Utilization zoning of cultivated land based on net primary productivity in Guanzhong-Tianshui Economic Region

As a populous nation, improve grain production capacity along with rational use and protection of cultivated land resources has posed a considerable challenge in domestic agriculture and land related research in China. Higher NPP for cultivated lands has suggested the existence of more organic biomass. This has been critical for the final production of food crops in the country. It was therefore likely for research on NPP to provide the basis for resolving food security issues. Functional zoning has been the commonly used method to guarantee sustainable use of land. Presently, however, heavily fragmented research merely described real supply of cultivated lands. A deeper understand on the potential reserves of cultivated lands was needed in this regard. Based on remote sensing observation, it is possible to have statistics of the output of a large number of cultivated lands within a short time. Compared with the yearbook data, remote sensing observation has advantages including timeliness and spatial precision. Remote sensing observations have therefore been strongly supplemental to statistical data. NPP estimated by remote sensing was used as crop biomass in cultivated lands instead of the traditional calculations based statistics data. Cultivated land in the Guanzhong-Tianshui Economic Region (GTER) was zoned by using neural network algorithm model and remote sensing data in 2001-2009 substituting for statistic crop yield data. Then the wavelet neural network was used to predict the NPP in the zoned regions. Three results were eventually attained. 1) From 2002 to 2009, total estimated NPP per year in GTER was 1.6×107 t. It showed large variation patterns between estimated NPP data and statistics grain data for cultivated lands in GTER. This suggested statistical and remote sensing data were not substitutable for one another. As clustering function was unknown, zoning via estimated NPP data reflected a more universal adaptability than via statistical data. 2) The final zonal type relatively corresponsed with common cognitions in the study area. It was important to emphasize counties in central GTER and Weihe River Valley (WRV) in the agriculture development of GTER. It was also important for government to set up precision agriculture and agricultural integration in these zones. 3) The prediction calculation by the wavelet neural network showed higher per unit area NPP as the principal trend in 2010 to 2015. Because of the reflected fluctuation patterns varied considerably for different data, it was important to note the differences in data sources and find the driving factors for the reflection of different pressures in cultivated lands. The discussions on data errors suggested that remote sensing data and statistical data should be compared in the study. As rapidly enhancing total crops biomass increase was difficult in the short term, the most effective way of remitting pressure on croplands was to improve use ratio of crop bio-energy.