Revision a5ab37e3
Added by Benoit Parmentier about 11 years ago
| climate/research/oregon/interpolation/master_script_temp.R | ||
|---|---|---|
| 10 | 10 |
#STAGE 5: Output analyses: assessment of results for specific dates... |
| 11 | 11 |
# |
| 12 | 12 |
#AUTHOR: Benoit Parmentier |
| 13 |
#DATE: 07/18/2013
|
|
| 13 |
#DATE: 07/21/2013
|
|
| 14 | 14 |
|
| 15 | 15 |
#PROJECT: NCEAS INPLANT: Environment and Organisms --TASK#363, TASK$568-- |
| 16 | 16 |
|
| ... | ... | |
| 71 | 71 |
|
| 72 | 72 |
#stages_to_run<-c(1,2,3,4,5) #May decide on antoher strategy later on... |
| 73 | 73 |
#stages_to_run<-c(0,2,3,4,5) #May decide on antoher strategy later on... |
| 74 |
stages_to_run<-c(0,0,0,4,5) #MRun only raster fitting, prediction and assessemnt (providing lst averages, covar brick and met stations)
|
|
| 74 |
stages_to_run<-c(0,2,3,4,5) #MRun only raster fitting, prediction and assessemnt (providing lst averages, covar brick and met stations)
|
|
| 75 | 75 |
#If stage 2 is skipped then use previous covar object |
| 76 | 76 |
covar_obj_file<-"/data/project/layers/commons/data_workflow/output_data_365d_gam_fus_lst_test_run_07172013/covar_obj__365d_gam_fus_lst_test_run_07172013.RData" |
| 77 | 77 |
#If stage 3 is skipped then use previous met_stations object |
| ... | ... | |
| 79 | 79 |
|
| 80 | 80 |
|
| 81 | 81 |
var<-"TMAX" # variable being interpolated |
| 82 |
out_prefix<-"_365d_gam_fus_lst_test_run_07182013" #User defined output prefix
|
|
| 83 |
out_suffix<-"_OR_07182013" #Regional suffix
|
|
| 82 |
out_prefix<-"_365d_gam_day_mult_lst_comb3_07202013" #User defined output prefix
|
|
| 83 |
out_suffix<-"_OR_07202013" #Regional suffix
|
|
| 84 | 84 |
out_suffix_modis <-"_05302013" #pattern to find tiles produced previously |
| 85 | 85 |
|
| 86 | 86 |
#interpolation_method<-c("gam_fusion","gam_CAI","gam_daily") #other otpions to be added later
|
| 87 | 87 |
#interpolation_method<-c("gam_CAI") #other otpions to be added later
|
| 88 |
interpolation_method<-c("gam_fusion") #other otpions to be added later
|
|
| 89 |
#interpolation_method<-c("gam_daily") #other otpions to be added later
|
|
| 88 |
#interpolation_method<-c("gam_fusion") #other otpions to be added later
|
|
| 89 |
interpolation_method<-c("gam_daily") #other otpions to be added later
|
|
| 90 | 90 |
#interpolation_method<-c("kriging_daily") #other otpions to be added later
|
| 91 | 91 |
#interpolation_method<-c("gwr_daily") #other otpions to be added later
|
| 92 | 92 |
|
| 93 |
#out_path <- paste("/home/parmentier/Data/IPLANT_project/Venezuela_interpolation/Venezuela_01142013/output_data",
|
|
| 94 |
# out_prefix,"/",sep="") |
|
| 95 |
out_path<-"/data/project/layers/commons/data_workflow/output_data" |
|
| 93 |
out_path<-"/home/parmentier/Data/IPLANT_project/Oregon_interpolation/Oregon_03142013/output_data"
|
|
| 94 |
|
|
| 95 |
#out_path<-"/data/project/layers/commons/data_workflow/output_data"
|
|
| 96 | 96 |
out_path <-paste(out_path,out_prefix,sep="") |
| 97 | 97 |
|
| 98 | 98 |
if (!file.exists(out_path)){
|
| ... | ... | |
| 236 | 236 |
#Set additional parameters |
| 237 | 237 |
#Input for sampling function... |
| 238 | 238 |
seed_number<- 100 #if seed zero then no seed? |
| 239 |
nb_sample<-1 #number of time random sampling must be repeated for every hold out proportion |
|
| 240 |
step<-0 |
|
| 239 |
nb_sample<-15 #number of time random sampling must be repeated for every hold out proportion
|
|
| 240 |
step<-0.1
|
|
| 241 | 241 |
constant<-0 #if value 1 then use the same samples as date one for the all set of dates |
| 242 |
prop_minmax<-c(0.3,0.3) #if prop_min=prop_max and step=0 then predicitons are done for the number of dates...
|
|
| 242 |
prop_minmax<-c(0.1,0.7) #if prop_min=prop_max and step=0 then predicitons are done for the number of dates...
|
|
| 243 | 243 |
#dates_selected<-c("20100101","20100102","20100103","20100901") # Note that the dates set must have a specific format: yyymmdd
|
| 244 |
dates_selected<-"" # if empty string then predict for the full year specified earlier |
|
| 244 |
dates_selected<-c("20100101","20100102","20100301","20100302","20100501","20100502","20100701","20100702","20100901","20100902","20101101","20101102")
|
|
| 245 |
#dates_selected<-"" # if empty string then predict for the full year specified earlier |
|
| 245 | 246 |
screen_data_training<-FALSE #screen training data for NA and use same input training for all models fitted |
| 246 | 247 |
|
| 247 | 248 |
#Models to run...this can be changed for each run |
| 248 | 249 |
#LC1: Evergreen/deciduous needleleaf trees |
| 249 | 250 |
|
| 250 |
#Combination for test run: |
|
| 251 |
|
|
| 252 |
list_models<-c("y_var ~ s(elev_s)",
|
|
| 253 |
"y_var ~ s(LST)", |
|
| 254 |
"y_var ~ s(lat,lon)+ s(elev_s)", |
|
| 255 |
"y_var ~ te(lat,lon,elev_s)", |
|
| 256 |
"y_var ~ s(lat,lon) + s(elev_s) + s(N_w,E_w) + s(LST)", |
|
| 257 |
"y_var ~ s(lat,lon) + s(elev_s) + s(N_w,E_w) + s(LST) + s(LC1)") |
|
| 258 |
|
|
| 251 |
#Combination 3: for paper baseline=s(lat,lon)+s(elev) |
|
| 252 |
list_models<-c("y_var ~ s(lat,lon) + s(elev_s)",
|
|
| 253 |
"y_var ~ s(lat,lon) + s(elev_s) + s(N_w)", |
|
| 254 |
"y_var ~ s(lat,lon) + s(elev_s) + s(E_w)", |
|
| 255 |
"y_var ~ s(lat,lon) + s(elev_s) + s(LST)", |
|
| 256 |
"y_var ~ s(lat,lon) + s(elev_s) + s(DISTOC)", |
|
| 257 |
"y_var ~ s(lat,lon) + s(elev_s) + s(LC1)", |
|
| 258 |
"y_var ~ s(lat,lon) + s(elev_s) + s(CANHGHT)", |
|
| 259 |
"y_var ~ s(lat,lon) + s(elev_s) + s(LST) + ti(LST,LC1)", |
|
| 260 |
"y_var ~ s(lat,lon) + s(elev_s) + s(LST) + ti(LST,CANHGHT)") |
|
| 259 | 261 |
#Default name of LST avg to be matched |
| 260 | 262 |
lst_avg<-c("mm_01","mm_02","mm_03","mm_04","mm_05","mm_06","mm_07","mm_08","mm_09","mm_10","mm_11","mm_12")
|
| 261 | 263 |
|
Also available in: Unified diff
master script, multisampling first predictions with new code using prop hold out 10% to 70%