基于净初级生产力的关中天水经济区耕地利用分区研究

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

耕地净初级生产力(NPP)值越大说明耕地有机生物量越高,其对粮食作物的最终产量起关键作用,因此,研究NPP的变化可为解决粮食安全问题提供重要依据。功能分区是保证土地可持续利用的常用手段,将不同耕地利用模式布局在不同行政区域,可作为地区政府通过管理手段提高产能利用率的有效参照。利用2001—2009年遥感数据,使用遥感估算NPP作为耕地作物生物量代替传统粮食统计数据,通过神经网络算法对关中天水经济区(关天经济区)耕地利用模式进行分区并利用小波神经网络模型对各区的NPP进行预测,得出:(1)关天经济区耕地NPP估算量除2001年外,总量保持在1 600万t,这与粮食统计数据之间的波动规律差异较大,分区中使用NPP数据分区比使用统计数据普适性更强。(2)关天经济区中部和渭河谷地的区县应作为关天经济区农业重点发展地区,率先对其进行精细农业与产业化经营试点的布局。(3)小波神经网络预测显示后6年(2010—2015年)中单位面积NPP上升为主要趋势。该结果证实了在农业作物生物总量短期内难以大幅提高的情况下,在管理上提升农作物能量利用效率是当下缓解耕地安全问题的最有效途径。

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.