Changes between Version 9 and Version 10 of udg/ecoms/RPackage/examples/drift


Ignore:
Timestamp:
May 15, 2015 2:55:17 PM (7 years ago)
Author:
juaco
Comment:

--

Legend:

Unmodified
Added
Removed
Modified

  • udg/ecoms/RPackage/examples/drift

    v9 v10  
    11= Analysing model drift in South-western Iberia
    22
    3 In this practice, we will analyse the model drift by using the forecast of daily mean (`time = "DD"`) surface temperature (`var = "tas"`) for July (`season = 7`) 2006 (`years = 2006`) considering 6 different forecast (lead) months, from January to June. The lead month 0 (i.e., the initialization of July itself) will be the reference from which we will compute the anomalies (`leadMonth = 0`). In this example, we will consider the first member (`members = 1`) of the CFSv2 hindcast as reference:
     3In this practice, we will analyse the model drift by using the forecast of daily mean (`time = "DD"`, `aggr.d = "mean"`) surface temperature (`var = "tas"`) for July (`season = 7`) 2006 (`years = 2006`) considering 6 different forecast (lead) months, from January to June. The lead month 0 (i.e., the initialization of July itself) will be the reference from which we will compute the anomalies (`leadMonth = 0`). In this example, we will consider the first member (`members = 1`) of the CFSv2 hindcast as reference:
    44
    55
    66{{{#!text/R
    7 ref <- loadECOMS(dataset = "CFSv2_seasonal_16",
     7ref <- loadECOMS(dataset = "CFSv2_seasonal",
    88                 var = "tas",
    99                 members = 1,
    … …  
    1313                 years = 2006,
    1414                 leadMonth = 0,
     15                 time = "DD",
     16                 aggr.d = "mean")
    1517                 time = "DD")
    1618}}}
    1719
    18 Messages on-screen inform about the loading process. Note that the selection of `leadMonth = 0` will give a NOTE on screen:
     20Messages on-screen inform about the loading process. Note that the selection of `leadMonth = 0` will give a NOTE:
    1921
    2022{{{
    21 [2014-09-03 17:50:20] Defining homogeneization parameters for variable "tas"
    22 NOTE: daily mean will be calculated from the 6-h model output
     23[2015-05-15 14:50:46] Defining homogeneization parameters for variable "tas"
    2324NOTE: 'leadMonth = 0' selected
    24 [2014-09-03 17:50:20] Defining geo-location parameters
    25 [2014-09-03 17:50:20] Defining initialization time parameters
    26 [2014-09-03 17:50:26] Retrieving data subset ...
    27 [2014-09-03 17:50:31] Done
     25[2015-05-15 14:50:46] Defining geo-location parameters
     26[2015-05-15 14:50:46] Defining initialization time parameters
     27NOTE: Daily aggregation will be computed from 6-hourly data
     28[2015-05-15 14:50:49] Retrieving data subset ...
     29[2015-05-15 14:50:56] Done
    2830}}}
    2931
    … …  
    4345{{{#!text/R
    4446cfs.list <- lapply(1:6, function(lead.month) {
    45       loadECOMS(dataset = "CFSv2_seasonal_16",
     47      loadECOMS(dataset = "CFSv2_seasonal",
    4648                var = "tas",
    4749                members = 1,
    … …  
    5153                years = 2006,
    5254                leadMonth = lead.month,
    53                 time = "DD")
     55                time = "DD",
     56                aggr.d = "mean")
    5457      }
    5558)
    … …  
    110113[[Image(image-20150130-185813.png)]]
    111114
    112 
    113115The results show how a increasing lead month leads to a negative bias of the forecast, demonstrating that the mean state of a variable of a forecast is not stationary along the runtime dimension.
    114116