Page 101 - Modern Spatiotemporal Geostatistics
P. 101

82      Modern Spatiotemporal Geostatistics —   Chapter 3

         There  are several  ways to  incorporate  such  knowledge into  BME analysis.  As-
         suming that  the  univariate  pdf  is the  same for  all  points  in space/time,  a sim-
         ple approach is to  consider a transformation  of  the  original  random field  into
         a  Gaussian  field  with  known  mean  and  covariance functions,  which  brings  us
         back to the earlier section on general knowledge in terms of statistical  moments
         (p.  75).
             In some other  circumstances, while the  multivariate  pdf  } ?  is unknown, an
         equation  that  describes the  evolution  of  a lower  level  pdf  may be derived from
         physical  or  mathematical  considerations  (e.g.,  a diffusion-type  equation).  In
         order  to  incorporate this  knowledge  into  BME  analysis,  one  may select  a set
        of g a  functions  so that  the  stochastic  expectations  <^  with  respect to  /§ are
         calculable (e.g.,  with the  help  of the evolution  equation of/ §).  The statistical
        functions  Tfe  are now tne  general  knowledge,  i.e.,  the original  situation  again
         has  been converted to  that of the  earlier section (p.  75).  It  may be  instructive
        to  demonstrate  the  approach in terms of  a very simple example.

         EXAMPLE  3.11:  Suppose that the  prior  pdf / ? satisfies the  equation




        X  €  [a,  6], with  initial  condition  f s(x',  0)  =  <Kx)-  We  cnoose  9o  =  1 (as
                                                 n
         usual)  and let gi(x)  De  a known function  of x  ° \y, with  <7i(0)  =  CQ.  We can
        then  write





         In light of the initial condition,  Equation  3.28 gives g\ (t) =  CQ  exp[/3t].

             Finally,  in  some applications,  useful physical knowledge  may be expressed
         in  the  form  of  integral  equations.  This  is the  case,  e.g.,  of  perturbation  ap-
        proximations (ordinary or diagrammatic) to  groundwater flow and solute trans-
         port  problems  (for  a  detailed  study  of  such  perturbations  see Christakos and
         Hristopulos,  1998).


        The    Specificatory     Knowledge       Base
        The  specificatory  or  case-specific  knowledge  base  S  is  knowledge  about  the
        specific situation.  For example, the  singular statements, "the  litmus  paper that
         Mr.  Nicolaou  gave me turned  red when immersed in the  liquid"  and "the  distri-
         bution of the  West Lyons field  porosity  data is shown in  Figure  7.4,"  constitute
        specificatory  knowledge.  Unlike  general  knowledge,  the  specificatory  knowl-
        edge  base  S  refers to  a particular  occurrence or state  of  affairs at  a  particular
         location  and  at  a  particular  time.  The  S  includes  both  external  or  demon-
        strative evidence (actual  measurements,  perceptual or data of  sense,  etc.)  and
   96   97   98   99   100   101   102   103   104   105   106