RESULTS AND DISCUSSION 



Desirable Features 



From the evaluation in table 2, it is apparent that 

 none of the surveyed systems fully implemented all 

 of the desirable features. This is not surprising be- 

 cause these features make up an ideal DBMS for 

 remeasured plots. 



Most of the systems could import and export data 

 (feature 1). There were a few that had trouble export- 

 ing data. All of the systems possessed a technique for 

 editing the data already in the data base (feature 2). 



Half of the systems did not meet the error-checking 

 criteria (feature 3). The criteria were purposely de- 

 signed to be stringent due to the importance we feel 

 this feature has to the quality of data. Most of these 

 systems failed to use previous measurements to check 

 the reasonableness of current measurements. This 

 error-checking technique is one of the most useful we 

 have found for detecting errors. Many of the systems 

 also did not check for unique tree numbers in each 

 plot, which becomes especially important for (1) larger 

 plots with many trees and (2) plots with a large num- 

 ber of new ingrowth trees. Only one of the systems 

 depended strictly on human error-checking as opposed 

 to machine-driven procedures. Though human error- 

 checking can be beneficial, we believe well-designed, 

 machine-coded procedures are more efficient and bet- 

 ter detectors of most errors. 



Storage of large amounts of data (feature 4) was 

 limited only by hardware constraints (a + rating) for 

 most of the systems. The DBMS software should not 

 limit the amount of data that is potentially storable. 



Most of the organizations have experience collecting 

 remeasurement data (feature 5). The criterion most 

 commonly failed was the inability to merge remeasure- 

 ment data with existing data. This is a very necessary 

 procedure because it is vital to assessing changes in 

 plot summary and tree attributes over time, probably 

 one of the main reasons for using remeasured plots. 



All of the systems had a procedure for safe backup 

 of the data base (feature 6). And most adequately 

 protected the integrity of their data base by assuring 

 that only approved persons have direct access to the 

 data and that safe procedures are used to enter new 

 data into the data base. Recording changes made to 

 the data base presented a major problem for many of 

 the systems. Only a few had a way of automatically 

 recording changes; most relied on the person making 

 the change to document it. Automating the procedure 

 guarantees that a listing of changes is maintained. 



We suspect that most of the systems were designed 

 to meet the data needs for a specific inventory or re- 

 search study. No place is this more apparent than 

 with feature 7, the recording of commonly measured 

 plot and tree attributes. Most provided the capability 

 to record some data for at least six measurements, 



but many had difficulty when it came to fully describ- 

 ing sampling designs and common measurements. 

 Many of the systems were adequate for one specific 

 design, namely their present inventory or research 

 study design, but were of limited use for describing 

 other designs. Many could not record information at 

 the subplot level, the hierarchical level commonly used 

 for collecting small tree measurements (see Byrne 

 and Stage [1988] for a detailed discussion of sampling 

 designs). Of the 27 systems that failed the criteria on 

 storage of common plot and tree characteristics, most 

 of them were not able to record comments (for example, 

 a unique feature about a plot or tree that existing 

 data fields do not address). More than one-third could 

 not describe plot location and/or types of stand distur- 

 bance. The inability to geographically reference plot 

 location seems to be a major omission. 



Nearly half of the systems do not have the capabil- 

 ity to add new data fields (feature 8). Because data 

 needs will often change over the time that remeasured 

 plots are monitored, the ability to add new data fields 

 is a real advantage, possibly preventing a major re- 

 programming effort in the future. 



Selective retrieval of data from the data base (fea- 

 ture 9) is a necessary part of the DBMS when analy- 

 sis of the data is required. Many of the systems have 

 retrieval capabilities, most often with plot-level data 

 fields as search criteria (for example, retrieving data 

 from all plots with a site index greater than 70, etc.). 

 Using tree-level data fields as search criteria was not 

 possible for many. An example of such a search would 

 be looking for all Douglas-fir (Pseudotsuga menziesii) 

 trees greater than 5 inches d.b.h. growing on south- 

 facing slopes. Tree-level searches become more impor- 

 tant when the remeasured plots are used as a basis 

 for improving an individual tree growth and yield 

 model. 



The next feature (10), accessibility of the data base 

 software at a reasonable cost, was difficult to evaluate 

 for many systems because of the lack of information 

 provided in the questionnaire (* in table 1). Cost of 

 the DBMS software was difficult for many to provide, 

 especially because many of the organizations had de- 

 veloped their DBMS software in-house. Some of the 

 organizations were unsure whether their DBMS soft- 

 ware was proprietary. For the systems for which cost 

 and proprietary nature were known, the main reason 

 for failing the criteria was expense of DBMS software. 



Adequate documentation and help facilities (feature 

 11) are present on only about one-third of the systems 

 (defined by a user's manual plus one of the following: 

 interactive help, user-support services, sample data 

 set, or tutorial). In addition, more than half of the 

 systems require a person that is experienced or expert 

 in computer knowledge to use the system. Because of 

 these two attributes it would probably be difficult for 

 a new user to quickly learn how to use many of these 



9 



