lib/CXGN/NOAANCDC.pm

   1
   2 =head1 NAME
   3
   4 CXGN::NOAANCDC - helper class for getting weather data from NOAA NCDC weather stations
   5
   6 =head1 SYNOPSYS
   7
   8 my $noaa = CXGN::NOAANCDC->new({
   9     bcs_schema => $schema,
  10     data_types => ['TMIN', 'TMAX', 'PRCP'],
  11     start_date => "2019-01-13", #YYYY-MM-DD
  12     end_date => "2019-12-19", #YYYY-MM-DD
  13     noaa_station_id => "GHCND:US1NCBC0005",
  14     noaa_ncdc_access_token => $noaa_ncdc_access_token
  15 });
  16 my $temperature_averaged_growing_degree_days = $noaa->get_temperature_averaged_gdd($gdd_base_temperature);
  17
  18 # STILL HAS BUGS WITH MULTIYEAR QUERIES STARTING/ENDING WITH YYYY/01/01 or YYYY/12/31 DUE TO do_query_max_one_year()...
  19
  20 # Musgrave stationid = GHCND:USC00300331
  21
  22 # Datatypes:
  23 # PRCP Total Precipitation (+Musgrave)
  24 # F2MN Faster 2 minute wind speed (-Musgrave)
  25 # TMAX Maximum Temperature (+Musgrave)
  26 # TMIN Minimum Temperature (+Musgrave)
  27 # TOBS Temperatue at observation time (+Musgrave)
  28 # TAVG Average Temperature (-Musgrave)
  29 # F5SC Fastest 5 second wind speed (-Musgrave)
  30 # AWND Average Wind Speed (-Musgrave)
  31 # FSMI Fastest Mile (ddfff) (-Musgrave)
  32 # FSMN Fastest One-minute Wind (ddfff) (-Musgrave)
  33 # PRES Station Pressure (-Musgrave)
  34 # RWND Resultant Wind Speed (-Musgrave)
  35 # SLVP Sea Level Pressure (-Musgrave)
  36 # TMPW Wet Bulb Temperature (-Musgrave)
  37 # FSIN Fastest Instantaneous Wind (ddfff) (-Musgrave)
  38 # WDMV 24_hour Wind Movement (-Musgrave)
  39 # MNTP Average Temperature (-Musgrave)
  40 # DPTP Dew Point Temperature (-Musgrave)
  41
  42 =head1 AUTHOR
  43
  44 Nicolas Morales <nm529@cornell.edu>
  45
  46 =head1 METHODS
  47
  48 =cut
  49
  50 package CXGN::NOAANCDC;
  51
  52 use Moose;
  53 use Data::Dumper;
  54 use Try::Tiny;
  55 use SGN::Model::Cvterm;
  56 use LWP::UserAgent;
  57 use JSON;
  58 use Math::Round;
  59 use Time::Piece;
  60 use Time::Seconds;
  61
  62 has 'bcs_schema' => (
  63     isa => 'Bio::Chado::Schema',
  64     is => 'rw',
  65     required => 1
  66 );
  67
  68 has 'data_types' => (
  69     isa => 'ArrayRef',
  70     is => 'rw',
  71 );
  72
  73 has 'start_date' => (
  74     isa => 'Str',
  75     is => 'rw',
  76     required => 1
  77 );
  78
  79 has 'end_date' => (
  80         isa => 'Str',
  81         is => 'rw',
  82     required => 1
  83 );
  84
  85 has 'noaa_station_id' => (
  86     isa => 'Str',
  87         is => 'rw',
  88     required => 1
  89 );
  90
  91 has 'noaa_ncdc_access_token' => (
  92     isa => 'Str',
  93         is => 'rw',
  94     required => 1
  95 );
  96
  97 sub get_noaa_data {
  98     my $self = shift;
  99     my $start_date = $self->start_date;
 100     my $end_date = $self->end_date;
 101
 102     my $data_types_string = '&datatypeid=';
 103     $data_types_string .= join '&datatypeid=', @{$self->data_types};
 104
 105     my $ua = LWP::UserAgent->new(
 106         ssl_opts => { verify_hostname => 0 }
 107     );
 108     my $server_endpoint = "https://www.ncdc.noaa.gov/cdo-web/api/v2/data?stationid=".$self->noaa_station_id."&limit=1000&datasetid=GHCND".$data_types_string."&startdate=".$start_date."&enddate=".$end_date;
 109
 110     print STDERR $server_endpoint."\n";
 111     my $resp = $ua->get($server_endpoint, "token"=>$self->noaa_ncdc_access_token);
 112
 113     my $message_hash = {};
 114     if ($resp->is_success) {
 115         my $message = $resp->decoded_content;
 116         $message_hash = decode_json $message;
 117         # print STDERR Dumper $message_hash;
 118     }
 119     else {
 120         my $message = $resp->decoded_content;
 121         print STDERR Dumper $message;
 122         $message_hash = decode_json $message;
 123     }
 124     return $message_hash;
 125 }
 126
 127 sub do_query_max_one_year {
 128     my $self = shift;
 129     my $start_date = $self->start_date;
 130     my $end_date = $self->end_date;
 131
 132     my $start_date_object = Time::Piece->strptime($start_date, "%Y-%m-%d");
 133     my $end_date_object = Time::Piece->strptime($end_date, "%Y-%m-%d");
 134     my $time_diff = $end_date_object - $start_date_object;
 135     my $time_diff_days = $time_diff->days;
 136     print STDERR Dumper $time_diff_days;
 137     my $end_date_epoch = $end_date_object->epoch;
 138
 139     my @time_ranges;
 140     if ($time_diff_days > 365) {
 141         my $day_start = Time::Piece->strptime($start_date, "%Y-%m-%d");
 142         my $year_counter = 0;
 143         my $day_end_year = $day_start + ONE_DAY * ( 365 - $day_start->day_of_year );
 144         my $day_end_year_start_original = $day_end_year;
 145         while (1) {
 146             if ($year_counter != 0) {
 147                 $day_end_year = $day_start + ONE_DAY * ( 365 - $day_start->day_of_year );
 148             }
 149
 150             if ( $day_end_year > $end_date_object ) {
 151                 push @time_ranges, [$day_start->strftime("%Y-%m-%d"), $end_date_object->strftime("%Y-%m-%d")];
 152                 last;
 153             }
 154             else {
 155                 if ($year_counter == 0) {
 156                     push @time_ranges, [$day_start->strftime("%Y-%m-%d"), $day_end_year->strftime("%Y-%m-%d")];
 157                     $day_start = $day_end_year_start_original + ONE_YEAR * $year_counter;
 158                 }
 159                 else {
 160                     push @time_ranges, [$day_start->strftime("%Y-01-01"), $day_start->strftime("%Y-12-31")];
 161                     $day_start = $day_end_year_start_original + ONE_YEAR * $year_counter + 1 * ONE_DAY;
 162                 }
 163                 $day_start = $day_start->truncate(to => 'day');
 164             }
 165             $year_counter++;
 166         }
 167     }
 168     else {
 169         push @time_ranges, [$start_date, $end_date];
 170     }
 171     print STDERR Dumper \@time_ranges;
 172
 173     my %weather_hash;
 174     foreach (@time_ranges) {
 175         sleep(1);
 176         $self->start_date($_->[0]);
 177         $self->end_date($_->[1]);
 178         my $message_hash = $self->get_noaa_data();
 179         # print STDERR Dumper $message_hash;
 180         if ($message_hash->{status} eq '400') {
 181             return {error => $message_hash->{message}};
 182         }
 183
 184         foreach (@{$message_hash->{results}}) {
 185             $weather_hash{$_->{date}}->{$_->{datatype}} = $_->{value};
 186         }
 187     }
 188     return \%weather_hash;
 189 }
 190
 191 sub get_temperature_averaged_gdd {
 192     my $self = shift;
 193     my $gdd_base_temperature = shift || '50'; #For Maize use 50
 194     my $result = 0;
 195
 196     $self->data_types(['TMIN','TMAX']);
 197
 198     my $weather_hash = $self->do_query_max_one_year();
 199
 200     foreach (values %$weather_hash) {
 201         if (defined($_->{TMIN}) && defined($_->{TMAX})) {
 202             #TMAX and TMIN are in tenths of C
 203             my $tmax_f = (9/5)*($_->{TMAX}/10) + 32;
 204             my $tmin_f = (9/5)*($_->{TMIN}/10) + 32;
 205             my $gdd_accumulation = (($tmax_f + $tmin_f)/2) - $gdd_base_temperature;
 206             if ($gdd_accumulation > 0) {
 207                 $result = $result + $gdd_accumulation;
 208             }
 209         }
 210     }
 211
 212     return {gdd => round($result) };
 213 }
 214
 215 sub get_averaged_precipitation {
 216     my $self = shift;
 217     my $result = 0;
 218
 219     $self->data_types(['PRCP']);
 220
 221     my $weather_hash = $self->do_query_max_one_year();
 222
 223     foreach (values %$weather_hash) {
 224         if (defined($_->{PRCP})) {
 225             $result = $result + $_->{PRCP};
 226         }
 227     }
 228
 229     return $result;
 230 }
 231
 232 sub get_daily_values {
 233     my $self = shift;
 234     my $result;
 235
 236     my $weather_hash = $self->do_query_max_one_year();
 237
 238     while (my ($date, $do) = each %$weather_hash) {
 239         while (my ($data_type, $value) = each %$do) {
 240             $result->{$date}->{$data_type} = $value;
 241         }
 242     }
 243     my @sorted_dates = sort keys %$result;
 244
 245     return ($result, \@sorted_dates);
 246 }
 247
 248 1;