fix: Standardized tavg availability filename and updated precipitation data filtering logic.

global.R

@@ -15,7 +15,7 @@ source("utils/generateLabel.R", local = T)
 
 # Read the metadata and availability data
 tavg_meta <- read.csv("www/data/tabs/tavg_meta.csv")
-tavg_avail <- read.csv("www/data/tabs/tavg_vaialability.csv")
+tavg_avail <- read.csv("www/data/tabs/tavg_availability.csv")
 
 # Merge the metadata and availability data on 'ID'
 stations_data <- merge(tavg_meta, tavg_avail, by = "ID")

misc/Python/ghcnm_read_prec.py

@@ -125,7 +125,8 @@ for i, file in enumerate(csv_files):
     # 4. Skip this file if any 'precip_mm' value is greater than 5000
     if (df['precip_mm'] > 2000).any():
         #continue
-        sys.exit(1)
+        print(f"Skipping {file} due to extreme precipitation value (> 2000mm)")
+        continue
 
     # 5. Append cleaned and filtered dataframe
     dataframes.append(df)
@@ -137,7 +138,7 @@ combined_df.to_parquet('www/data/tabs/prec_long.parquet', engine='pyarrow', inde
     
 
 
-!rm -rf misc/data/*
+
 
 combined_df.describe()
 

misc/Python/ghcnm_read_tavg.py

@@ -110,7 +110,7 @@ station_summary = long_format_data.groupby('ID')['YEAR'].agg(first_year='min', l
 print(station_summary)
 
 
-station_summary.to_csv("www/data/tabs/tavg_vaialability.csv", index=False)
+station_summary.to_csv("www/data/tabs/tavg_availability.csv", index=False)
 
 
 ### read text  file 

www/data/tabs/{tavg_vaialability.csv → tavg_availability.csv}

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:31f40d20ae1ff0b88e81e3dd89fd557a4e39ba2f8e8aa924763b0e3720971dfa
-size 611162
+oid sha256:26e585ce08c48911e59b22f2ceec9eb6994a21b5a2d56eac3df275b9c879ac22
+size 611360

www/data/tabs/tavg_long.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ee63b2b4bc2daa6626a47cd080804d15155ee95052cc572fc2aa01fa3d068e58
-size 29906393
+oid sha256:899fd5826cdb0d597a405a0922011dac853f837f08776bc77fdf049ed1f05ede
+size 29958269

www/data/tabs/tavg_meta.csv

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e82a456b4d169606c2c4dd0b02bcfbb73a293697ee515e22448e5b90b0de01f2
-size 1295927
+oid sha256:2726a073d64963e8e8410830ab59e5990554a3780183c3d5b41a1a1c93a62ad8
+size 1296342