The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B4. Beijing 2008 
46 
language such as XQquery, XPath but not conversion to SQL or 
OQL (Object Query language); ©provide rich development 
interfaces. Most native XML database support XML:DB API of 
XML:DB.org development. 
GML database as a tool for managing XML data with special 
format, its research could be advanced on the basis of current 
XML database technology combined with characters of GML. 
Combined with characters of native XML database and GML 
spatial data, this paper designed a more effective native XML 
database based GML spatial data storage model-G2NXDBM, 
develop and implement GML data storage prototype system, 
directly access GML data in native XML database. This paper 
also evaluates this storage model on aspects of storage space 
and query efficiency. 
2. NATIVE XML STORAGE BASED GML STORAGE 
MODEL- G2NXDBM 
2.1 Architecture of G2NXDBM 
2.1.1 GML spatial data storage mechanism 
The main approach of native XML database based GML 
data storage is: firstly analyses structure relationships between 
features according to GML application schema, then map 
structure relationship to collection of native XML database, and 
establish collection in database to reflect document structure. 
Figure 1. Storage Mechanism of GML Spatial Data 
Assume that every GML document corresponds to at least one 
specific GML schema, this schema file has existed in system or 
provided by the owner of the GML document. When import a 
specific schema, system would generate a one to one mapping 
rule between schema and storage structure, and establish 
corresponding collection. Thus, when read a GML document, 
system would classify document by schema. And partition this 
document by unit of feature, then store the partitioned 
document in corresponding data collection according to 
mapping rule. 
2.1.2 GML Spatial data storage structure 
In this storage model, when GML data is stored in native XML 
database, the storage is implemented by three types of files, 
(1) GML fragment file; (2) GML schema file; (3) GML 
storage log file. GML fragment file is the fragments after 
partition imported GML document according to certain 
granularity. GML schema file is used to define content, 
structure and constrain rule of GML document. It could be used 
to conjecture storage mapping rule between GML document 
and native XML database and testify validity of GML 
document. GML storage log file mainly save partitioned GML 
document name, GML fragment storage position, fragment 
order and metadata about stored GML document such as 
storage time, storage person and so on. 
Two root collections are defined in database, one is data 
collection, and another is management collection. Data 
collection is used for store GML schema and log file. In data 
collection, document belong to different schema correspond to 
different document collection. In the same document collection, 
the GML fragment file with the same type would be stored in 
the same feature collection. Also, the management collection is 
also stored according to classified schema. Its storage structure 
is shown as figure 2. 
Native 
Database 
Figure 2. Storage Structure of GML Spatial Data GML 
Native XML database based GML storage model has characters 
as follows: 
(1) Partitioning storage granularity of GML document 
according to feature type not only has certain semantic 
repetition but also enhance storage efficiency. 
(2) Record nested relationship between features by 
hierarchy of collections in native XML database. 
(3) One of the most important characters of Native XML 
database is providing “out and home access” function, thus, 
storing GML document in native XML database can retake the 
“same” document. 
(4) Store document with the same feature type in the same 
collection can accelerate query speed. Because querying on one 
collection would be faster than multi-collections. 
(5) saving partition information and storage information in 
log file can facilitate later operations such as information access 
and document restore. 
2.2 Storage granularity of G2NXDBM 
Current native XML database could be classified to three types 
according to its storage granularity: element based, sub-tree 
based and document based. Lore[J. McHugh ,1997] and 
TIMBER[H. V. Jagadish,2002] adopt element based storage 
strategy, each element is the minimal storage unit; Natix[T. 
Fiebig,2002] adopt sub-tree based storage strategy, partition 
XML document into a series of sub-trees according to physic 
storage pages size, take each sub-tree as storage record; Xindice 
adopt document based storage strategy, take a XML document 
as storage record. 
Two types of data stored in GML database system: one is GML 
schema file; another is GML document instance. GML schema 
determines storage structure of data, and document instance 
contain the true data content in system. GML schema defines 
content and structure of GML document, which could be 
adopted as the standard of GML documents classify and basis 
of storage granularity selecting. GML adopt feature to describe 
geographic entries and geographic phenomenon, data in GML 
document is organized by units of feature. Thus determine 
storage granularity of GML document model according to 
feature type is appropriate.