each stratum subpopulation in light of one's own spatial
variability of each stratum.
As Appendix: Figure 4 is shown, the total population was
divided into six strata in terms of the Getis ord G;, and a
proportion (percentage) of winter wheat planting area in this
study region obtained by spatial sampling (computation) is
42.98 (%) and its relative error of -1%. Additionally, the result
of integrating the sample point distributions in all the strata
subpopulations by overlay operation shows that the total
sample-point distances are not less than the corresponding
average minimum sample-point distances of each stratum
subpopulation and the previous obtained optimum synthetic
sample-point distances (7.5km~22.5km). This could obtain a
random sample and meet the requirement of probability random
spatial sampling, so the sample point layout didn’t need to be
readjusted and optimized.
Appendix: Figure 5 shows that, based on the Moran’s /;, the
total population was also divided into six strata, and a
proportion (percentage) of winter wheat planting area in this
study region is 50.78 (%) and its relative error of 3.9%.
Nevertheless, the result of integrating the sample point
distributions in all the strata subpopulations by overlay
operation illustrate that parts (which had 14 sample-point pairs
checked as for the six strata) of the total sample-point distances
are less than the above optimum synthetic sample-point
distances (7.5km~22.5km) (See Appendix: Figure 6). Thus they
didn’t meet the requirement of probability random spatial
sampling and the sample point layout needed to be readjusted
and optimized in order to make them not less than the optimal
synthetic distances. Appendix: Figure 7 shows the results of the
readjusted and optimized sample point distribution, and through
computation a proportion (percentage) of winter wheat planting
area in this study region is 50.897(%) and its relative error of
4.1%. Although its relative accuracy was reduced somewhat,
still in an actually acceptable scope had it considerably well
accuracy falling within more than 95%, and even more
importantly, it made the sample-point spatial layout more
reasonable and more geographically representative for
probability random spatial sampling.
Using the Geary’s C;, Appendix: Figure 8 shows that the total
population was divided into four strata, and a proportion
(percentage) of winter wheat planting area in this study region
is 42.05 (%) and its relative error of -14%. As is the same as that
above of the Getis ord G;, its sample point spatial distribution
could meet the requirement of probability random spatial
sampling and didn't thus need to readjust and optimized.
5. DISCUSSION
In order to provide adaptive analysis and reporting according to
report units (such as province, county, township), based upon
spatial structuring a priori knowledge/information, the Kriging
spatial sampling technique(Feng, 2010) can be used to
implement optimal local spatial prediction (i.e., extrapolation)
and infer regionalized element population, whereafter inference
result is gridded on basis of sampling grid basic cells and a data
field map of regionalized elements of spatial sampling of study
areas (that is, called a data field of spatial elements) is thus
obtained. According to practical reporting needs, adaptive
reporting patterns are able easily to perform using the obtained
data fields and selected reporting units, depending on map
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXIX-B7, 2012
XXII ISPRS Congress, 25 August — 01 September 2012, Melbourne, Australia
algebra of GIS (Geographical information systems, such as
ARCGIS and MAPINFO).
Additionally, it is greatly important that remote sensing a priori
knowledge/information (e.g., spatial structuring features)
collaboratively associating with other pertinent helpful (a priori)
knowledge or information (such as the relevant historical data
of, spatial probability distribution mode of sampling objects of
and traffic accessibility of study areas) is used in a spatial
sampling procedure, which can more effectively improve
efficiency, effectiveness and accuracy of spatial sampling. This
is a research direction of our follow-up work.
6. CONCLUSIONS
In this paper, the MODIS remote sensing data, featured with
low-cost, high-timely and moderate/low spatial resolutions, in
the NCP as a study region were firstly used to carry out mixed-
pixel spectral decomposition to extract an useful RIP/RIV from
the initial selected indicators. Then, the RIV values were
spatially analyzed, and the spatial structure characteristics (i.e.,
spatial correlation and variation) of the NCP were achieved,
which were further processed to obtain the scale-fitting, valid a
priori knowledge or information of spatial sampling.
Subsequently, founded upon an idea of rationally integrating
probability-based and model-based sampling techniques and
effectively utilizing the obtained a priori knowledge or information,
the spatial sampling models and design schemes and their
optimization and optimal selection were developed, as is a
scientific basis of improving and optimizing the existing spatial
sampling schemes of large-scale cropland remote sensing
monitoring. In addition, in terms of the adaptive analysis and
decision strategy, the optimal local spatial prediction and
gridded system of extrapolation results were able to excellently
implement an adaptive reporting pattern of spatial sampling in
accordance with report-covering units in order to satisfy the
actual needs of researches or running operation of sampling
surveys. This study can further effectively enhance the level of
spatial sampling survey and decision-making analysis of the
pertinent departments of national and local governments.
ACKOWLEDGEMENTS
This work was supported by the China National Science
Foundation (No. 41171340, 41101390) and China National 863
Project(No. 2006AA 120103).
REFERENCES
Anselin L., 1995. Local Indicators of Spatial Association: LISA.
Geographical Analysis, 27(5), pp. 93-115.
Anselin, L., 1988. Spatial Econometrics: Methods and Models.
Boston: Kluwer Academic Publishers.
Atkinson P. M. and D. R. Emery, 1999a. Exploring the relation
between spatial structure and wavelength: Implications for
sampling reflectance in the field. International Journal of
Remote Sensing, 20, pp. 2663-2670.
Atkinson P. M., 1999b. Geographical information science:
Geostatistics and uncertainty. Progress in Physical Geography,
23, pp. 134-140.
