Agriculture Reference
In-Depth Information
maps, or other geo-referenced information layers. Remote sensing can significantly
contribute to provide a timely and accurate picture of the agricultural sector,
because it is very suitable for gathering information over large areas with a high
A commonly used auxiliary variable for crop area estimates is the Land
Use/Land Cover (LULC) data. LULC refers to data that are a result of classifying
raw satellite data according to the registered values of the satellite image (see
Chap.
4
). LULC have been widely applied to estimate crop areas. Hung and Fuller
(
1987
) estimated crop areas by combining satellite data with data collected using
area surveys. Gonz´lez and Cuevas (
1993
) used a thematic map to estimate crop
areas. These estimates used regression methods. Pradhan (
2001
) presented an
crop forecasting systems at a regional level. The overall system combined spatially
referenced sampling frames and remote sensing.
Remotely sensed data also provide information on different factors that influ-
ence the crop yield. For a comprehensive review of different ways to use remote
sensing for agricultural
statistics
see Gallego (
2004
) and Carfagna and
Gallego (
2005
).
The availability of remotely sensed data does not eliminate the need for ground
data, because satellite data do not always have the required accuracy. However, this
information can be used as auxiliary data to improve the precision of the direct
estimates. In this framework, the calibration estimator can improve the efficiency of
crop acreage and yield estimates for a large geographical area, using classified
satellite images and NDVI (normalized difference vegetation index, see Sect.
4.7
)
as auxiliary information.
The most important application of covariate methods for survey estimation is the
treatment of nonresponses. This topic is a typical and undesirable feature of a
survey, and has always received a great deal of attention in survey literature
(S¨rndal and Lundstr¨m
2005
). The problem is obviously much more relevant in
business and social surveys (a unit may refuse to be interviewed) than in agricul-
tural surveys based on a spatial definition of statistical units. In agricultural surveys,
there may be no interviews and the data collection is based primarily on direct
observations. However, these observations can sometimes be difficult, if not
impossible. Consider, for example, the difficulties of observing a sample point on
the top of a very high mountain or in the interior of a fenced unit.
Using estimation procedures that expand the sample results to the population
using sampling weights, it is clear that we need to use a correction mechanism so
that any nonresponses do not automatically generate a negative bias. Then, we are
interested in adapting the estimation procedures so that they are efficient in the
presence of a considerable nonresponse rate.
The concept of the quality of the estimates is strongly linked with the possibility
that we can measure and control the sampling errors of the design-estimator. Much
of the basic theory of sample surveys concerns variance estimation with a linear
estimator and a non-complex design. The common solutions suggested by the HT or
Search WWH ::
Custom Search