International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004 | In
equals 0.5 during low tide, and further follows a sine form, ie. Positiona| TIR EE T | 3
ned() = 0.75+0.25-cos(2m1/12.5), with / expressed in hours in Accuracy Accuracy Compl. ,
relation to high tide. Finally, the function nb(f) is included to Objects el Ged. CC QAA De AIM | W
Na x aii ht = VENT ; oie E E eigen rigen EE epic creed ti EE 3 EE m
describe m SCR EL elev ES met Fa a pr C id 486m 303m 86.796 < | year =
we correct the slopes of the membership functions derived [rot W : s
6 cotrect fhe slopes af fio ineimbervim rinetiaris den ed (rom Opi, NR 5h30 86.7% < 1 year el
equations (1) and (2) with a correction factor related to the : x se |
; Pallet Lex Pis rap at Dai BD.id 48.6 m 30.3 m 86.7% <1 year Wi
temporal (un)certainty of the vegetation- and wetness index, as BS id 48.6 m 303 SG TA
described above. Zod AIO MSL. M 56./%0 ^ ! yeat | in
Attributes — 1 | his
2.4 Quality Elements and Quality Matrix C.vol 48.6 m 30.3 m mv(i,t) £0.28 m 86.776 nb(t) ust
: C.vol90 48.6 m 30.3 m mv(i,t) NR nb(t) su]
2.4.1. Quality elements: For the spatial uncertainty of the CL.geo NR 30.3 m 86.7%
beach compartments, we encounter the following ISO quality BD.geo 48.6 m 30.3 m mv(i,t) 86.7% In
elements and subelements as most essential (ISO 2003): BD.ndvi 48.6 m 30.3 m NR 86.7% nv(t) ele
o Positional accuracy : + BD.z NR 30.3 m + 0.28 m 86.7% be
Relative or internal: closeness of the relative positions BS 2e0 48.6 m 303 m mv(i,t) 86.794 ser
of objects in a dataset to their respective relative e e : ge cor
positions accepted as or being true. 5m a m Je m NR 28 inh i the
d rd Tar S.z 30.3 n .28 m 86.
Gridded data position: closeness of gridded data E 86.0303 ; c d Ti ;
position values to values accepted as or being true. et RE IR mv() 20.7% nol), 3.2
o Thematic accuracy Processes d uer t
Accuracy of quantitative attributes: the correctness of . lr voli «10 year *
quantitative attributes and of the classifications of PL id < I year | 0 |
objects and their relationships. TL.trend < | year
Classification correctness: comparison of the classes ec
assigned to objects or their attributes to a universe of Ron
discourse (e.g. ground truth or reference dataset). Table 1. Quality elements for the ontological features for 1995. Or
o Completeness Abbreviations: CC = classification correctness, QAA ha ;
Data completeness: the commission and omission of = quantitative attribute accuracy, ATM = accuracy of d
: z ist
datasets. time measurement, NR = not relevant. fan
For the temporal uncertainty of the compartments, we recognize — — v Positional 7 [hematic Temp. an
or different time scales: Cid C.vol Compl. com
for different time scales: Accuracy Accuracy p CC. plat
o Temporal accuracy ret ener in puta TC. " sen D: ATM : à
a ti 3 | at | His
Accuracy of a time measurement: correctness of the aiid gle peu REL. He. ee Qi fata P B su
al 'efer AC X € ifo “a "ti o f star 1 A 0% 0 0 ace
temporal references of an item (reporting ol error in # m3 m m % m % % :
; erre visu
time measurement). 260 1.02E+04 486 30.3 0212 028 0.133 0.326 E
; : : 280 6.87E+03 48.6 30.3 0.185 0.28 0.133 0.331 para
2.4.2. Quality matrix: By applying an ontological approach, 300 2.01E+03 486 30.3 0.169 0.28 0.133 0.275 - l
we can construct a quality matrix, whereby ontological features 301 482E+02 486 30.3 0,123 028 0.133 0.235 b y
j s 82E+02 6 30.3 0.123 0.28 0.135 0.25 ec
as objects, attributes relationships, processes and events are 302 2.56E+03 48.6 30.3 0.253 0.28 0.133 0.324 4
projected against quality elements, as described above. Table ! 303 3.18E+03 486 303 0.167 0.28 0.133 0.313
describes the quality of objects, attributes and processes in a 304 5.33E+03 486 303 0.198 0.28 0.133 0.290 The
general fashion that applies to the case study. Different 320 | Ue |
membership functions occur, whereas spatial and temporal d
accuracy apply to a limited set of objects. Table 2. Amount of beach volume for each compartment each
: : es : (C.vol) and its quality elements for 1993. each
The prominent feature of interest i$ the amount of beach Abbreviations: C.id = compartment id., C vol = beach i
volume, represented by C.vol in table 1. Within one year, C.vol volume per compartment, CC = classification Fo
can be calculated for each compartment, as well as its quality correctness, QAA = quantitative attribute accuracy, son
parameters (see table 2). ATM = accuracy of time measurement, NR = not value
relevant. exam
Furth
3. VISUALIZATION OF THE QUALITY MATRIX Here, we discuss some aspects to fulfil the aim to construct à
; ah ; : prototype illustrating the qualit elements involved in a beach ;
In the beach nourishment application, we can easily depict a I YP aüng nequa y : ain R iv 33
. T nourishment process. First, it should incorporate interactivity
map with beach compartments suitable for nourishment. t TU. M AS 7
Eau between separate windows, re. it should have dynamically 0 d
However, trends and associations between compartments, ; ; * sa le hie =
a d live] sr volved. in she decision linked views. Second, the prototype should handle ^g visual
changes n ume and qua ine ements uo reni li e eee j dimensionality of the attributes, using multivariate visualization coast]
making, Bro mole complicated sou ve As qualite Sn tools. Last, it should be able to deal with multi temporal variat
| ar rariate pose — ie. there are many qualı : : M
I oe multivar e m purae i N are Dey datasets and to detect trends In the beach nourishment process There
ale $5 - each ontological feature — ari : : ;
elements sue ied for igacn ontoioe s de and its quality elements during multiple time observations. order
visualization tools are the most appropriate display technique. sont à
Furthermore, to detect trends and the evolution of the quality seit
elements in time, a temporal element should be included in the com
visualization tool. p.
1192