I wrote the other day about the new datapoint API from the MetOffice (there were some great links to other weather APIs in the comments, if you like weather). I've been using it to create a detailed forecast of the weather over the next few days, mixing in some lovely weather icons by Adam Whitcroft, from The Noun Project (the same site that the icons on my own site came from) - so I have something like this for each kind of weather:

To achieve this, I hit the datapoint API from PHP (but if you use any other scripting language this should be easy enough to follow):

<?php

$leeds_site_id = '352241';
$fcast_url = 'http://datapoint.metoffice.gov.uk/public/data/val/wxfcs/all/json/' . $leeds_site_id . '?res=3hourly&key=' . $api_key;
$fcast = json_decode(file_get_contents($fcast_url), true);
foreach($fcast['SiteRep']['DV']['Location']['Period'] as $day) {
    $fdate = new DateTime($day['value']);
    echo $fdate->format('D jS M') . "\n";;
    foreach($day['Rep'] as $slot) {
        if($slot['$'] > 0) {
            // makes 3, 12 into 0300 and 1200 respectively
            echo str_pad(($slot['$'] / 60) . "00", 4, "0", STR_PAD_LEFT);
        } else {
            // avoid divide by zero
            echo "0000";
        }
        echo ": "
            . "<img src="images/" /> "
            . $weathers[$slot['W']] . "<br />\n";
    }
}

// lovely nounproject icons need some attribution
echo $attr;

I live in Leeds so I've used the site_id for that city (annoyingly they don't seem to document them, just request all locations for one time period and then grab whichever one you want is my advice). When you build the URL like this, the response is in JSON format so I convert that quickly to an array and then foreach over the various forecasts. One challenge is figuring out what time the forecast relates to, I'm certain this can't be the most elegant solution but it does seem to work!

I also added pretty images, by taking the list of weather types and grabbing some icons to work with those. So my HTML output ends up (not styled because I'm pasting it into the middle of my blog, but you get the idea) looking like this:

Tue 18th Dec

1500:  Overcast

1800:  Cloudy

2100:  Cloudy

Wed 19th Dec

0000:  Mist

0300:  Overcast

0600:  Overcast

0900:  Overcast

1200:  Overcast

1500:  Heavy rain

1800:  Heavy rain

2100:  Heavy rain

Thu 20th Dec

0000:  Heavy rain

0300:  Heavy rain

0600:  Heavy rain

0900:  Heavy rain

1200:  Light rain

1500:  Heavy rain

1800:  Light rain

2100:  Overcast

Fri 21st Dec

0000:  Cloudy

0300:  Cloudy

0600:  Cloudy

0900:  Cloudy

1200:  Cloudy

1500:  Partly cloudy (day)

1800:  Partly cloudy (night)

2100:  Partly cloudy (night)

Sat 22nd Dec

0000:  Cloudy

0300:  Cloudy

0600:  Cloudy

0900:  Light rain

1200:  Heavy rain

1500:  Heavy rain

1800:  Light rain

2100:  Light rain

Icons by Adam Whitcroft, from The Noun Project

This is a nice way of illustrating one of the data fields that comes from the datapoint API, and I love the excellent icons, they are wonderfully clear and well made (most of the icons on this site are noun project icons). In fact you could use any of the various data points with any output mechanism you please ... my next datapoint post will probably be about how I use this data with my blink(1) :)