### then substract missing tiles...

  r_day_predicted <- r_overlap_m - datasum

  writeRaster(r_day_predicted, NAflag=NA_flag_val,filename=raster_name_number_prediction,overwrite=TRUE)  

  r_table <- ratify(r_day_predicted) # build the Raster Attibute table

  col_mfrow <- 1

  row_mfrow <- 1

  title_str <-  paste("Number of predicted pixels for ",variable_name," on ",date_str, sep = "")

  col_mfrow <- 1

  row_mfrow <- 1

  title_str <-  paste("Number of predicted pixels for ",variable_name," on ",date_str, sep = "")

  #extension_str <- extension(lf_files)

  #raster_name_tmp <- gsub(extension_str,"",basename(lf_files))

  #out_suffix_str <- paste0(region_name,"_",out_suffix)

  writeRaster(r_missing_day, NAflag=NA_flag_val,filename=raster_name_missing,overwrite=TRUE)  

  ### generate return object

                                                 FUN=function(i){mask(raster(list_tiles_predicted_m[i]),r_mask,filename=list_mask_out_file_name[i],overwrite=T)},

                                                       mc.preschedule=FALSE,mc.cores = num_cores))                         

  #r_stack_masked <- mask(r, m2) #, maskvalue=TRUE)

  ########################

  #### Step 3: combine overlap information and number of predictions by day

  ##Now loop through every day if missing then generate are raster showing map of number of prediction

  #function_product_assessment_part0_functions <- "global_product_assessment_part0_functions_11152016b.R"

  #source(file.path(script_path,function_product_assessment_part0_functions)) #source all functions used in this script 

  if(nrow(df_missing_tiles_day)>0){

    obj_number_pix_predictions <- mclapply(1:nrow(df_missing_tiles_day),

                                        mc.preschedule=FALSE,

                                        mc.cores = num_cores)

  }

}