基于深度学习模型的种植结构复杂区农作物精细分类研究

目的 卫星遥感技术具有覆盖范围广、探测周期短、调查成本低等优势而广泛应用于大区域农作物分类。然而在种植结构复杂区（如城乡结合部）,因其地块破碎、同期生长的作物种类多且分布分散,利用传统的统计分类或机器学习方法进行农作物分类时仍存在精度不高的问题。为提高种植结构复杂区农作物分类精度。方法 文章选取河北省廊坊市广阳区为研究区,以GF-1 PMS全色多光谱融合影像为数据源,采用U-Net、PSPNet及DeepLabv3+,3种深度学习模型进行农作物分类研究。分析模型参数对农作物分类精度的影响,评价3种深度学习模型的农作物分类精度,优选农作物精细分类方法。结果 （1）学习率与3种深度学习模型的分类精度呈正相关关系,较大的学习率（0.01,0.001）下,3种模型收敛速度快,分类精度高。批样本量与模型分类稳定性相关,批样本量设为100时,3种模型的分类稳定性最好。（2）相比PSPNet、DeepLabv3+模型,U-Net模型分类效果最好,总体分类精度为89.32%。（3）GF-1 PMS影像结合U-Net模型可有效提升种植结构复杂区农作物分类精度,大宗作物春玉米、夏玉米的分类精度在80%以上,花生、红薯、蔬菜小宗作物分类精度在60%以上。结论 该研究可为准确获取种植结构复杂区的农作物类型、面积及空间分布信息提供参考依据。     

基于多时相遥感影像的花生种植面积提取——以彰武县北部为例

本文以彰武县北部4个乡镇(大冷乡、章古台镇、四合城乡和阿尔乡)种植花生的耕地为研究对象,基于不同农作物类型之间的物候特征差异,借助3个时期的遥感影像提取农作物的不同NDVI特征值和影像的特征信息,采用监督分类的方法提取花生的种植面积及空间分布情况。结果表明该方法对于区分不同物候特征下的作物分类精度高,较好的反映作物的分布状况,可为该地区主要作物种植分布情况提供科学依据。     

面向农业区划的作物种植结构遥感提取

[目的]尝试将作物种植结构提取结果由行政单元发展为基于相对均质的地理网格单元,解决当前农作物种植结构信息提取的空间局限性问题,文章在大尺度的土地利用/覆被分类与地块尺度的作物分类之间提出了作物种植结构单元概念,并构建了一种快速、低成本、准确的区域尺度作物种植结构提取方法。[方法]利用黑龙江省2014年250m分辨率的植被指数产品构建时间序列曲线提取物候信息,在耕地物候分区基础上对各物候区进行面向对象的多尺度分割,提取作物种植结构单元,利用光谱特征和NDVI指数构建特征空间,最终采用最邻近分类方法提取作物种植结构。[结果](1)利用MODIS时间序列数据提取物候特征进行多尺度分割的方法,能够有效的提取区域尺度农作物种植结构单元;(2)作物种植结构提取总体精度为95.70%;(3)黑龙江省2014年作物种植类型共有12种。其中,三江平原主要是水稻单一种植区、水稻混作区;松嫩平原以玉米单一种植区以及玉米-大豆混作区种植为主;西北部种植结构较复杂;东南部因地势等影响多种植玉米、大豆。[结论]利用物候数据进行种植结构提取可以有效划分农业区划,研究成果不仅为作物种植结构调整和农业发展布局提供科学依据,也是不同区域产量预测的基础,为合理布局农业生产、改进耕作制度以及引入和推广新产品等提供依据。     

多时相GF-1卫星PMS影像提取农作物种植结构

[目的]及时准确获取农作物种植结构信息,对农作物种植结构调整具有重要意义。[方法]文章以多期GF-1卫星PMS影像为主要数据源,通过分层分类方法成功提取了朱仙镇一个种植年内秋播蔬菜、冬小麦、春花生、早熟西瓜、夏玉米、夏花生等农作物,并通过叠加分析得到朱仙镇的农作物种植模式。[结果]朱仙镇有10种种植模式,以一年两熟的冬小麦-夏玉米为主要倒茬模式,占比49.7%,其他9种种植模式所占比例之和接近50%;从整体景观的角度对朱仙镇的主要作物种植模式进行衡量,斑块丰富度PR达到9.0,香农多样性指数SHDI达到1.483 6,说明朱仙镇种植模式多样,有利于多元农业种植结构发展;分层分类的方法能够延续最优提取结果的精度,各类作物提取精度均达到了80%以上。[结论]PMS影像的高分辨率能够清晰的表达地块边界,一定程度上呈现了非大宗作物的空间分布,GF-1卫星PMS为多尺度的农业监测提供了更多的选择。     

利用高空间分辨率遥感数据的农作物种植结构提取

农作物种植结构是掌握粮食种植面积和产量的重要前提,也是进行作物结构调整与优化的依据。该研究以黑龙江肇东市为研究区域,以高空间分辨率RapidEye影像为遥感数据, 基于最大似然监督分类方法提取了肇东市2011年农作物种植结构空间分布,利用地面样方调查数据进行了线状及细小地物扣除系数计算,实现遥感提取的农作物种植面积的精细提取,然后从面积数量和空间位置两个方面对遥感提取的农作物种植结构进行了精度评价。研究结果表明,利用RapidEye数据提取的农作物种植面积数据总体精度为97.00%,位置精度为96.15%,高空间分辨率数据在农作物种植结构遥感提取中具有重要潜力,线状及细小地物扣除系数可以有效减小线状及细小地物对高分提取的农作物种植结构的精度。     

若羌绿洲特色林果种植信息遥感提取方法适用性分析

目的 为快速掌握若羌绿洲特色林果业种植信息,提高特色林果资源调查的信息化水平,服务若羌特色林果业的发展。方法 文章通过野外调查和室内文献资料分析建立遥感分类的先验知识,采用面向对象的影像分析方法,构建多尺度分割结果的光谱、植被指数、形状、纹理等特征,采用CART和随机森林算法进行遥感分类,分析使用不同分类算法、特征集对分类结果的影响。结果 （1）随机森林算法相比于CART算法总体分类效果好,分类精度高,特征数据集的加入对CART算法的分类精度影响较小,而对随机森林算法分类精度的提升较为明显。（2）形状特征对形状差异较大的地物具有较强的识别能力,植被指数特征能有效识别植被和非植被地物,纹理特征则对分类精度的影响较小。（3）综合所有特征集并结合随机森林算法的分类结果最优,总体精度88.43%,Kappa系数85.47%,面积精度96.89%。结论 形状、植被指数、纹理等特征集的加入对各个地类的分类精度的影响不同,应具体问题具体分析,随机森林算法比CART算法更适用于多维特征数据集的作物遥感分类任务。     

基于中分辨TM数据的水稻提取方法对比研究

水稻种植面积监测是当前农业土地变化科学的热点问题,但运用遥感技术对水稻种植面积精确实施监测一直是难点。中分辨率遥感影像能够满足我国大面积水稻作物监测,成为业务化运行的主要数据源。为此,该研究尝试以中分辨率 TM影像为数据源,结合神经网络和面向对象 (SVM)两种算法对对黑龙江省富锦市2010年两期不同时相影像分别进行水稻分类提取,并对分类结果进行滤波处理及混淆矩阵精度评定。结果表明:(1)在高纬度单季稻生长区,面向对象分类算法的精度显著高于神经网络的分类精度,水稻用户精度和生产者精度在6月份分别高0.55%、1.37%,在8月份分别高0.62%、2.34%;(2)对神经网络分类的结果进行 Majority滤波处理,在一定程度上可以改善水稻分类的精度,水稻用户精度和生产者精度在6月份分别提高0.14%、0.5%,在8月份分别提高1.56%、1.43%;(3)选取关键水稻物候期的遥感影像获取水稻种植面积的精度更高,返青期水稻提取精度要高于乳熟期,其中神经网络算法的水稻用户精度及生产者精度分别提高 2.67%、3.45%;面向对象算法的水稻用户精度及生产者精度分别提高 2.6%、2.48%。未来需要重点考虑建立全国水稻物候历信息、面向对象算法中自动化最优尺度分割方法来提高水稻分类的精度。     

基于高分与多时相中分影像的苹果园地提取

目的 卫星影像已经被广泛用于小麦、玉米、水稻等种植范围较广的粮食农作物空间分布信息提取和面积估算。但是,利用遥感影像研究苹果园光谱与纹理特征,获取苹果园地面积与分布的应用研究较少。以往的研究已经指出高分纹理特征有利于苹果园地分类,但是中分影像的数据保障性会更强,可提供时序的光谱特征,因此,文章提出了一种综合利用非最佳观测期获得的高分辨率影像纹理信息和多时相中分辨率影像光谱信息提取苹果园地的方法。方法 该方法以GF-2号卫星及哨兵卫星数据为数据源,首先利用灰度共生矩阵（GLCM）提取GF-2的纹理特征,然后将提取的GF-2的纹理特征和光谱特征与5期不同时期的哨兵影像相结合,形成5组GF-2纹理和光谱特征与某一期哨兵影像的组合特征（记录为GF-2+单时相哨兵）,1组GF-2纹理和光谱特征与所有哨兵影像的组合特征（记录为GF-2+多时相哨兵）,以7组不同的组合特征（GF-2纹理和光谱特征、5组GF-2+单相哨兵和1组GF-2+多时相哨兵）为输入,使用3种不同的分类方法（C4.5、CART和Random Forest）提取苹果园地的分布,比较分析不同特征组合对苹果园分类精度的影响以及3种不同决策树学习方法的优劣。结果 苹果园提取时,GF-2+单时相哨兵特征组合与GF-2+多时相哨兵特征组合明显优于GF-2纹理和光谱特征组合,GF-2+4月（花期）哨兵是最佳特征组合,其苹果园的分类精度和整个研究区域的总体精度分别比GF-2单独分类提高至少8.14%和8.29%,比GF-2+多时相哨兵分类提高至少1.14%和0.95%。当采用相同组合特征时,随机森林分类方法提取苹果园地效果最好,苹果园提取精度均在90%以上。结论 利用非最优时相高分影像与花期中分影像提取苹果园能够得到最优的精度;如果没有花期中分影像,其他时相中分影像与非最优时相高分影像组合也能够得到比较优的精度。     

多时相极化SAR数据的旱地作物分类研究

[目的]及时、准确地获取旱地作物类型、种植面积及空间分布信息,可为农业生产管理,国家粮食政策提供重要依据。文章主要是对河北省冀州市棉花、玉米、水体和建筑进行分类,比较不同时相及分类方法下RADARSAT-2数据对4种地物的分类精度。[方法](1)计算得到每个时相(2018年7月14日、8月7日、9月24日)全极化RADARSAT-2数据的39个特征;(2)结合随机森林分类器比较不同分解方法(Freeman分解、Yamaguchi分解、MCSM模型和Cloud分解)得到的特征对旱地作物的分类精度影响;(3)分析参与分类的特征数量和时相对分类精度的影响;(4)将多时相多特征相结合,确定研究区内旱地作物的最佳识别方案。[结果]不同分解方法得到的特征越多、分类精度越高;就仅使用单景影像而言,在9月24日(玉米成熟期早期、棉花吐穗期中期)总体分类精度最高;不同分解方法得到的极化特征之间有一定的相关性,同一时相下,增加参与分类的极化特征数量不能有效提高分类精度;使用3个时相上117个极化特征,结合随机森林分类器,可以得到最佳分类精度(总体分类精度达92.89%,Kappa系数为0.885 9)。[结论]结合多时相与多特征相RADARSAT-2数据,能够有效提高复杂种植结构下旱地作物的识别精度,该研究可为旱地作物种植面积的快速提取提供参考。     

基于纹理特征与MODIS NDVI时间序列的耕地面积提取研究

[目的]耕地是粮食安全的重要前提与保障,通过对MODIS时间序列数据的特征分析和提取,讨论了在大尺度条件下的耕地面积提取的可行方法,以期为当地合理利用耕地资源进行农业生产规划与布局提供参考。[方法]文章以呼伦贝尔市为研究区域,选择MOD13Q1为数据源,采用Savizky-Golay滤波方法对MODIS-NDVI时间序列进行滤波处理,并分析各地物间的时间变化特征,最后结合纹理特征信息进行CART决策树分类的结果比较。[结果](1) MODIS-NDVI时间序列能较好的区分不同的土地利用类型;(2) Savizky-Golay滤波降噪能够明显提高分类精度;(3)纹理特征的加入进一步反映地物间差异性;(4)该方法的总体分类精度为83. 72%,Kappa系数为0. 789,其中耕地的提取准确度为86. 33%。[结论]纹理特征使数据像元间的灰度相关性更加丰富,并通过提高像元之间的差异改善结果的精度。该文结果为进一步利用MODIS数据进行土地利用信息与农业资源调查提取提出了新思路。     

The Effect of the Structural Adjustment Program on the Output,Prices and Exports of Major Agricultural Export Crops in Nigeria

The primary objective of this paper is to examine changes in the level of output, prices and exports of major export crops resulting from the Structural Adjustment Program (SAP) in Nigeria. THere two main sources of data: the Central Bank of Nigeria (CBN) through its annual reports, Economic and Financial Review, and the Nigerian Federal Office of Statistics (FOS) through its Trade Summaries. Although the SAP led to appreciable increases in output and producer prices, its effects on exports were insignificant. The implications of these for the implementation of the program are explored.     

Application of an ecosystem function framework to perceptions of community woodlands

Owners, local residents, government, and conservation organisations can express divergent preferences in the development and management of local woodlands. The perceptions of these four groups were examined, in the context of three community woodlands in Eastern England, using an ecosystem function framework. In a pilot study, residents were able to allocate a relative importance to woodland ecosystem services which were then related to “regulation”, “habitat”, and “production” or “information” functions. However residents also placed importance on negative services or “dis-services” associated with the woodland ecosystem. Therefore a fifth category of “dis-services” was included in the main survey which included 84 local residents, three woodland owners, three government institutions, and six representatives from conservation groups. Each of the four groups placed greatest importance on services associated with habitat (16–39% of the total importance) and information (30–50%) functions suggesting, in this example, mutual interest in the use of woodlands as a habitat or recreational resource. By contrast a potential area of difference was the particularly high importance placed by one owner on dis-services such as fly tipping. In addition the woodland owners placed higher importance (10–20%), than local residents and conservation groups (7–9%), on the productive services of the wood. This suggests a need for communication when production-related operations affect recreation. The ecosystem function framework appears to be a useful approach for highlighting potential tensions and areas of mutual interest in the management of semi-natural ecosystems.     

土地学科的建设和发展——"土地学科发展规划研讨会"综述

中国土地科学学科发展已取得很大成就,表现为专业教育和人才培养成绩突出、初步形成学科体系、取得大批研究成果.而存在的问题主要是基础理论研究欠缺、独特的土地科学研究方法体系尚未形成、研究角度单一、学术交流质量不高和学科带头人的影响力不足.土地科学学科发展目前面临着巨大的机遇和挑战,应明确土地科学学科发展的基本思路、发展目标及2010-2020年的主要建设任务和保障措施.     

Revealing different understandings of soil held by scientists and farmers in the context of soil protection and management

This paper aims to analyse and draw together results from similar studies in England, Switzerland and France which investigated farmers’ understanding of soil and compared it with that of scientists, researchers and advisors (collectively called scientists in this analysis). A range of methods were used across the three studies and different theoretical approaches, looking at forms of knowledge, local practice of knowledge production and conceptions of reality, were employed to explain the results. Despite the different contexts, methodologies and theoretical approaches in the three studies, the results reveal similar patterns of difference in farmer and scientist understanding of soil. In the English study, farmers demonstrate a ‘know-how’ form or intuitive working knowledge of soil while advisors rely on scientifically established forms of ‘know-why’ and seek to understand and explain soil processes. Similarly in the Swiss study farmers’ and scientists’ differing perceptions are directed and shaped by their respective aims, methods and context of work. In the French study, farmers and researchers are shown to have different conceptions of soil, they attribute different meaning to the same activities, and use different words and language to describe the same features. In all three studies understanding is shown to be cultural and contextual, as such an integrative theoretical framework is proposed.     

An evaluation framework for earthquake-responsive land administration

In recent years earthquakes and their secondary hazards have claimed the largest number of lives of all large natural disasters. Some of the world’s most earthquake-prone zones are also areas of high population density. The impact is magnified by vulnerability factors including non-enforcement of building codes, knowledge gaps, urban poverty and poor governance capacity to manage and reduce earthquake risks. Poor security of land tenure and property rights increases the vulnerability of people and affects their ability to respond to natural disasters.Earthquake recovery and reconstruction provides very significant challenges for land agencies, with these challenges differing from one country to the next due to differences in the local context. Drawing on contrasting case studies in Haiti, Nepal and New Zealand this paper identifies the common post-earthquake land administration functions and challenges that may apply to many contexts. These lessons provide land agencies and other key stakeholders with a summary of the challenges an earthquake poses for land administration at different post-disaster stages. We also discuss the policy and regulatory, institutional, operational and preparedness lessons for land administration. From these lessons we propose a framework for evaluating the earthquake-responsiveness of a land administration system. This framework can be used by a land agency in an earthquake prone region, or where an earthquake has recently occurred, to assess what challenges to land administration might occur in the event of an earthquake, and the preparedness of their land administration system.     

森林生物多样性评价模型的改进

针对2002年的评价模型中的不足,提出了新的改进后的森林生物多样性变化的评价模型,并在我国第六次森林资源清查(1999～2003)资料的基础上,对1973～2003年我国森林生物多样性的变化进行了补充评价。评价结果表明:我国森林生物多样性的压力越来越大,压力指数由1973年的100,上升到2003年的199.10,上升了1.99倍;森林物种多样性总体的下降趋势已有一定幅度的改变,森林物种多样性指数增加了31.12%。森林生态系统多样性和森林生物多样性先逐渐上升,尔后又下降,第六次森林资源清查时又有所恢复,并分别上升了4.67个百分点和17.90个百分点。     

Evaluating Alternative Policy Responses to Higher World Food Prices: The Case of Increasing Rice Prices in Madagascar

Higher world food prices have led many developing countries to adopt policies to mitigate the impact on low-income households. This article sets out a partial equilibrium framework to evaluate the efficiency, distributional, and revenue implications of alternative policy responses. The model is applied to evaluate tariff reductions and targeted transfers in Madagascar. Although lowering tariffs generates substantial efficiency gains, these accrue mainly to the top half of the welfare distribution, and poor net sellers are actually worse off. Developing a system of targeted direct transfers to poor households is likely to be a substantially more cost-effective approach to poverty alleviation.     

转变地质工作发展与服务方式的实践和建议

天津市地矿局注重加快转变经济发展方式,发挥地质工作优势,推进低碳清洁能源的开发利用创新和技术进步,紧紧依托地质工作的人才、技术和行业优势,加大清洁环保资源的勘查、研究、开发、利用。针对天津市水资源缺乏的情况,积极加强后备和应紧水源勘查。按照“两种资源,两个市场”的思路,通过实施中外合作勘查项目,以境外勘查工作为载体和平台,实施引进来和走出去战略,为天津市经济社会发展提供地质技术支撑。     

Trust-A monolithic panacea in land use planning?

Although trust research has been extensive and substantial, the constitution and actual merits of trust in land use planning remain nebulous. What do participants of a local land use planning process actually mean when they say they trust the planning committee? How important is participants’ trust in the planning committee in shaping their perceptions of the planning process and their cooperation in it? To explore the everyday meaning of trust in the specific context of local land use planning, semi-structured interviews with the participants of a local planning project in Switzerland were conducted (n = 9). Based on these results, the constitution and the relevance of trust in the responsible planning committee among a wider circle of participants were assessed (n = 66). The results show that participants trust the planning committee if they perceive its members as competent, honest, open, fair, reliable, reciprocating, respectful and committed. Interestingly, a considerable part of trust in the planning committee hinges explicitly on whether the committee incorporates the personal interests of the participant. Trust is indeed vital to participants’ belief that a meaningful discussion and cooperation with the members of the planning committee is possible. However, trust is neither a guarantee for people's approval of planning decisions nor for their cooperation. One implication of the findings is that in order to evaluate trust-building practice comprehensively, it is necessary to first understand the specific everyday meaning of trust for ‘real people’ in real planning processes.     

Agrarian Poverty,Nutrition and Economic Class – A Study of Gujarat,India

This paper analyses poverty and calorific undernourishment in the Indian state of Gujarat, where high and market‐led industrial growth has resulted in rapid economic improvement. The study is carried out through a combination of secondary and survey‐based data. We conclude that the neoliberal agenda of uncontrolled, outward‐looking growth has not resulted in significant reduction of poverty or malnourishment in rural areas. Furthermore, while land ownership is officially used as a proxy for wealth distribution, class position appears a better predictor of poverty status in the rural areas than landownership per se. At the policy level, there is a need to revive the agrarian economy and create new non‐agricultural assets, and the primary focus in the state must shift to the distribution of created assets rather than a single‐minded focus on growth.