Over View of the Book
Chapter 1 describes the data sets relevant to the integration of remote sensing within agricultural statistics. First,
the remote sensing data sources themselves will be addressed. The role of reference data and ancillary data layers,
and their use in stratification and aggregation, will also be discussed. A key issue arising recently in data access is
the general trend towards open access. Thus, while the major commercial remote sensing missions will be listed for
reference, attention will be paid to open-access sources for a number of reasons: (1) to create wider awareness on
data sets that are available under free and open licenses;
(2) to provide a thorough understanding of how these data can be assessed; and (3) to discuss how open data can be used to optimize the acquisition – i.e. minimize the costs – of commercial data. Remote sensing data, from both open and commercial sources, usually requires post-processing for follow-up use in statistical analyses. The use of open-source software for image processing and geospatial analysis is another important development that accelerates the broader adoption of remote sensing data. The use of open-source software is discussed, and a listing of commercial software alternatives is provided.
The chapter concludes with a discussion of the more recent trend to move data analytics into cloud computing environments.
(2) to provide a thorough understanding of how these data can be assessed; and (3) to discuss how open data can be used to optimize the acquisition – i.e. minimize the costs – of commercial data. Remote sensing data, from both open and commercial sources, usually requires post-processing for follow-up use in statistical analyses. The use of open-source software for image processing and geospatial analysis is another important development that accelerates the broader adoption of remote sensing data. The use of open-source software is discussed, and a listing of commercial software alternatives is provided.
The chapter concludes with a discussion of the more recent trend to move data analytics into cloud computing environments.
Chapter 2 deals with land cover mapping. It first introduces the concept of land cover and then reviews some key
elements of land cover mapping. Existing land cover maps are discussed systematically, based on a set of welldefined criteria. While land cover map supporting stratification always refers to previous years, recent experiences of
map production throughout the ongoing season will also be explored. Today, land surface can be described in several
ways, thanks to the unprecedented development of information technology and observation capabilities, ranging
from Unmanned Aerial Vehicles (UAVs) to in-orbit Earth Observation (EO) platforms.
Satellite remote sensing is an undisputed source of land information for a vast range of users at all geographical scales. Due to the increasing gap between remote sensing producers and map users, which is very much supported by spatial data infrastructure making a great deal of geographic information widely available, it is important to understand the different concepts and constraints underlying land cover mapping.
This becomes even more critical when considering the use of a land cover map to support stratification at the sampling design level in the context of agricultural statistics. Indeed, maps derived from remote sensing that show, for instance, crop intensity classes, may significantly reduce sampling variances or, simply, reduce ground sampling effort and its associated costs. A land cover map can highlight the nonagricultural strata which should not be sampled or those strata which could be sampled differently. The efficiency of stratification is obviously related to the relevance of the land cover map selected for the stratification.
Satellite remote sensing is an undisputed source of land information for a vast range of users at all geographical scales. Due to the increasing gap between remote sensing producers and map users, which is very much supported by spatial data infrastructure making a great deal of geographic information widely available, it is important to understand the different concepts and constraints underlying land cover mapping.
This becomes even more critical when considering the use of a land cover map to support stratification at the sampling design level in the context of agricultural statistics. Indeed, maps derived from remote sensing that show, for instance, crop intensity classes, may significantly reduce sampling variances or, simply, reduce ground sampling effort and its associated costs. A land cover map can highlight the nonagricultural strata which should not be sampled or those strata which could be sampled differently. The efficiency of stratification is obviously related to the relevance of the land cover map selected for the stratification.
Chapter 3 focuses upon the use of remote sensing at design level in list and area frames. In the context of censuses,
surveys or registers, satellite imagery can be of great support when defining or optimizing the design options.
The imagery may be of primary importance when reference maps are absent or obsolete, as they enable a clear
delimitation of the Enumeration Area (EA), the counting of dwellings and the planning of the workload. With
reference to surveys, stratification on classified imagery will lead to a reduced sampling variance and a variation
of sampling fraction (or of the probability proportional to size – PPS) that is proportional to agricultural intensity.
Particular attention is paid to the creation of list frames, starting with the point area frame.
With regard to area frames, if, in stratification, the strata should be as different as possible, in two-stage sampling, the Primary Sampling Units (PSUs) should be as similar as possible. In both scenarios, the imagery is of great help. The chapter reviews practical examples in developing and developed countries, thus illustrating the type of efficiency and homogeneity that can be achieved. Recommendations are given on segment size optimization in function of field pattern complexity.
With regard to area frames, if, in stratification, the strata should be as different as possible, in two-stage sampling, the Primary Sampling Units (PSUs) should be as similar as possible. In both scenarios, the imagery is of great help. The chapter reviews practical examples in developing and developed countries, thus illustrating the type of efficiency and homogeneity that can be achieved. Recommendations are given on segment size optimization in function of field pattern complexity.
Chapter 4 is to provide an overview of remote-sensing-based approaches for detailed
(field-level) annual crop mapping at national scale. First, an overview of the existing approaches based on remote
sensing used for cropland mapping is presented. This includes a brief overview of supervised image classification
and pixel- versus object-based classification. Second, the various types of satellite data, ground data and secondary
data used for detailed crop mapping are discussed.
Third, the operational implementation of a national crop mapping program is demonstrated with specific reference to Canada’s Annual Crop Inventory. Finally, the main challenges and opportunities for crop type mapping at national scales in the future are outlined. The past decade has borne witness to several attempts to articulate the spatially explicit requirements of remote sensing data to map cropping systems, and, particularly, where, when and how frequently,
over which spectral range, and at what spatial resolution, data are needed. Elucidating the best data and methodologies for crop mapping remains a high priority on the international research agenda. Indeed, several international efforts have been made to achieve a convergence of approaches and develop monitoring and reporting protocols and best practices for a variety of global agricultural systems (e.g. the GEOGLAM initiative, which includes the Joint Experiment of Crop Assessment and Monitoring (JECAM), the Asian Rice Crop Estimation and Monitoring initiative (Asia-RiCE), the Stimulating Innovation for Global Monitoring of Agriculture activity (SIGMA), and contributions from the Sentinel-2 for Agriculture system (Sen2-Agri).
Third, the operational implementation of a national crop mapping program is demonstrated with specific reference to Canada’s Annual Crop Inventory. Finally, the main challenges and opportunities for crop type mapping at national scales in the future are outlined. The past decade has borne witness to several attempts to articulate the spatially explicit requirements of remote sensing data to map cropping systems, and, particularly, where, when and how frequently,
over which spectral range, and at what spatial resolution, data are needed. Elucidating the best data and methodologies for crop mapping remains a high priority on the international research agenda. Indeed, several international efforts have been made to achieve a convergence of approaches and develop monitoring and reporting protocols and best practices for a variety of global agricultural systems (e.g. the GEOGLAM initiative, which includes the Joint Experiment of Crop Assessment and Monitoring (JECAM), the Asian Rice Crop Estimation and Monitoring initiative (Asia-RiCE), the Stimulating Innovation for Global Monitoring of Agriculture activity (SIGMA), and contributions from the Sentinel-2 for Agriculture system (Sen2-Agri).
Chapter 5 deals with crop area estimation using remote sensing. The chapter introduces the history of crop area
estimation, reviewing the evolution from the use of conventional methods to the use of satellite data, with the
attendant challenges and complexities. The major initial crop area estimation programmes using satellite data,
such as LACIE and AgRISTARS, are discussed. The various approaches to crop area estimation,
such as the Area Sampling Frame (ASF), pixel counting, and regression or calibration estimators are described with examples. Details of current major programmes for use of remote sensing in crop area estimation are provided under three categories: national (USDA-NASS’s CDL and India’s FASAL); regional (the European Commission’s MARS); and global (USDA’s FAS, China’s CropWatch and GEOGLAM).The concluding section deals with the major issues and limitations in remote-sensing-based estimates and the way forward.
such as the Area Sampling Frame (ASF), pixel counting, and regression or calibration estimators are described with examples. Details of current major programmes for use of remote sensing in crop area estimation are provided under three categories: national (USDA-NASS’s CDL and India’s FASAL); regional (the European Commission’s MARS); and global (USDA’s FAS, China’s CropWatch and GEOGLAM).The concluding section deals with the major issues and limitations in remote-sensing-based estimates and the way forward.
Chapter 6 reviews the fundamental concepts relating to Early Warning Systems (EWSs) and crop yield forecasting,
to better address the climatic risks that bear an impact on food security. System-based dissemination of timely
alerts and specifications of the probability of hazard occurrence are fundamental components of early warning
information; systematic linkages to early action options and possibilities would go a long way towards saving lives
and livelihoods. Forecasting crop yields and aggregate production is of significant importance in early warning
systems that seek to assess the food supply and demand situation of a given country or region.
Accurate analyses of market conditions, and identifications of the surplus and deficit areas in a country or region will contribute greatly to design appropriate policy responses to mitigate food security problems. Robust and accurate agricultural statistics are also crucial to achieve such important objectives. In this context, information derived from remote sensing plays a vital part in improving the production of agricultural statistics because it is capable of introducing independent verifying mechanisms, particularly when area frame or multiple frame sample designs are used. Remotely sensed data and information can be introduced at both design and estimator levels.
Accurate analyses of market conditions, and identifications of the surplus and deficit areas in a country or region will contribute greatly to design appropriate policy responses to mitigate food security problems. Robust and accurate agricultural statistics are also crucial to achieve such important objectives. In this context, information derived from remote sensing plays a vital part in improving the production of agricultural statistics because it is capable of introducing independent verifying mechanisms, particularly when area frame or multiple frame sample designs are used. Remotely sensed data and information can be introduced at both design and estimator levels.
Chapter 7 deals with the estimation of forest cover and deforestation from global to national scales using Earth
Observation technology. Considering the specificities of forestry statistics (permanence of the stands from year
to year, plot sizes far exceeding pixel sizes, long-term management, availability of management registers in nonnatural forests), a special chapter is dedicated to forest resources and deforestation. The main approaches to the use
of remote sensing for forest cover assessment and evolution are reviewed, with particular focus on specificities and
results as shown in the recent literature.
After reviewing the background information on the use of remote sensing for monitoring forest cover, the Remote Sensing Survey of FAO’s Global Forest Resources Assessment is described, as well as other examples of remote sensing surveys used for forestry statistics. Finally, the complementarity between estimates of changes occurring in forests and agriculture is analysed
After reviewing the background information on the use of remote sensing for monitoring forest cover, the Remote Sensing Survey of FAO’s Global Forest Resources Assessment is described, as well as other examples of remote sensing surveys used for forestry statistics. Finally, the complementarity between estimates of changes occurring in forests and agriculture is analysed
Chapter 8 presents fundamental requirements and criteria for an organization that is beginning to use geospatial
analysis and, in particular, remote sensing for producing agricultural statistics. It also elucidates the need for
resources and the competences necessary for application of remote sensing systems in the contexts of agricultural
data collection and training needs. Furthermore, consideration is given to the human resources required in the
multidisciplinary team, its qualifications, size and to the budget required.
Examples of collaboration between statistical services and mapping agencies are also provided, as well as
explanations on the importance of close interaction with stakeholders.
The necessary budgets and business plans are presented.
Finally, Chapter 9 explains how to evaluate the cost-efficiency of remote sensing. Examples are given of past and
recent uses, showing why and where clear cases of cost-efficiency exist. Based on the current trend for free and open
access to satellite imagery, agricultural complexity may soon be expected to become manageable with the images’
increasing information content (spectral, spatial or textural)
Download Here
Posts
No comments:
Post a Comment