JIA-2018-09

2104 WANG Di et al. Journal of Integrative Agriculture 2018, 17(9): 2096–2106 The implication was that most sampling units with maize were contained in the second class of ground slope, and therefore, there was a strong spatial autocorrelation within this slope class. For rice, the Moran’s I was lower when the average ground slope of sampling units was higher, except when the sampling size was greater than or equal to 3000 m. This indicated that the spatial autocorrelation of the sampling units can be estimated, when the sampling unit size is greater than or equal to 3 000 m. Moreover, when the ground slope was in the minimum class, the Moran’s I value was the highest. Rice is often planted in low and flat areas because of its need for water. Therefore, the smaller the ground slope, the more intensive the rice planting and, consequently, the greater the Moran’s I value. 4. Discussion Crop acreage information is the crucial basis for formulating national food policies and economic planning. Spatial sampling, the combination of traditional sampling methods and remote sensing and geographic information system technology, provides an effective measure for crop acreage estimation at the regional scale. Traditional sampling requires that the sampling units should be independent each other, but in practice there is often spatial autocorrelation among crop areas included in the sampling units. To improve the rationality of the spatial sampling scheme for crop acreage estimation, this study investigated the spatial autocorrelation among crop acreage included in sampling units of different sizes, and analyzed the effects of different stratification criteria on the level of spatial autocorrelation of crop acreage within the sampling units, based on crop thematic mapping, cultivated land plots and DEM data covering the whole study area. The results indicated that there was strong spatial autocorrelation among maize and rice acreage included in the sampling units, no matter what the sampling unit size is. When the sampling unit size was less than 3 000 m, the stratification design with CPI was selected as the stratification criterion, the stratum number was 5 and the stratum interval was 20%, we found that the Fig. 11 Moran’s I values for cultivated land fragmentation in maize (A) and rice (B) sampling units. Seven solid lines denote different sampling unit sizes. Fig. 12 Moran’s I values for different ground slopes in maize (A) and rice (B) sampling units. Seven lines denote different sampling unit sizes. 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 A B 0.25–0.50 0.50–0.75 0.75–1 .00 1.00–1.50 1.50–9.20 Moran’s I Fragmentation of cultivated land (%) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.25–0.50 0.50–0.75 0.75–1 .00 1.00–1.50 1.50–9.20 Moran’s I Fragmentation of cultivated land (%) 200 m 500 m 1 000 m 2 000 m 3 000 m 4 000 m 5 000 m 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.34–0.87 0.87–1.01 1.01–1.23 1.23–1.82 1.82–11.64 Slope of cultivated land (%) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 A B 0.34–0.87 0.87–1.01 1.01–1.23 1.23–1.82 1.82–11.64 Slope of cultivated land (%) Moran’s I Moran’s I 200 m 500 m 1 000 m 2 000 m 3 000 m 4 000 m 5 000 m