The Key Laboratory of Agro Informatics
The Key Laboratory of Agro Informatics
Tang P.,The Key Laboratory of Agro Informatics |
Tang P.,Chinese Academy of Agricultural Sciences |
Yang P.,Chinese Academy of Agricultural Sciences |
Chen Z.,The Key Laboratory of Agro Informatics |
And 9 more authors.
Nongye Gongcheng Xuebao/Transactions of the Chinese Society of Agricultural Engineering | Year: 2013
The crop's spatial-temporal pattern is critical to many agricultural studies. In the last 30 years, the rice area in Northeast China increased significantly, and has become one of the major rice producing regions in China. To explore the rice spatial-temporal change characteristics and enrich the crop spatial distribution information acquisition methods, this study combined such multi-source datasets as agricultural statistics data, cropland data, irrigation map and crop suitability, and uses a crop spatial distribution model SPAM (Spatial Production Allocation Model) which has been based on the cross-entropy theory and applied successfully in Brazil and Africa. We further modified the model and applied it to China, and named it SPAM-China. This model mainly has four modules that include a data consistency processing module which makes the multisource data a better spatial match, an agricultural statistical data input module, and a multisource data discrimination module that uses GAMS to optimize the procedure and improve the efficiency of the model, and a result output module. With the SPAM-China model, the study simulates the rice spatial distribution of Northeast China on the pixel scale in the past 30 years (1980-2008) and obtains four rice distribution maps. The result shows that the model has a better capacity to simulate rice spatial distribution, and can reflect the main rice cultivation region and temp-spatial change characteristics. In the last 30 years, rice spatial distribution change characteristics were very significant. The rice cultivation area obviously increased, and the cultivation region constantly expanded northeastward. Rice cultivation gravity was northward about 1.76 latitude (140 km) in the nearest 30 years, but the rice cultivation gravity of 2008 is southward about 0.23 latitude (16km) than that of 2000. The rice area rose sharply and has a distinct upward trend in the middle and north of this area, where the north region has the most significant increasing trend. Sanjiang plain and Songnen plain are the main factors for rice increasing in the north region. However, in the south of the region, the rice area change shows no significant trend. On the pixel scale, the rice area of the most pixels has a significant increasing trend and only a few has a decreasing trend in the south of Northeast China. The increasing pixels mainly lie in the middle of Jilin province, south and east of Heilongjiang province. Compared with the remote sensing interpretation crop pattern of Northeast China based on the moderate-resolution satellite MODIS in 2009, this model's simulation results are validated and prove that the SPAM-China model has a better capacity to simulate the rice spatial distribution of Northeast China on the pixel scale. On the pixel scale, the spatial distribution consistency pixels mainly lie on the traditional rice cultivation region and Sanjiang Plain. Although there is some discrepancy in a few regions, these pixels have a good spatial distribution consistency occupying a greater advantage in pixel amount and rice area. It accounts for 77.20% of the total rice area and 59.57% of the whole rice pixels for SPAM-China and accounts for 91.69% of the rice area and 73.09% of the whole rice pixels for the remote sensing interpretation result. Although the SPAM-China model can better simulate the crop spatial distribution, it must consider the multi-source data consistency, data update, statistics data accuracy, and agricultural and geographical characteristics of study region because these factors can reflect, to a large extent, the simulation accuracy. At the same time, the simulation spatial resolution promotion is also the main direction to improve the model.