Special report: Why are Pakistan’s weather forecasts so cloudy?

Raise your hand if you have recently cancelled plans due to unexpected rain or intense heat. Do you check the Pakistan Meteorological Department’s (PMD) weather forecast before stepping out of the house? Do you manage your plans according to these predictions?

If you’re shrugging off these questions, worry not because you aren’t alone. A survey conducted by Dawn.com via Instagram reveals that 75 per cent of Pakistanis (based on 730 respondents) don’t bother checking the PMD forecast at all, while 71pc (out of 778 respondents) find them as trustworthy as a politician’s promise.

By now, scepticism towards the Met Office’s predictions is second nature to Pakistanis. They’ve come to accept the harsh reality of the country’s unpredictable weather patterns and the PMD’s incapability to gauge these patterns.

Muhammad Usman is one such person who recently encountered the shortcomings of our weathermen. After five years abroad, he visited Pakistan this July, eager to reunite with friends in Chitral. He was, however, wary of venturing into the mountains — due to a sea of flood and rain warnings issued by the PMD — ultimately forcing him to cancel his plans.

Despite several assurances from friends, who kept telling him that the skies were clear in Chitral, Usman ended up leaving without visiting them.

One might think that this is because the weather is beyond the reach of technology, or maybe climate change has made it too random to predict. Surprisingly, neither is true; the issue lies elsewhere, behind the walls of Pakistan’s ramshackle weather forecasting department.

Guided by gut instinct

While Usman was just relying on the PMD for a travel weather forecast, there are millions of citizens whose very bread and butter is intrinsically linked to our weathermen’s ability to predict the weather.

Marwan Khan, a Charsadda-based farmer, has been unable to prepare his fields for sugarcane season. Last year, he incurred substantial losses due to unpredictable rains, which washed away fertiliser that was essential to his crops, snatching away all their nutrients from the soil.

According to Khan, sugarcane is usually planted in August. But this time, the weather was unfavourable. “I check all the apps and official forecasts, but none are accurate enough for me to plan my planting and harvesting dates,” he said.

These uncertainties have forced farmers to rely on their own gut — harkening back to the old days, before long-range weather radar was a thing.

“Our farmers observe the clouds and make decisions on their own instead of solely relying on official forecasts from the MET Department, because they are almost always wrong,” said Akram Khaskheli, president of the Hari Welfare Association.

Khaskheli is unequivocal; accurate weather predictions would be a significant blessing for people in his line of work. “For example, if we know that rain is expected on a particular day, we can harvest crops like cotton beforehand, as wet cotton loses its quality,” he explained, noting that this uncertainty has led many a farmer to harvest his crop prematurely.

Fishermen are also affected by inaccurate forecasts.

Majeed Motani, president of the Karachi Fisherfolk Forum, said that his community has taken a hit from erratic weather patterns; with several fishermen losing their lives and boats during cyclones in recent years.

Their faith in the official weather forecasting system has effectively evaporated.

But the question remains: what is the PMD doing wrong?

Predictive puzzles

The PMD’s weather observational network consists of 195 stations at 110 locations, some of which are over a century old. Of these, 110 are manual observation centres, while 45 are Automatic Weather Stations (AWS).

In addition to these, the PMD has four radars, but only two are functional — Karachi and Islamabad.

Atif Hassan, a meteorological assistant at a manual weather observation centre near Karachi’s Jinnah International Airport, takes four daily temperature readings from instruments such as wet and dry bulb thermometers.

At the centre, which was built in 1928, these devices are placed inside an insulated box with gaps for air circulation. This allows him to record the maximum and minimum temperatures, the current temperature, dew point, and relative humidity before sunrise and sunset.

Rainfall is measured with a rain gauge — typically every three hours, and at hourly intervals during heavy downpour. All instruments are placed on uncemented ground to avoid heat absorption from affecting the readings.

A barometer on the roof of the centre measures air pressure — low pressure indicates rain and high pressure suggests dry/hot weather. A sunshine recorder, consisting of a glass ball that burns a paper strip placed beneath it using focused sunlight, tells Hassan the duration of sunshine on that day. He also has a wind vane, which indicates wind direction, and an anemometer that measures wind speed.

The readings taken by Hassan are then uploaded into a software, used by the PMD to transmit readings from the observation stations to the forecasters stationed at their main office.

“In a densely populated metropolis like Karachi, the World Meteorological Organisation (WMO) requires weather observation stations to be set up every 50kms,” Hassan told Dawn.

Karachi’s five weather stations, located miles from each other, are quite a ways off the standards set by the global authority. Two of them are located near the airport, while there is one each in PAF Masroor Base, Keamari and Gulistan-i-Johar.

Due to this sparse network, predicting rapidly developing weather phenomena like cloudbursts is a major challenge. To mitigate this, the department relies on radar data for short-range weather predictions and warnings of potential calamities, such as thunderstorms or tsunamis, Hassan explains.

Observations from the five weather stations do not provide sufficient data to predict the sprawling metropolis’ weather, admitted Anjum Nazir, senior officer at the PMD’s Early Warning Centre.

In the past, PMD had approached several departments, seeking permission to install Automatic Weather Stations (AWS), “but none of them have cooperated,” Nazir claimed.

Asked if the lack of data points was the reason for the inefficient forecasts, he says: “On ground weather stations’ readings are not the only ones we use to predict the next day’s weather.”

To draw up a forecast, a forecaster has four different reports at their disposal.

• The Aircraft Meteorological Data Report (AMDAR) tells forecasters about the weather in the upper reaches of the atmosphere. These observations are obtained using sensors carried by commercial aircraft, which land at airports across Pakistan.

• A Meteorological Aerodrome Report (METAR) is generated internationally every 30 minutes, and domestically every hour. It tells forecasters the visibility, cloud position, temperature and dew point in an area. Based on the readings in these reports, a weather trend can be drawn up.

• The Significant Meteorological Warning (SIGMET) report, which is issued by the WMO, is primarily meant for aircraft pilots and gives them updates on what the weather would be en route to their destination, or warns them about other phenomena that may have an impact on the safety of their aircraft.

• A Terminal Area Forecast (TAF), which is a message in a defined format, reports the weather forecast for a single airport and its immediate vicinity. These reports are generated four times a day and are valid for 24 hours. The message is generated by the International Civil Aviation Organisation (ICAO) based on information it receives from its Flight Information Regions (FIR), located around the world. In Pakistan, these regions are located in Karachi and Lahore.

Using readings from the 110 local MET observation centres, satellite images and reports, weather models, surface charts of weather from 5,000ft to 10,000ft above sea level, as well as other data, a team of forecasters can draw up weather predictions for the next day, month and season.

“We collaborate with departments in different cities before drawing up forecasts; it is a long, tiresome process,” said Nazir.

He acknowledged gaps in weather station coverage in Sindh and Balochistan, but notes that a World Bank-funded project is under way to install AWS at the tehsil level across Pakistan, which is expected to take five years.

Meanwhile, he said, the current system is “good enough”.

But is it really?

Lack of resources

A quick look at the PMD website yields vague ‘now-casts’ and short-term forecasts — three-day predictions — that lack specificity. They just tell you if the weather is going to be sunny, or rainy.

One may be tempted to give them the benefit of the doubt, given Pakistan’s tropical location and erratic weather changes. But neighbouring India, which has similar geographical conditions, provides far more detailed forecasts.

When asked inquired why India is able to issue better forecasts, PMD’s Chief Meteorologist Dr Sardar Sarfaraz attributed this to their use of far superior technology.

He explained that Pakistan’s meteorological network was quite limited. International standards dictate that there should be at least 353 weather stations in different locations for a country of Pakistan’s size, as opposed to the 110 currently in place.

Furthermore, while India operates its own weather observation satellites, Pakistan relies on the satellites of neighbouring countries.

Addressing the absence of an effective local weather model, Dr Sarfaraz said that the department, with the help of a World Bank soft loan, will soon acquire the much-needed supercomputers required to make these mammoth calculations.

“Hopefully, after that, we can develop our own weather model,” he said.

However, as the department struggles to equip itself with the necessary technology — a process that officials say could take around half a decade — social media-savvy citizens have turned towards private weather channels, whose forecasts often do not line up with the PMD’s predictions.

One Karachi-based journalist said she usually goes by Pak_Weather, a private channel, for everyday updates and finds it to be far more reliable. Several other people Dawn spoke to for its survey expressed similar views.

The PMD, however, views these newcomers with suspicion, saying that they do nothing but “spread panic among people because that is the only way they can get attention to generate money for running their operations”.

“Their machines are placed on top of buildings, whose height can impact the readings. Who knows if they are up to the standards [set] by the World Meteorological Organisation (WMO),” Nazir wondered.

“PMD, on the other hand, has to calibrate their machines every six months and comply with international standards.”

He added that private weather stations mostly depend on international forecasting websites, combining them with their on-ground observations based on data from one or two individual AWS.

“If we stop providing ground data to the WMO, the international websites won’t get Pakistan’s data,” said Nazir, stressing that at the end of the day, the private channels also rely on data from the PMD.

Private forecasting solutions

Junaid Yamin, who runs WeatherWalay, another local private weather channel, said that his company’s raison d’être was a lack of trust in PMD and the need for better forecasts. “We realised very soon that the main reason for the inaccuracy was a lack of data points,” Mr Yamin told Dawn, explaining that to address this issue, the company has installed more than 250 AWS across Pakistan.

The AWS measures current conditions such as temperature, precipitation, rainfall, air pressure, wind direction, among other readings. For example, instead of collecting water in a bottle to measure rainfall — as is done by the rain gauge used by PMD — an AWS uses a spoon-shaped instrument that tips each time 0.1mm of water falls on it, ultimately providing a more precise rainfall measurement for the area.

When asked if the placement of these instruments on the rooftops of buildings affected the reliability of their readings, WeatherWalay forecaster Faizan Khan said the systems are built with the ability to overcome possible errors.

Once the data is recorded, it is synced to the cloud used by forecasters at their main station, in Islamabad.

At their headquarters, a team of forecasters analyses the on-ground weather conditions, comparing them with past weather patterns, looking at international numerical weather model predictions, satellite imagery and weather charts.

They then draw up short-range forecasts that predict the weather for 24 hours; mid-range forecasts that cover the next 10 days; long-range forecasts for the next 15 to 30 days; and, extended-range forecasts that look up to 42 days ahead.

“We have achieved almost 90pc accuracy in now-casting and short-term weather forecasting and are working towards improving our mid-range and long-range forecasts,” he said.

However, Yamin noted that around the world, forecast accuracy halves for every 10 days, the further you go.

Khan concurred, adding that their inability to maintain 90pc accuracy with now-casts and short-range forecasts was due to a lack of sufficient AWS around Pakistan.

In Karachi, WeatherWalay has 10 AWS located in Bahria Town, Gulshan-i-Iqbal, Gadap Town, Airport, Malir, Ghulshan-i-Hadeed, PIB Colony, Keamari, Korangi, and North Karachi. “But in a densely populated city like Karachi, we need at least 50-60 such stations to provide an accurate picture,” said Khan.

Yamin also mentioned that the company has a long way to go. “In the first stage, we divided Pakistan into 600 climate zones, each of which requires an AWS, based on topography.”

With almost 300 AWS nationwide, Yamin claims they are already predicting weather better than the PMD, but admits that challenges remain.

“The lack of network infrastructure makes it difficult to get data from Balochistan, which is a virtual ‘black hole’ in our forecasts.”

He also said that without mobile network coverage, data transmission from centres to the head office becomes a challenge. A similar issue exists in the country’s northern areas, where frequent electricity outages and internet disruptions hinder data-syncing from the AWS, delaying forecasts.

Lack of a weather model

Another major challenge is the absence of a dedicated weather model for Pakistan.

“Currently, we use international models that are downscaled to understand Pakistan’s weather conditions,” said Yamin.

However, these models are not calibrated with Pakistan’s data, nor are international agencies pushed to make the adjustment. “They aren’t interested because updating these models requires a significant amount of money, and the returns from Pakistan wouldn’t be profitable enough,” he noted.

In his view, predictions made using these models are only 40pc accurate, due to a lack of detailed information on Pakistan’s topography, climate conditions and current weather conditions.

“For now, we need to get Pakistan’s data calibrated into an international model, but in the long run, we need our own weather model,” said Yamin. “A local weather model could include the smallest of details, each of which would help improve the accuracy of predictions.”

Yamin admitted that they faced challenges initially. “PMD always had a monopoly over forecasting. When we entered the market, there were many reservations from them.” However, with time, relations have improved, and now the company collaborates with government departments like PMD and the National Disaster Management Authority (NDMA).

“Pakistan is the world’s sixth most vulnerable country to climate change, and the absence of spring and autumn seasons is proof of its impact,” he warned.

“The weather in the coming years will become even more erratic, making it essential for us to have the capability to predict it accurately and prepare for the disasters it may bring.”