Loading ...
Sorry, an error occurred while loading the content.

Re: GEOSTATS: Who is going to do the best map ?

Expand Messages
  • Toshifumi Matsuoka
    Hi I have a question for these discussions. ... Do we need to define best in this case? Cross Validation technique (Chapter 15, Isaaks and Srivastava s
    Message 1 of 12 , Feb 19, 1997
    View Source
    • 0 Attachment
      Hi

      I have a question for these discussions.

      At 3:37 PM 1997.19.02, Chuck Ehlschlaeger wrote:
      >
      > I agree "best" needs to be defined. Is it simply a matter of which
      > model gets values closest to the missing values?

      Do we need to define "best" in this case? Cross Validation technique
      (Chapter 15, Isaaks and Srivastava's book) can tell us the estimation
      errors for different estimation methods. I think this concept can be
      applied to compare different estimation algorithms.

      Do I misunderstand the cross validation?

      Toshi

      ***********************************************
      Dr. Toshifumi Matsuoka
      Geophysical Department Japex, Tokyo Japan
      Tel : 81-3-5461-7324 Fax : 81-3-5461-7397
      E-mail : toshi@...
      ***********************************************


      --
      *To post a message to the list, send it to ai-geostats@....
      *As a general service to list users, please remember to post a summary
      of any useful responses to your questions.
      *To unsubscribe, send email to majordomo@... with no subject and
      "unsubscribe ai-geostats" in the message body.
      DO NOT SEND Subscribe/Unsubscribe requests to the list!
    • Agustin Lobo
      ... The problem would be that there will be different a best for each family of functions used to create the artificial surface- On the other hand, I think
      Message 2 of 12 , Feb 20, 1997
      View Source
      • 0 Attachment
        On Wed, 19 Feb 1997, joseph loon wrote:

        > We will need to define "best"!
        > Here is a variation on this theme:
        > Define a synthetic surface Z=f(X,Y). This surface can be artificially
        > manufactured or ( to closely resemble some known surface) it could be
        > defined by a double Fourier series, etc.
        > Select a sample ( random, systematic,stratified, etc) then use a number of
        > different interpolation methods to predict Z-values at certain X,Y
        > locations. THESE Z-VALUES ARE ALREADY DEFINED BY THE SYNTHETIC SURFACE!
        > therefore absolute errors are obtained. You will find that the character
        > of the original surface (how can this be determined?) will determine which
        > interpolation method will give the most accurate results. In other words,
        > a sophisticated interpolation method will give accurate interpolations on
        > a complex surface, but may give bad results on a fairly smooth surface.
        > If anyone is interested in looking into this please contact me.
        > Joe


        The problem would be that there will be different a "best" for each family
        of functions used to create the artificial surface-


        On the other hand, I think that interpolation of rainfall needs of
        a deterministic-based approach, which would take into account
        not only absolute distances, but factors as dominant wind, distance to
        the coast, even month of the year etc.

        Agus


        Dr. Agustin Lobo
        Instituto de Ciencias de la Tierra (CSIC)
        Lluis Sole Sabaris s/n
        08028 Barcelona SPAIN
        tel 34 3 330 2716
        fax 34 3 411 0012
        alobo@...
        alobo@...




        --
        *To post a message to the list, send it to ai-geostats@....
        *As a general service to list users, please remember to post a summary
        of any useful responses to your questions.
        *To unsubscribe, send email to majordomo@... with no subject and
        "unsubscribe ai-geostats" in the message body.
        DO NOT SEND Subscribe/Unsubscribe requests to the list!
      • Åsa Danielsson
        Hi! When I read the mail, from Gregoire Dubois, I recalled a paper I read some time ago. It presents something similair to what is discussed here. They had a
        Message 3 of 12 , Feb 20, 1997
        View Source
        • 0 Attachment
          Hi!

          When I read the mail, from Gregoire Dubois, I recalled a paper I read some
          time ago. It presents something similair to what is discussed here. They had
          a data set, sent it to several people who should do a geostatistical
          analysis on the data and then they compared the different results to see
          which of the methods that performed the best. For you that are interested
          the paper is:

          Englund, E., A Variance of Geostatisticians, Mathematical Geology 22(4):417-455.

          Greetings,

          Åsa



          ____________________________________________________________________________
          Åsa Danielsson Linkoping University Phone:+46 13 282922
          __ Dept.of Water & Env. Studies Fax: +46 13 133630
          __ /##\ 581 83 Linkoping
          /o \/####\_ Sweden
          // \\
          E-mail:asada@...

          ______________________________________________________________________________

          --
          *To post a message to the list, send it to ai-geostats@....
          *As a general service to list users, please remember to post a summary
          of any useful responses to your questions.
          *To unsubscribe, send email to majordomo@... with no subject and
          "unsubscribe ai-geostats" in the message body.
          DO NOT SEND Subscribe/Unsubscribe requests to the list!
        • Craig Johns
          ... It seems clear that Best in one science will be less than Best for another. Why not have multiple (sensible) criterion for Best ? Then, the GIS
          Message 4 of 12 , Feb 20, 1997
          View Source
          • 0 Attachment
            On Wed, 19 Feb 1997, Jeff Wolfe wrote:
            > Date: Wed, 19 Feb 1997 19:32:23 -0500
            > From: Jeff Wolfe <wolfe@...>
            > To: ai-geostats@...
            > Subject: Re: GEOSTATS: Who is going to do the best map ?
            >
            > In message <330B8EB6.ABD@...>, Chuck Ehlschlaeger writes:
            >
            > > I agree "best" needs to be defined. Is it simply a matter of which...

            > If I may add a few cents to the pot as well:
            >
            > Assuming we can come up with a reasonable approximation of "best", one tack to take with this project would be to try and develop some procedures/algorithms for automated selection of the "best" method, it strikes me that this would be useful in GIS and other computer mapping applications.
            >
            >
            > Just a thought..
            >
            > -Jeff
            >


            It seems clear that "Best" in one science will be less than "Best" for
            another. Why not have multiple (sensible) criterion for "Best"? Then,
            the GIS algorithm suggested elsewhere would depend on how the user
            defines best.


            -- Craig Johns
            351 Kerr (752-3687)
            Division of Statistics
            UC Davis

            --
            *To post a message to the list, send it to ai-geostats@....
            *As a general service to list users, please remember to post a summary
            of any useful responses to your questions.
            *To unsubscribe, send email to majordomo@... with no subject and
            "unsubscribe ai-geostats" in the message body.
            DO NOT SEND Subscribe/Unsubscribe requests to the list!
          • Mohammad J Abedini
            Greetings, Neat idea, When I noticed G. Dubois s recent posting on Who is going to do the best map? I recalled a paper I read a while ago entitled: Naef, F.,
            Message 5 of 12 , Feb 20, 1997
            View Source
            • 0 Attachment
              Greetings,

              Neat idea, When I noticed G. Dubois's recent posting on "Who is going to
              do the best map? I recalled a paper I read a while ago entitled:

              Naef, F., (1981). "Can we model the rainfall-runoff process today"
              Hydrological science Bulletin, 26(3):281-289.

              In that paper, the author was referring to a similar project being
              organized by WMO(1975) entitled:

              WMO (1975). "Intercomparison of conceptual models used in Operational
              Hydrological Forecasting. WMO Operational Hydrological Forecasting Report
              no. 7.

              Part of the first paper reads: "In that project, several model owners
              calibrated their models using data sets supplied by WMO. For a period of 2
              years, the model owners had to simulate the discharge without knowing the
              measured discharge. These simulations had to be sent to WMO, were then
              evaluated by VAW in Zurich, and finally judged by a team of experts. .."

              In order not to repeat some of the pitfalls of the similar project, I do
              suggest to go through the cited document.

              One thing that I want to emphasize and know that it was also acknowledged
              by other repected member is that we need to put the project into a
              scientific framework such as

              Research Hypothesis: blah blah ...

              Objective: blah blah ...

              Assumptions: blah blah ....

              Approach: blah blah .....

              As for my own viewpoint of the project, I do suggest everybody to read the
              following paper:

              Oreskes, et. al. (1994). "Verification, Validation, and Confirmation of
              Numerical Models in the Earth Sciences", Science, Vol. 263, pp. 641-646.

              Hope this helps paving the road for the project.

              Best of luck.
              Abedini
              \|||/
              (o o)
              +=================oOO==(_)==OOo=================+
              |_____|_____|_____|_____|_____|_____|_____|_____|
              |__|_____|_____|_____|_____|_____|_____|__|__|__|
              |_____| |_____|
              |__|__| Mohammad J. Abedini |__|__|
              |_____| School of Engineering |_____|
              |_____| University of Guelph |_____|
              |_____| Guelph Ont, N1G 2W1 |_____|
              |__|__| Tel.:(519) 824-4120 ext. 4321 (W) |__|__|
              |__|__| Tel.:(519) 821-1199 (H) |__|__|
              |_____| Fax :(519) 836-0227 |_____|
              |_____| e-mail: mabedini@... |_____|
              |__|__|___________________________________|__|__|
              |_____|_____|_____|_____|_____|_____|_____|_____|
              +===============================================+
              The world is your exercise-book, the pages on which you do your sums.
              You are also free to write nonsense, or lies, or to tear the pages.

              Richard Bach






              --
              *To post a message to the list, send it to ai-geostats@....
              *As a general service to list users, please remember to post a summary
              of any useful responses to your questions.
              *To unsubscribe, send email to majordomo@... with no subject and
              "unsubscribe ai-geostats" in the message body.
              DO NOT SEND Subscribe/Unsubscribe requests to the list!
            • Mohammad J Abedini
              In my recent posting, I did not comment on the meaning of the best to let you read the suggested paper in Sceince Journal. As a PhD student, I have
              Message 6 of 12 , Feb 20, 1997
              View Source
              • 0 Attachment
                In my recent posting, I did not comment on the meaning of "the best" to
                let you read the suggested paper in Sceince Journal.

                As a PhD student, I have conceptual difficulty defining the meaning of
                "the best". To me, "the best" is objective dependent. Let me elaborate a
                little bit more on this point. Everybody will agree that precipitation is
                the driving mechanism for almost all processes being in effect in a
                typical watershed. Of course, the rainfall field being interpolated or
                whatever, should have an end result. By interpolating rainfall, do we want
                to find the optimal spatial location of ground rain gauges or do we want
                to find optimal temporal resolution of datalogger to measure rainfall or
                do we want to have a best match in runoff at the mouth of the watershed or
                finally do we want to delineate the source of contaminat (say sediment)
                throughtout the waterhsed. Based on my experience working in the same
                field, I am quite sure that "the best" will be dramatically different
                depending which objective is in our mind.

                Perhaps, someone out there might want to shed some light on me because
                these are something that I am struggling with on a continous basis.

                Thanks
                Abedini


                On Thu, 20 Feb 1997, Mohammad J Abedini wrote:

                > Greetings,
                >
                > Neat idea, When I noticed G. Dubois's recent posting on "Who is going to
                > do the best map? I recalled a paper I read a while ago entitled:
                >
                > Naef, F., (1981). "Can we model the rainfall-runoff process today"
                > Hydrological science Bulletin, 26(3):281-289.
                >
                > In that paper, the author was referring to a similar project being
                > organized by WMO(1975) entitled:
                >
                > WMO (1975). "Intercomparison of conceptual models used in Operational
                > Hydrological Forecasting. WMO Operational Hydrological Forecasting Report
                > no. 7.
                >
                > Part of the first paper reads: "In that project, several model owners
                > calibrated their models using data sets supplied by WMO. For a period of 2
                > years, the model owners had to simulate the discharge without knowing the
                > measured discharge. These simulations had to be sent to WMO, were then
                > evaluated by VAW in Zurich, and finally judged by a team of experts. .."
                >
                > In order not to repeat some of the pitfalls of the similar project, I do
                > suggest to go through the cited document.
                >
                > One thing that I want to emphasize and know that it was also acknowledged
                > by other repected member is that we need to put the project into a
                > scientific framework such as
                >
                > Research Hypothesis: blah blah ...
                >
                > Objective: blah blah ...
                >
                > Assumptions: blah blah ....
                >
                > Approach: blah blah .....
                >
                > As for my own viewpoint of the project, I do suggest everybody to read the
                > following paper:
                >
                > Oreskes, et. al. (1994). "Verification, Validation, and Confirmation of
                > Numerical Models in the Earth Sciences", Science, Vol. 263, pp. 641-646.
                >
                > Hope this helps paving the road for the project.
                >
                > Best of luck.
                > Abedini
                > \|||/
                > (o o)
                > +=================oOO==(_)==OOo=================+
                > |_____|_____|_____|_____|_____|_____|_____|_____|
                > |__|_____|_____|_____|_____|_____|_____|__|__|__|
                > |_____| |_____|
                > |__|__| Mohammad J. Abedini |__|__|
                > |_____| School of Engineering |_____|
                > |_____| University of Guelph |_____|
                > |_____| Guelph Ont, N1G 2W1 |_____|
                > |__|__| Tel.:(519) 824-4120 ext. 4321 (W) |__|__|
                > |__|__| Tel.:(519) 821-1199 (H) |__|__|
                > |_____| Fax :(519) 836-0227 |_____|
                > |_____| e-mail: mabedini@... |_____|
                > |__|__|___________________________________|__|__|
                > |_____|_____|_____|_____|_____|_____|_____|_____|
                > +===============================================+
                > The world is your exercise-book, the pages on which you do your sums.
                > You are also free to write nonsense, or lies, or to tear the pages.
                >
                > Richard Bach
                >
                >
                >
                >
                >
                >
                > --
                > *To post a message to the list, send it to ai-geostats@....
                > *As a general service to list users, please remember to post a summary
                > of any useful responses to your questions.
                > *To unsubscribe, send email to majordomo@... with no subject and
                > "unsubscribe ai-geostats" in the message body.
                > DO NOT SEND Subscribe/Unsubscribe requests to the list!
                >

                --
                *To post a message to the list, send it to ai-geostats@....
                *As a general service to list users, please remember to post a summary
                of any useful responses to your questions.
                *To unsubscribe, send email to majordomo@... with no subject and
                "unsubscribe ai-geostats" in the message body.
                DO NOT SEND Subscribe/Unsubscribe requests to the list!
              • krivoruchko@gray.isir.minsk.by
                Dear colleagues, First of all, the rainfall data linked with the major accident in the human history - the destroy of Chernobyl NPP. Rainfalls are directly
                Message 7 of 12 , Feb 21, 1997
                View Source
                • 0 Attachment
                  Dear colleagues,

                  First of all, the rainfall data linked with the major accident in the human
                  history - the destroy of Chernobyl NPP. Rainfalls are directly correlated
                  with dose loads on population in the very first days after the accident,
                  which leads to enormous growing of the different diseases in the southern
                  Belarus and northern Ukraine. I know it, because I'm processing morbidity
                  data with geostatistics, ANN and others with our GIS MapStudio since last
                  year.
                  Of course, there are uncertainties and difficulties in interpolation. For
                  example, some radioactive clouds were deposited with bromium silver on the
                  earth for preventing their propagation to Moscow and Kiev.
                  I think that the way to select the best result is to find the 25%
                  remainder of the data with the best accuracy without taking into account
                  any mathematical tricks.
                  I am agree to enter this work, if only some contingency will not stand in
                  the way.

                  Bye, Konstantin.

                  ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
                  Konstantin Krivoruchko, Ph.D.
                  Head, GIS Laboratory
                  International Sakharov Institute of Radioecology
                  220009, Belarus, Minsk, Dolgobrodskaya 23
                  Email: krivoruchko@...
                  Tel: 375 172 306 292
                  Fax: 375 172 306 897
                  ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~


                  --
                  *To post a message to the list, send it to ai-geostats@....
                  *As a general service to list users, please remember to post a summary
                  of any useful responses to your questions.
                  *To unsubscribe, send email to majordomo@... with no subject and
                  "unsubscribe ai-geostats" in the message body.
                  DO NOT SEND Subscribe/Unsubscribe requests to the list!
                • Joanne Halls
                  Gregoire, Count me in as a participant in the geostats test for predicting the missing 25%. Joanne Halls Joanne Halls, Ph.D. GIS Manager Research Planning Inc.
                  Message 8 of 12 , Feb 21, 1997
                  View Source
                  • 0 Attachment
                    Gregoire,
                    Count me in as a participant in the geostats test for predicting the
                    missing 25%.

                    Joanne Halls

                    Joanne Halls, Ph.D.
                    GIS Manager
                    Research Planning Inc.
                    1200 Park Street
                    Columbia, South Carolina 29202
                    Phone: (803) 256-7322
                    Fax: (803) 254-6445
                    Email: joanne@...


                    --
                    *To post a message to the list, send it to ai-geostats@....
                    *As a general service to list users, please remember to post a summary
                    of any useful responses to your questions.
                    *To unsubscribe, send email to majordomo@... with no subject and
                    "unsubscribe ai-geostats" in the message body.
                    DO NOT SEND Subscribe/Unsubscribe requests to the list!
                  Your message has been successfully submitted and would be delivered to recipients shortly.