Revision 19d44fdd
Added by Benoit Parmentier about 8 years ago
| climate/research/oregon/interpolation/global_product_assessment_part0.R | ||
|---|---|---|
| 469 | 469 |
rat$legend <- tb_freq$value |
| 470 | 470 |
levels(r_table) <- rat |
| 471 | 471 |
|
| 472 |
|
|
| 473 | 472 |
res_pix <- 800 |
| 474 | 473 |
#res_pix <- 480 |
| 475 | 474 |
col_mfrow <- 1 |
| ... | ... | |
| 488 | 487 |
|
| 489 | 488 |
dev.off() |
| 490 | 489 |
|
| 491 |
#r <- raster(matrix(runif(20),5,4)) |
|
| 492 |
#r[r>.5] <- NA |
|
| 493 |
#g <- as(r, 'SpatialGridDataFrame') |
|
| 494 |
#p <- as(r, 'SpatialPixels') |
|
| 495 |
#p_spdf <- as(r_overlap_m,"SpatialPointsDataFrame") |
|
| 496 |
#plot(r) |
|
| 497 |
#points(p) |
|
| 498 |
|
|
| 499 | 490 |
### now assign id and match extent for tiles |
| 500 | 491 |
|
| 501 | 492 |
lf_files <- unlist(list_predicted) |
| ... | ... | |
| 541 | 532 |
#r_tiles_s <- r_tiles_stack |
| 542 | 533 |
names_tiles <- basename(in_dir_reg) |
| 543 | 534 |
|
| 535 |
|
|
| 536 |
list_param_generate_raster_number_pred <- list(list_tiles_predicted_masked,df_missing_tiles_day,r_overlap_m, |
|
| 537 |
num_cores,region_name, |
|
| 538 |
NA_flag_val,out_suffix,out_dir) |
|
| 539 |
|
|
| 540 |
names(list_param_generate_raster_number_pred) <- c("list_tiles_predicted_masked","df_missing_tiles_day","r_overlap_m",
|
|
| 541 |
"num_cores","region_name", |
|
| 542 |
"NA_flag_val","out_suffix","out_dir") |
|
| 543 |
debug(generate_raster_number_of_prediction_by_day) |
|
| 544 |
|
|
| 545 |
obj_number_pix_predictions <- generate_raster_number_of_prediction_by_day(1,list_param_generate_raster_number_pred) |
|
| 546 |
|
|
| 547 |
obj_number_pix_predictions <- mcapply(1:nrow(df_missing_tiles_day), |
|
| 548 |
FUN=generate_raster_number_of_prediction_by_day, |
|
| 549 |
list_param=list_param_generate_raster_number_pred, |
|
| 550 |
mc.preschedule=FALSE, |
|
| 551 |
mc.cores = num_cores) |
|
| 552 |
|
|
| 544 | 553 |
generate_raster_number_of_prediction_by_day <- function(i,list_param){
|
| 554 |
##This function generates raster of missing pixels and number of predictions for days with missing tiles for a given region. |
|
| 555 |
#INPUTS |
|
| 556 |
#1) list_tiles_predicted_masked |
|
| 557 |
#2) df_missing_tiles_day |
|
| 558 |
#3) r_overlap_m |
|
| 559 |
#4) num_cores |
|
| 560 |
#5) region_name |
|
| 561 |
#6) NA_flag_val |
|
| 562 |
#7) out_suffix |
|
| 563 |
#8) out_dir |
|
| 545 | 564 |
|
| 546 |
list_names_tile_coord |
|
| 547 |
df_time_series |
|
| 548 |
missing_tiles <- df_missing_tiles_day[i,] |
|
| 549 |
date_str <- missing_tiles$date |
|
| 550 | 565 |
|
| 551 |
#r_tiles_s <- list_param$r_tiles_s |
|
| 552 |
list_param$list_tiles_predicted_masked |
|
| 566 |
###### Start script ##### |
|
| 553 | 567 |
|
| 568 |
#### read in paramters |
|
| 569 |
list_tiles_predicted_masked <- list_param$list_tiles_predicted_masked |
|
| 570 |
df_missing_tiles_day <- list_param$df_missing_tiles_day |
|
| 571 |
r_overlap_m <- list_param$r_overlap_m |
|
| 572 |
num_cores <- list_param$num_cores # 6 #PARAM 14 |
|
| 573 |
region_name <- list_param$region_name #<- "world" #PARAM 15 |
|
| 574 |
NA_flag_val <-list_param$NA_flag_val |
|
| 575 |
out_suffix <- list_param$out_suffix |
|
| 576 |
out_dir <- list_param$out_dir |
|
| 577 |
|
|
| 578 |
### To add or explore later...could have differences between predictions and rmse |
|
| 579 |
layers_option <- c("var_pred") #arg 17 ,param 17, options are:#res_training, res_testing,ac_training, ac_testing, var_pred
|
|
| 580 |
NA_value <- NA_flag_val |
|
| 581 |
#metric_name <- "rmse" #to be added to the code later... |
|
| 582 |
|
|
| 583 |
##### Select relevant day and create stack of missing tiles |
|
| 584 |
|
|
| 585 |
missing_tiles <- df_missing_tiles_day[i,] |
|
| 586 |
date_str <- missing_tiles$date |
|
| 554 | 587 |
selected_col <- names(list_tiles_predicted_masked) |
| 555 | 588 |
missing_tiles_subset <- subset(missing_tiles,select=selected_col) |
| 556 | 589 |
selected_missing <- missing_tiles_subset==1 |
| 557 |
#names(df_missing_tiles_day)[selected_missing] |
|
| 558 |
|
|
| 559 |
#names(list_tiles_predicted_masked)[selected_missing] |
|
| 590 |
|
|
| 560 | 591 |
list_missing_tiles_raster <- list_tiles_predicted_masked[selected_missing] |
| 561 | 592 |
r_tiles_s <- stack(list_missing_tiles_raster) |
| 562 | 593 |
|
| 563 |
### first sum missing |
|
| 594 |
##### Sum missing tiles in the stack and generate number of predictions by pixels |
|
| 595 |
|
|
| 564 | 596 |
datasum <- stackApply(r_tiles_s, 1:nlayers(r_tiles_s), fun = sum) |
| 565 | 597 |
|
| 566 | 598 |
### then substract missing tiles... |
| 567 | 599 |
r_day_predicted <- r_overlap_m - datasum |
| 568 |
|
|
| 600 |
raster_name_number_prediction <- file.path(out_dir_str,paste("r_day_number_of_prediction_sum_day_mosaiced","_",region_name,"_masked_",date_str,file_format,sep=""))
|
|
| 601 |
writeRaster(r_day_predicted, NAflag=NA_flag_val,filename=raster_name_number_prediction,overwrite=TRUE) |
|
| 602 |
|
|
| 569 | 603 |
r_table <- ratify(r_day_predicted) # build the Raster Attibute table |
| 570 | 604 |
rat <- levels(r_table)[[1]]#get the values of the unique cell frot the attribute table |
| 571 | 605 |
#rat$legend <- paste0("tile_",1:26)
|
| ... | ... | |
| 612 | 646 |
|
| 613 | 647 |
dev.off() |
| 614 | 648 |
|
| 615 |
### generate retunr object |
|
| 649 |
### writeout data |
|
| 650 |
#extension_str <- extension(lf_files) |
|
| 651 |
#raster_name_tmp <- gsub(extension_str,"",basename(lf_files)) |
|
| 652 |
#out_suffix_str <- paste0(region_name,"_",out_suffix) |
|
| 653 |
raster_name_missing <- file.path(out_dir_str,paste("r_missing_day_mosaiced","_",region_name,"_masked_",date_str,file_format,sep=""))
|
|
| 654 |
writeRaster(r_missing_day, NAflag=NA_flag_val,filename=raster_name_missing,overwrite=TRUE) |
|
| 655 |
|
|
| 656 |
### generate return object |
|
| 657 |
obj_number_day_predicted <- list(raster_name_number_prediction,raster_name_missing) |
|
| 658 |
names(obj_number_day_predicted) <- c("raster_name_number_prediction","raster_name_missing")
|
|
| 659 |
|
|
| 616 | 660 |
returm(r_day_predicted) |
| 617 | 661 |
} |
| 618 | 662 |
|
Also available in: Unified diff
modifying generate_raster_number_of_prediction_by_day function