrics for capturing the variance of plant health, and (4) a means for measuring the covariance of location and production. Refinement of data mining through the addition of satellite data for routine use to reduce fraud, waste, and abuse will ultimately diminish the frequency of false positives.

The Potential and Uptake of Remote Sensing in Insurance: A Review

“…The system is, like any insurance system, not free of fraud. Utilization of data mining techniques of the remote sensing record to seek for anomalies and suspect cases resulted in a cost reduction of USD 450 million over a six-year period [28]. Remote sensing derived crop classification was used, for example, to verify whether farmers had planted the crop for which losses were claimed [29].…”