Work methodology
The INFES; poses a challenge of logistics and integration of diverse information collected, for which an operational strategy must be developed that allows obtaining the maximum of information, considering its quality, avoiding human errors as much as possible. Being the training of the key and operational team, a core point in the successful development of the INF, within a planning framework that allows efficiently the development and systematization of field and office activities.
In this section, the systems to be used to collect the necessary information to generate the expected products will be discussed in general.
The various techniques for managing remote censors (satellite images), photogrammetric, photointerpretation, sampling, syndrome cartography and processing to be used in the relation of a forest inventory depend on the characteristics of each inventory. However, many of the elements of the overall design system will be common to all inventory levels.
The National Forest Inventory of El Salvador (INFES) will be developed through the recommended standards for the development of management forest inventories and will include the following phases
Phase I. Planning and design of the forest inventory
This forest inventory can be defined as the joint work of information related to the location, ownership, nature and productive capacity of forests, as well as their possible evolution, in order to provide both the state and the company private, timely and necessary information, for correct decision-making on the sustained use of the country's forest resources. The components that make up phase I are the following:
Component No. 1 Conformation of the team
The team of principal technicians assigned to the project is made up of a select group of professionals with extensive experience in the forestry field, specializing in GIS, photointerpretation, planimetric restitution, stratification and photointerpretation of forests, as well as solid knowledge on sampling methods and excellent abilities and skills for the management of forest measurement equipment.
The composition of the team will be:
One Project Manager
Three photo interpreters
Component No. 2 Training
In this phase, the training activity gives priority to strengthening knowledge about Dasometry, photointerpretation and geographic information systems. The methodology used will be that of "learning by doing", for which they will be trained directly in the field.
In compliance with what is established in the terms of reference, In addition to the inventory executing personnel, three technical members of the forest service staff and three technicians from institutions related to forestry activities will be trained, or a total of six technicians.
Component No. 3 Search for existing information
The largest amount of existing information will be collected, in order to adequately support the low level and avoid duplication of efforts, in cases where there are already results applicable to this work.
Among the information to look for we have:
- Images of satellite existents; Aerial photos existing from 1979 to date; Land use maps; Maps of terrestrial and aquatic ecosystems; Maps topographical, at a scale of 1: 25,000; Digital base map of El Salvador; Map of vegetation cover of El Salvador; Maps hydrological; Maps pedologíca; Maps Agrological.
Component No. 4 Typification and Stratification
Typification and stratification are tools; that seek to reduce the statistical bias at the time of information capture in the field, under the premise of homogenization of the field sample. It can be considered that this component is the most important office activity, using secondary information (vegetation cover map) and photo-interpretation (analysis of aerial photography).
Subcomponent No. 4.1 Typification
The forest cover map developed by the Ministry of the Environment and Natural Resources (MARN) is in line with the classification used by the FAO Forestry Department for the classification of forests at the global, regional and national levels.
The types of forests presented in the coverage map are
Subcomponent No. 4.2. Stratification
Properly identified each type of forest, we will proceed to stratification by type of forest. The vertical structure is the most used stratification method, considering that vertical stratification responds to the characteristics of the species that compose it and to the microenvironmental conditions present at the different heights of the profile.
These differences in the microenvironment allow species with different energy requirements to be located at the levels that best meet their needs. In addition to the vertical structure, the density of the tree population that makes up the forest mass, will be another factor to use for the stratification process.
Stratification is very important to obtain greater precision in estimating the average data of the forest structure. Stratification identifies forest areas that are distinguished by biophysical characteristics and / or floristic composition and vegetation structure. Inventory plots in different strata should differ more from each other than plots within the same stratum.
It is important to have a clear description of the strata, with easily recognizable characteristics in the forest itself. Thus, it is possible to verify in the field the stratification will initially be carried out based on photos or satellite images.
Under the previous premise, INFES will use the vertical stratification system, using the following classification:
1. High Forest
1.1. High Dense Forest
1.2. Semi-dense High Forest
1.3. Alto Ralo or Open Forest
2. Middle Forest
1.1. Medium Dense Forest
1.2. Medium Semi-dense Forest
1.3. Medium Rare or Open Forest
3. Low Forest
3.1. Low Dense Forest
3.2. Low Semi-dense Forest
3.3. Low Forest or Open
Component No. 5 Preparation of base map
Making use of planimetric restitution methods, for the elaboration of the forest base map, the results of the stratification work will be emptied on the corresponding basic maps; of Vegetal Cover of El Salvador, Topographic Maps, at scale 1: 25,000 and Digital Base Map of El Salvador.
NHA has the most modern technology and a complete Geographic Information System (GIS). In this study, the technical and human resources necessary for the advice and development of the various works included in this methodological offer will be put at the service of the executing team.
At the service of the "National Forest Inventory of El Salvador", in addition to the material and means that NHA already has, those agreements that are considered necessary to implement for the normal development of the works described below will be placed:
To frame this diagnosis, it is essential to develop a Base Cartography of forest resources and a thematic cartography that reflects the understanding and scope of the expected products.
For the cartographic development, the GIS software “ArcView” will be used, and its applications, which were acquired for use in the study and will be delivered to the contractor, at the end of the inventories.
The study will start from the available Cartography, which includes Quadrants of the IGN at a scale of 1: 25,000, Aerial Photographs, Satellite Images existing in MARN and Satellite Images existing in other institutions. In addition, it will make use of new images that will be acquired during the inventory development: IRS panchromatic mode, 5 meters resolution with a 70 x 70 cm scale.
In relation to basic cartography, the use of topographic squares of El Salvador at a scale of 1: 25,000 is foreseen, in digital format, both planimetric and altimeter; The Vegetation Map of El Salvador, recently prepared by MARN, will be superimposed on this cartography in order to constitute the starting point for the typification of the forests.
To support field work and data verification, GPS technology with metric and submetric precision will be used to geo-reference satellite images, strata and sample plots.
The digital coverage will constitute what we call "Basic topographic cartography", and the base and thematic information generated will be superimposed on this basis for verification.
With this information, the required maps will be interpreted, digitized and elaborated, with a precision scale of 1: 25,000 at the departmental level and 1: 300,000 for the country.
Component No. 6 Sampling system.
In the original methodology presented by NHA, a preliminary pre-sampling stage is considered, which is unnecessary if a sampling model based on adequate classification and stratification is used. The costs of the pre-sampling stage would raise the general costs of the inventory phase, considering the time factor, having to carry out a pre-sampling stage for each stratum determined in the cabinet phase, the forestry measurement literature cites that the pre-sampling phase is They are carried out basically in forests of defined dimensions (duly delimited forests with a known area), as well as the methodological standards of national forest inventories, do not consider the pre-sampling stage as part of the stages to follow.
Preferably, NHA will focus on developing a cabinet stage to achieve an adequate stratification and a sampling system model appropriate to the national reality.
The stratification process consists of dividing a population into subpopulations or strata that are sampled independently, so that estimates can be made by stratum and an overall estimate of its parameters.
The strata can correspond to the different types of forest delimited on aerial photos and satellite images. Forest types can differentiate according to species composition, management or state of intervention, structure, state of development, etc.
The strata can be constituted based on ordinary cartography, by subdividing the forested area according to environmental variables, such as altitude, exposure, etc. The strata do not necessarily have to be continuous. A stratum can be made up of several homogeneous units (stands) of the same class, without having geographic continuity. The minimum size of the surfaces or stands that make up a stratum can vary between fractions of a hectare and hundreds of hectares, depending, among other factors, on the type of inventory system to be used.
In some occasions, the extensive surfaces are subdivided according to communal limits, property, etc. This process cannot properly be called stratification, but rather specification of independent inventory units.
According to PRETO (1992) the following criteria must be taken into account when adopting the stratification procedure:
- Strata delineation is only possible when the boundaries between stands are sharp; it is preferable to adopt relatively broad classes of clearly differential parameters in the images being interpreted and also in the field. Very detailed classifications can be ineffective, as forest dynamics makes the limits then only detectable in very recent images. On the other hand, the effect of stratification does not become much more efficient when establishing classes with very narrow ranges. It is possible to show that the best stratification variable is the same internal variable. Therefore, it is convenient to establish the strata based on parameters that, when combining their values, generate classes strongly related to the levels adopted by the variable of iteres in the population.
Stratification by classification of photoparcels, without delimitation, may in many cases be the best option, for several reasons:
- The photointerpretation in small plots is more exact than the delimitation of large areas, especially when the characteristics of the forest depend a lot on the environmental conditions (natural forest) and this is very varied.The variability of the parameters interpreted in photoplots is less than that of the parameters used in a conventional photointerpretation to delimit strata. A permanent network of photoplots allows better control of the evolution of the roof through repeated photointerpretations.
The effect of stratification is a reduction in the estimation error for the population, as a result of a lower variability between individuals within the strata, compared to the variance between individuals of the population without stratification. The smaller the variation within strata, the more efficient the stratified estimation. The way in which the units are distributed in the different strata also includes error reduction.
High access costs in national inventories often lead to a concentration of the field sample. The optimal degree of concentration depends, in part, on the structural characteristics of the forest and in part, on the relationship between the cost of access to a point in the area and the cost of measuring each unit (plot) in the vicinity of said point..
The most common stratified sampling procedure is the conglomerate, which consists of the division into blocks that constitute units of the first stage. In a sample of randomly selected blocks, a certain number of units (plots) that constitute the units of the second stage are located, also at random. Currently the blocks or units of the first stage are of equal size, as is the sample size of the second stage, in the selected blocks.
The principles of the systems are based on:
- The population is divided into L strata or subpopulations, whose boundary dimensions are exactly known, before distributing the sample. Being Yh the total stratum h, the population total is:
- In each stratum h, an independent sample of size nh is randomly distributed, and in each sample unit one or more attributes (yh, zh,…..) are valued. Based on the sample, the total population is estimated in each stratum (Yh) in such a way that in practice it can be considered unbiased, at least for practical purposes, and with a known variance SYh (or the estimation error), also unbiased.
We then have the variables Y1, Y2, …….Yh, ………..YL, all of them independent random variables, with their estimated variances:
The respective estimators of the methodology, start from the total population estimated unbiased as follows
Depending on the sample estimation procedure, it is possible to estimate other population characteristics, such as averages per sample unit or the global total per unit area.
Each estimator Yh has a variance V (Yh), the value of which depends on the sample design used. Since the variance of a sum of independent random variables is equal to the sum of their variances, the variance of the estimated population total is as follows:
The estimation error starts from:
Unbiased estimators of variance
An unbiased estimator of the estimation error of the total population is:
The mistrust limit is presumed that the total stratified population tends to be normally distributed, the confidential limits for a confidence level 1 - a.
The t value depends on the confidence level and the degrees of freedom. These in turn depend on whether the variance in the strata can be considered homogeneous or not. In general, n - L degrees of freedom are assumed.
Starting from the statistical analysis for the stratified method, the analysis is carried out at the conglomerate level. Considering a population (wooded area), it is divided into N blocks (within the stratum), all of equal size. Each block is integrated, in turn, by M units (parcels), also of constant size.
In a first stage n blocks are randomly selected. Within each selected block, m sample units (plots) are located. In each unit j of each block i, an attribute yij is valued.
The total population of y is as follows:
Where the population average per unit is:
The following is an unbiased estimate of the population average per unit of the second stage (plots)
Where
yi is the mean in the primary unit (block) i:
The total population is estimated, also unbiased, like this:
The total variance can be divided into two additive components:
Figure No. 2
"Stratified cluster sampling method"
Source: Own elaboration (Melgar, M. 2002)
Figure No. 3
"Flow of the INFES process"
Component No. 6 Preliminary tour
After the development of the different initial cabinet stages, a preliminary tour will be carried out with the inventory team, with the main objective of validating the typification, stratification and designed sampling system; as well as establishing the logistical and human resource needs necessary in Phase II.
The preliminary tour must not exceed 15 (fifteen) days, and must be properly planned to optimize the resources that will be invested in the process. It is important to consider that the preliminary tour is part of the standard methodology used by the TFAP and conventional inventories and ignoring it could limit actions that could improve the quality of the work to be developed.
