Objective Sampling Estimation of Crop Area Based on Remote Sensing Images

Objective Sampling Estimation of Crop Area Based on Remote Sensing Images

Alfredo Luiz (Embrapa, Brazil), Antonio Formaggio (INPE, Brazil) and José Epiphanio (INPE, Brazil)
DOI: 10.4018/978-1-61692-871-1.ch005
OnDemand PDF Download:
List Price: $37.50


Having the ability to estimate crop areas is a necessity ever more pressing for all the stakeholders of productive chains. For many scientists involved in agricultural research it is also important to know the location of crops and the area they occupy. With this information, considered together with data on soil, climate, availability of infrastructure for storage and transportation, among others, it is possible, for example, to build scenarios and fit models to attend multiple demands. In this chapter the authors propose a simple method combining the techniques of statistical sampling with the characteristics of images obtained by remote sensing, to construct estimates of acreage in a micro-region scale. The objective to be achieved is to produce an estimate of crop area in a defined territory, with estimated statistical error associated.
Chapter Preview


Area estimation is one of the most obvious applications of remote sensing because, among other reasons, it has a direct economic impact and very early found a user community with clearly defined accuracy specifications, at least for applications to agriculture statistics (Gallego, 2004). In his chapter, Gallego (2004) cited an abundant literature that made use of different ways to use satellite images for land cover area estimation. However, the linkage with statistical sampling techniques and field work to check or substitute image interpretation are not presented in a simple way yet.

Context and History

Proper utilization of statistics within any scope depends primarily on how well the nature of employed data is known, and on how clearly the goals are established. Remote sensing offers a rather particular set of data, which almost predetermines the characteristics that must be taken into consideration when choosing the statistical methods to be used in analysis. Application of those data in agriculture, particularly when the aim is to quantify crop areas, will define quite specific goals that should influence the selection of statistical analysis techniques. An alliance between statistics and remote sensing, based on theory and used in proper ways to estimate crop areas, shall result in a step ahead in the efficient use of data from orbital sensors for agriculture purposes. In this direction, this document presents: 1) a method to prepare and use satellite images in agricultural surveys by sampling; 2) the way to calculate estimates over objective data of crop area and their respective variances; 3) a case study consisting in the estimates of soybean area in a municipality, using remote sensing as auxiliary data; and 4) how to use stratification to improve the quality of estimates.

The availability of reliable information about agricultural production is an increasingly fundamental demand in the decision making process. One of the main variables involved in the assessment of agricultural production is the sowed or planted acreage of important crops (Epiphanio et al, 2009). Methods of survey that take into consideration the increasing availability of remote sensing images, and the use of Global Positioning Systems (GPS) and Geographic Information Systems (GIS) may turn out to be the most practical way for a country produce basic data on major farming commodities (FAO, 1996).

The challenge that arises is the establishment of a method which allows associating the technology provided by orbital remote sensing to the procedures used in agricultural statistics surveys.


The purpose of this chapter is to present a simple and reliable method based on the use of remote sensing images and statistical sampling, which allows the quantification of areas occupied by a particular crop in micro-scale regional or municipal level. This method was developed by Luiz (2003).

To achieve this general goal the following specific objectives are established:

  • a.

    Provide a guide for the preparation of images from orbital remote sensing in order to allow their use in agricultural sample surveys;

  • b.

    Establish a procedure to extract a random sample of image’s pixels from a set of pixels spatially delimited;

  • c.

    Present formulas for the calculation of estimated planted area by micro-region or municipality, as well as its variance from data obtained by sampling.

Complete Chapter List

Search this Book: