3. Implementation of an NFA

The phase of implementing an NFA typically starts with the planning of data acquisition and comprises a series of activities ending with data analysis and dissemination of results. Our starting point for the discussion is that the overall design of the inventory has been decided upon and that it is the task of the organization established to carry out the inventory and compile the results. The conditions and needed work packages will be slightly different depending on whether an inventory is run on a continuous basis or not. With an inventory run at certain time intervals only, the data acquisition phase may extend over several years being followed by an analysis and dissemination phase. With an inventory run on a continuous basis, each year may be a cycle of data acquisition and analysis; in this case the analyses typically will be based on average data from the last 3-5 years.

The main elements below will remain the same regardless of approach, although we base the discussion on a case where the inventory is run on a continuous basis. The cycle of implementing an NFA may be presented as comprising the following steps, focusing on an NFA mainly based on field sampling:
  1. Conversion of theoretical design into practice
  2. Development or up-dating of manuals for the inventory
  3. Development or up-dating of data capture procedures
  4. Acquisition of materials needed for the sample surveys
  5. Hiring and training of staff for the data collection
  6. Data collection
  7. Conduction of independent check assessments
  8. Data control and compilation of databases
  9. Analyses
  10. Dissemination of results
In addition to this, at certain intervals there is a need for evaluating the entire NFA and implement changes due to possible shortcomings identified and new emerging needs from the user society. We do not treat this issue further here.

In the following sections, each part of the inventory implementation cycle will be discussed in more detail.

Conversion of theoretical design into practice
The theoretical inventory design needs to be converted into practice. Regarding field sampling, this involves determining the actual sample sizes and the distribution of sample units over the area to be surveyed. In many cases there will be a need for an initial stratification of the land area, allocating, e.g., areas that are not accessible to specific strata to be handled according to separate procedures. With digital maps and GIS available, all this work may be straightforward. With maps only available in paper form, quite substantial work may be involved in determining land areas and plotting the sample plots on the maps according to the design. In the latter case, this work may need to start a long time before the actual field sampling starts.

Although modern GPS equipment is available in most parts of the world, low and high resolution maps over the areas to be field sampled still are important for the field teams. For example, maps are needed for finding the best routes to the plots to be visited, and the map information may also be helpful in identifying land ownership in countries where permits must be worked out prior to visiting plots in the field.

Manual for the inventory
Written instructions need to be worked out in good time before the actual fieldwork, so that they are available for the training of the field staff. For securing the data quality, it is very important that the manuals provide unambiguous routines and definitions for the assessments to be made.

Data capture procedures
Many different issues need to be considered in choosing between using field computers or field forms for the capture of data in the field. There are many advantages with a well functioning field computer system for the data capture. Fieldwork will often be simplified, especially if definitions and assessment options are displayed in the field computer. In addition, checks for valid values and simple plausibility checks of data can be performed. However, large resources may be needed both for developing the system and for maintaining it, and use of computers increases the demand for training of the field staff.

Thus, often it may be advisable to use procedures based on field forms, although field computers - when working well - considerably simplify fieldwork. Use of field forms also requires a subsequent step of registering data on computer media before the analyses. This step may be expensive and it may also introduce registration errors in the data. On the other hand, as a back-up system easily available to more than a few persons, field forms do have advantages. In the part of the data collection which is based on interviews, computers might also alienate the interviewers from the interviewees.

Acquisition of materials
Forest inventories require a number of different measurement instruments, like callipers, hypsometers, and equipment for determining distances. During the last decade, the traditional instruments have been supplemented with modern instruments based on computer techniques. Thus, today there is an option to use the traditional instruments or buying the new digital instruments.

A general advice in this regard is to adapt to the level of technical competence in the country, also considering the labour costs in relation to the increased efficiency of using the new instruments, and the potential measurement errors induced by using different instruments. A general remark is that the new instruments generally reduce the risk for errors, provided they are properly used. However, they need to be properly maintained and the larger the extent of digital instruments in an inventory is, the more vulnerable the inventory is to technical problems.

Hiring and training of staff for the data acquisition
Due to limited resources, field staff may need to be hired only on a temporary basis during the field data collection phase. Thus, part of the personnel is likely to be new every year. This introduces a need for repeated and consistent training of the field staff. The issue of consistency is very important. It may be intuitively obvious that the ambition should be to always increase the level of competence of the field staff. However, this may complicate trend assessments, which would then be based on data of varying quality. In NFAs the problem with varying systematic errors between years may be severe, and in the worst case it may lead to completely erroneous conclusions regarding trends in the development of forest resources. Thus, the issue of training and maintaining the competence of the field staff on a stable level is important (cf. Bell 1998). If possible, routines should be established whereby assessments can be consistently performed between years. This may involve training always following certain standard routines.

Data collection
In many countries an important part of the planning of field data collection is to check who owns or manages the land where a field plot will be situated. The reason for this may be three-fold. Firstly, there is a need to obtain a permit for accessing the land. Secondly, ownership information may be important in the analyses. Thirdly, there is a need to identify the stakeholders before interviews can take place.

During the field data acquisition, it is important to keep continuous contact with the field teams. One way of assuring this is to have specific people visiting the field teams during their work.

In the actual work, GPS equipment generally is a very good (and rather cheap) complement to the field maps; with GPS, the teams can locate the plots in the field with high accuracy. In particular this is important when permanent plots are to be established or revisited. When planning procedures for the exact location of field plots, it may be advisable to use some kind of final randomization procedure regarding where a plot should be situated rather than using the GPS all the way to the plot center. The reason for this is that there is otherwise a risk for systematic errors in the plot locations due to, e.g., poor GPS receiving conditions in dense forests.

Using permanent plots there is a need to establish the plots in a manner so that they can be found at a later time point when they are to be re-inventoried. It is important that the marks introduced for this reason be very discreet, so that the plot locations will not become known to land managers. Otherwise there may be a risk that the permanent plots within the NFA develop in a different manner than other land areas, and thus the results of the NFA will be biased.

From the organizational point of view, there are many administrative routines that need to be worked out to assure that correct salaries and reimbursements be properly paid. Moreover, the routines for submitting data from the field to the NFA office need to be properly established, in order to avoid loss of data.

Check assessments
Check assessments should be considered an integral part of any field data acquisition campaign within an NFA. Firstly, by being clear about that a certain percentage of the plots will be revisited and checked by a control team, the ordinary teams obtain incentives for maintaining ¿high moral¿ during their work. This may be the most important reason for performing check assessments. In addition, through check assessments it is possible to obtain insight into data quality issues that would otherwise not be possible.

In analyzing data from the check assessment for some types of variables it is possible to assess the actual errors committed during the ordinary inventory. This may be the case for variables such as the number of trees on a plot and tree diameters. For other variables, e.g. like the land-use class, subjectivity in the assessments necessarily needs to be used also by the control team. In this case, the check assessment will show the variability among different crews rather than the deviation of the ordinary team from the true value.

With very ambitious check assessments, it may be relevant to adjust the original estimates based on the results from the check. However, this introduces additional complexity to the NFA.

Data control and compilation of databases
In addition to the consistency checks that are - hopefully - performed already in the field, and the check assessments, at final stages of the data compilation completeness tests must be performed in order to ascertain that all data have been collected. Moreover, increased checks based on logical relationships between different variables may be conducted.

Having finished all checks of data, the final databases can be established.

As pointed out previously, the analyses may be of many different kinds. As part of the inventory cycle every year, it may be advisable only to include certain standard results. The reason for this is that integrated analyses including scenario analysis are very demanding, and can thus - due to the limited resources available - be carried out at certain intervals. Also, it may be wise to concentrate on different analyses different years.

In case the NFA is based on a complex design, standard routines for assessing state or change could be implemented based on the NFA databases.

Dissemination of results
The final stage of the inventory cycle is the dissemination of major results to the major stakeholders. This may be conducted in different ways, and although we do not elaborate far on different options here, a few remarks are motivated.

Especially when an NFA is newly established, there is often a great need to make stakeholders aware of its existence, in order to discuss how it can be used and in what form results should be presented, etc. Therefore, it is not advisable to wait until all results are finalized before starting dissemination. Otherwise, there is an obvious risk that data will not be used according to what was planned, but remain in files. Many inventory projects have actually faced the situation that results have been finalized only in the form of huge amounts of statistics, when monetary resources have been too limited, and interest and attention have faded. A careful planning of continuous dissemination of results can prevent this situation.