Impact of feature selection on the accuracy and spatial uncertainty of per-field crop classification using Support Vector Machines

Crop mapping is one major component of agricultural resource monitoring using remote sensing. Yield or water demand modeling requires that both, the total surface that is cultivated and the accurate distribution of crops, respectively is known. Map quality is crucial and influences the model outputs. Although the use of multi-spectral time series data in crop mapping has been acknowledged, the potentially high dimensionality of the input data remains an issue. In this study Support Vector Machines (SVM) are used for crop classification in irrigated landscapes at the object-level. Input to the classifications is 71 multi-seasonal spectral and geostatistical features computed from RapidEye time series. The random forest (RF) feature importance score was used to select a subset of features that achieved optimal accuracies. The relationship between the hard result accuracy and the soft output from the SVM is investigated by employing two measures of uncertainty, the maximum a posteriori probability and the alpha quadratic entropy. Specifically the effect of feature selection on map uncertainty is investigated by looking at the soft outputs of the SVM, in addition to classical accuracy metrics. Overall the SVMs applied to the reduced feature subspaces that were composed of the most informative multi-seasonal features led to a clear increase in classification accuracy up to 4.3%, and to a significant decline in thematic uncertainty. SVM was shown to be affected by feature space size and could benefit from RF-based feature selection. Uncertainty measures from SVM are an informative source of information on the spatial distribution of error in the crop maps.     

Projection Pursuit Evaluation Model: Optimizing Scheme of Crop Planning for Agricultural Sustainable Development and Soil Resources Utilization

Planting structure influences the economic, social, and ecological benefits of crop farming as well as the use efficiency of water and arable land resources, and so crop planning (CP) benefits for agricultural sustainable development and soil resources utilization. The projection pursuit evaluation (PPE) model is put forward to solve the problem of selecting an optimizing scheme for CP by considering the indices of water‐saving and economic, social, and ecological benefits. The real‐coding‐based accelerating genetic algorithm (RAGA) is introduced to accelerate the calculation process. The model can translate multi‐indices into a single index by transforming high‐dimensional data to low‐dimensional space, which helps evaluate CP optimizing schemes. For example, the model is used to evaluate and select an optimal scheme of CP in the middle reaches of the Heihe mainstream basin in the arid area of northwest China. According to four criteria (high efficiency of resources use, economic rationality, social equity, and ecological security) 19 indices were chosen to evaluate 12 optimizing schemes of four kinds (economic‐benefit, food‐security, ecological‐benefit, and water‐saving programs) in 2006, 2020, and 2030. The result shows that, in the 3 years, the water‐saving program is always the optimized scheme in an arid region with water deficiency and fragile ecology. The evaluated results match up to the developmental conditions of crop farming in recent years. Moreover, the direction of the optimal projection could reflect the weight and orientation of indices objectively and accurately.     

Quantitative analysis of agricultural land use change in China

This article reviews the potential impacts of climate change on land use change in China. Crop sown area is used as index to quantitatively analyze the temporal–spatial changes and the utilization of the agricultural land. A new concept is defined as potential multiple cropping index to reflect the potential sowing ability. The impacting mechanism, land use status and its surplus capacity are investigated as well. The main conclusions are as following;

• 1.During 1949–2010, the agricultural land was the greatest in amount in the middle of China, followed by that in the country’s eastern and western regions. The most rapid increase and decrease of agricultural land were observed in Xinjiang and North China respectively, Northwest China and South China is also changed rapid. The variation trend before 1980 differed significantly from that after 1980.
• 2.Agricultural land was affected by both natural and social factors, such as regional climate and environmental changes, population growth, economic development, and implementation of policies. In this paper, the effects of temperature and urbanization on the coverage of agriculture land are evaluated, and the results show that the urbanization can greatly affects the amount of agriculture land in South China, Northeast China, Xinjiang and Southwest China.
• 3.From 1980 to 2009, the extent of agricultural land use had increased as the surplus capacity had decreased. Still, large remaining potential space is available, but the future utilization of agricultural land should be carried out with scientific planning and management for the sustainable development.

     

Analyzing the agricultural transition in Mato Grosso, Brazil, using satellite-derived indices

The Amazonian state of Mato Grosso is the main production area for soybeans in Brazil and contains 31.3% of the national production as of 2009. The rapid evolution of the agricultural systems in this area shows that the region is experiencing a rapid agricultural transition. In this paper, we broke down this transition process into three steps: crop expansion, agricultural intensification and ecological intensification. We used remote sensing products to develop and compute satellite-derived indices describing the main agricultural dynamics during the cropping years from 2000-2001 to 2006-2007. Our results indicated that Mato Grosso is continuing to expand its agricultural sector, with a 43% increase in the net cropped area during the study period. Although this expansion mainly occurred in the cerrado ecoregion until the early 2000s, the forest ecoregion is experiencing expansion at this time. We observed that 65% of the crop expansion in Mato Grosso from 2000 to 2006 occurred in this ecoregion. However, we did not identify this crop expansion as the major driver of deforestation in Mato Grosso because only 12.6% of the cleared areas were directly converted into croplands. Agricultural intensification also evolved rapidly, as the proportion of the net cropped area cultivated with double cropping systems harvesting two successive commercial crops (i.e., soybean and corn or soybean and cotton) increased from 6% to 30% during the study period. Finally, we found that ecological intensification occurred because the region’s farmers planted a non-commercial crop (i.e., millet or sorghum) after the soybean harvest to prevent soil erosion, improve soil quality, break pest cycles, maintain soil moisture and set the conditions for high-quality no-tillage operations. In 2006-2007, 62% of the net cropped area was permanently covered by crops during the entire rainy season. This practice allowed the farmers to diversify their production, as shown by the positive evolution of the Area Diversity Index. Future scholars can use the method proposed in this paper to improve their understanding of the forces driving the agricultural dynamics in Mato Grosso.     

Coping with land scarcity. Farmers' changing land-use and management practices in two mountain watersheds of Nepal

In this study some light is shed on farmers' changing land-use and management practices in two mountain watersheds located in the Western Hills of Nepal. The study is based on a survey of 300 households, group discussion, interviews of key informants, and field observation in project and non-project watersheds conducted from April to September 1999. Confronted with shrinking landholding size, owing to a steadily growing population and scarcity of non-farming employment opportunities, farmers in both watersheds have gradually intensified land use and cultivated new crops to increase farm production and income. They are shifting from cereal crops to livestock husbandry, particularly along the road in the project watershed, and to vegetables and other cash crops in the accessible foothills of the non-project watershed. They have also adopted various structural and biological measures to control soil erosion, landslides, gully expansion and soil nutrient loss in order to maintain or enhance land productivity. The degree of adoption of the structural and biological measures is higher in the project watershed than in the non-project watershed. Contrary to the traditionally held belief of some researchers, population pressure on a finite land resource has brought about positive changes in land-use and management practices. Farmers have innovated and adopted different land management technologies to increase farm production as they are exposed to the risk of food insecurity because of shrinking landholding size and land degradation.     

60年来我国主要粮食作物适宜生长区的时空分布

模拟作物适宜生长区的时空分布是分析气候变化对作物生长影响、提高作物生长适应能力的重要内容。选择影响主要粮食作物(小麦、玉米和水稻)生长的气候要素,结合地表土壤和地面高程要素与农业观测站数据,模拟和分析1953—2012年主要粮食作物适宜生长区的变动,评估气候变化下作物的适应能力。研究发现:(1)60年来3种粮食作物适宜生长区对气候变化响应程度从大到小依次是小麦、水稻和玉米。(2)同一时空尺度上,主要粮食作物适宜生长类型区在南方农区较北方农区多样化,在山地较盆地多样化,在高原较平原多样化。(3)小麦生长适应气候变化的能力在多数农区略有上升。玉米生长的适应能力在北方和南方农区分别略微提高和下降。水稻生长的适应能力在长江中下游区、西南区和华南区相对稳定,在黄淮海区和东北区分别下降和提高。(4)60年来,主要粮食作物综合生长适应气候变化的能力在黄淮海区和长江中下游区下降,在其余农区升高。(5)玉米和水稻适宜生长区分别与播种面积和作物产量显著相关,这为模拟未来不同气候情景下二者适宜生长区的分布提供了可行性。小麦适宜生长区与播种面积和产量均不显著相关,未来需要考虑更多因素精准识别小麦适宜生长区,以便更为有效地提高小麦生长对气候变化的适应能力。     

Application of AMD to the Determination of Crop Protection Agents in Drinking Water. Part IV: Fundamentals of a Confirmatory Test

A method is described for the determination of fifteen active ingredients of plant protecting products in ground-, raw, and drinking water. After extraction and enrichment of the pesticides from the water sample with solid-phase extraction, the extract is fractionated on silica-coated TLC plates, first in a so-called screening gradient by AMD (automated multiple development). The detection of the pesticides is performed by UV multidetection. Positive results from the first separation are confirmed by a second separation on silica in an AMD gradient of distinctly different selectivity, compared to the first separation. Following this strategy, it is possible to distinguish nearly all nonvolatile, from water extractable, active ingredients of plant protecting products in the market. UV spectra are taken for further confirmation of positive results. In most cases, this is possible even at the limit of determination. All chromatograms and spectra shown are raw data, resulting from the confirmatory test DIN 38407 part 11, in the laboratory of one participant*). The method is standardized in the meantime as DIN V 38407-11.     

冠层CO2分布廊线的农业气象数值模拟研究

本文将农田微气象模式与冬小麦冠层光合模式进行耦合，建立了一个具有较强机理的冬小麦冠层CO2分布廓线模式，冠层光合模式中考虑了气孔对叶片光合的调节作用，具有明确的生物学意义，实测资料验证表明，模式可以较准确地模拟拔节期冬小麦冠层CO2分布状况，平均相对误差为6．47％，数值分析表明：当冬小麦冠层风速为1m/s时，CO2廓线弯曲最为明显，随着风速加大，CO2廓线弯曲程度变小，CO2浓度升高后风速对廓线影响的基本规律没有发生改变，但是冠层中CO2浓度差将进一步加大。     

作物模拟模型在农业气象业务应用中的研究初探

针对农业气象业务的需求，利用东北地区1971～2000年5个代表气象站的气象资料和作物资料，应用玉米生长模拟模型评估东北区域气象条件对玉米生长发育及产量形成的影响。探讨了作为评价标准的平均气候状态下平均模拟生物量的确定方法，提出了从生物量变化、温度和水分条件影响等方面进行实时动态评价和全生育期农业气象条件综合评价的方法，给出了实际评价个例。验证结果表明该方法是可行的。