A post about the organization, planning, execution, and control of weather radar surveillance. (En Español aquí)
I have known institutions and countries that got very excited about acquiring one or several radars, whether thanks to a timely donation or because they secured a budget to buy them.
Unfortunately, the prevailing idea is that the «turnkey» motto that vendors usually wave around means they have bought «the automatic forecaster»—a machine you plug in, let it run, and it will warn any would-be users of any dangerous weather phenomenon within its range. I hate to be a spoilsport, but things are far from that today.
Maybe AI will make it possible in the future, but I don’t see it happening anytime soon. Allow me to explain why.
Operational availability and measurement accuracy: the radar’s non‑negotiable standards
Like any other undertaking, using radars for a weather surveillance service has to be organized, planned, executed, and controlled if you want to get the results you can expect from such an expensive investment. Kind reader, please bear in mind that of all the tools available to meteorologists, radars are, by a long shot, the most costly investment—and therefore the one that demands the greatest effort to pay off.
You buy a car. It’s reasonable to expect that over the next five years, that car will be available whenever you need it. You use it every day, several times a day. Then one bad morning, when you’re heading to work… it won’t start. You exclaim angrily: «Today, of all days, when I needed it most, this damn thing won’t start!»—Liar! You need it every day, at any hour you happen to need it.
The same goes for radars. You need it every single day. If one day it breaks down, right when a storm is brewing, the meteorologists will be furious and the bosses will look for someone to blame. From this we can draw a conclusion: when you buy an expensive radar (they’re all expensive), you are entitled to uninterrupted operation 24/7/365. Every observation it is supposed to make is a fraction of the price you paid (amortizing the cost). If it doesn’t make that observation, that’s money you threw away and cannot recover—just as you cannot recover lost time.
So far we’ve talked in simple terms: works / doesn’t work. But it’s much more complicated than that. It’s not enough that the transmitter radiates, the receiver receives, and the antenna spins. The problem is that the radar is a measuring instrument that measures six variables: Reflectivity, Radial Velocity, and Spectral Width of radial velocities, repeated in two channels (horizontal and vertical). Like any measuring instrument, it must be calibrated in those three variables and between the two channels. So far, these are things that can be calibrated with precision and accuracy.
Planning
Planning is the foundation of successful surveillance. «A goal without a plan is just a wish» — Exupéry.
You wouldn’t buy a house without first having seen the neighborhood, without having asked about the quality of the neighbors, without having checked whether the roof can withstand the rains in your region, and without having calculated whether your budget can sustain the taxes and maintenance. Well, the same goes for a weather radar. The excitement of the purchase or the donation cannot be the only engine driving the decision. Planning is the foundation on which the entire service will rest for the next twenty or thirty years. If those foundations are made of mud, no matter how pretty the building is, it will come down with the first storm—and that’s precisely what you bought it for.
In this post we take for granted that the radars have already been acquired, but we cannot take for granted that they were planned properly. That is why, even if you already have them, it is worth reviewing what should have been done—because it will help you correct course in time and avoid repeating mistakes in future acquisitions.
Threat assessment: what do we want to see, and why?
Before even thinking about which radar to buy, you must sit down with the meteorologists, the civil protection officers, the airports, the farmers, and the water resource managers. Each of them has a list of phenomena they worry about. For some it’s flash floods; for others, destructive winds; for others, wind shear on runways; for others, hail that ruins crops.
Not all radars do everything. An S‑band radar (10 cm wavelength) is excellent for seeing deep storms and penetrating heavy rain without attenuation, but it is expensive and bulky. A C‑band radar (5 cm) is more compact and economical, but suffers more attenuation in very heavy rain. An X‑band radar (3 cm) is ideal for local and short‑range applications, but it is practically blind behind an intense storm. Planning means knowing which phenomena predominate in your region and choosing the radar that best detects them—not the one they offer you cheapest or the one they give you with the most pomp.
Radar siting: the art of seeing without being seen
The radar needs a clear line of sight to the horizon. Any obstacle—mountain, tall building, dense forest—produces blockages that hide entire areas. In mountainous regions, a single radar can never see into deep valleys; you need several, or a very high location, such as a mountain top. But beware: putting a radar on an isolated summit sounds great, but then comes access: how does the technician get there with the oscilloscope when there’s snow or mud? How is it powered reliably? How is it protected from lightning? Siting planning is an exercise in compromise between optimal coverage and logistical viability.
Also, you must consider interference with other radars. If you put two radars too close together and they operate on the same frequency, you’ll get an annoying «clutter» on the screens that will ruin the data. If you put them too far apart, you leave gaps. In countries with multiple radars, network planning is a puzzle that demands wave‑propagation studies and detailed topography.
Support infrastructure: the radar doesn’t live alone
A radar is not just a spinning antenna and an electronics cabin. It needs:
- Stable electrical supply, with generator and battery backup that can last at least 48 hours of autonomy. In many countries, the power grid fails precisely when there is a storm—just when you need the radar most, the lights go out. Planning means installing generators of sufficient power, with stored fuel and scheduled maintenance.
- Redundant communications: fiber optics are wonderful, but a single cut in the trench leaves the radar mute. You must plan for microwave links, satellite, or even backup radio links. It cannot be that the radar is seeing the storm perfectly, but the forecasters can’t see it because the link broke.
- Physical access: passable roads year‑round, even in winter or rainy season. A radar that’s down because the technician couldn’t get to it is just as useless as if it were broken.
- Lightning protection: radar towers are the highest points in the landscape; they attract lightning with fury. A well‑sized lightning‑rod system and a good grounding are mandatory investments, not luxuries.
Spare parts and critical consumables management: the logistics of survival
I’ve left this point for the end of planning because it’s often the most forgotten. And yet, it’s the one that saves the most headaches.
When you buy a car, you know that within five years you’ll need to change the tires, the brakes, the oil, and maybe the battery. For some brands, you can find spare parts on any corner; for others, you have to import them and wait weeks. With radars it’s the same, but with much longer lead times and much higher prices.
The magnetron or klystron—the tube that generates the transmission power—has a useful life of several tens of thousands of hours. But it doesn’t warn you when it’s going to fail. One day, it simply stops radiating. If you don’t have a spare one in the warehouse, your radar will be out of service for weeks or months, depending on import procedures, manufacturer availability, and available budget.
Planning means identifying, together with the manufacturer, the list of critical spare parts: those that, if they fail, leave the radar completely inoperative. Those must be in the warehouse from the day of installation. In addition, you must plan for periodic wear‑and‑tear spares: antenna motors, power supplies, electronic boards, high‑voltage cables. All of this requires an inventory, controlled rotation so they don’t deteriorate in storage, and a backup supplier (quite an art) in case the main manufacturer has delays.
You also need to plan for the consumables and third‑party services—auxiliary systems without which the radar doesn’t work: power, air conditioning, communications, and protections. You must identify the providers of these services and seek redundancy so that a failure in one of these systems doesn’t stop your radar. In certain countries this can be quite tricky, especially if the radars are in remote and hard‑to‑reach locations.
Also plan a multi‑year budget for these spare parts. It’s not a one‑time expense; it’s an annual fixed cost (estimated at 10‑15% of the radar’s price). Many institutions buy the radar with a million‑dollar budget and then have no funds for maintenance. It’s like buying an airplane and having no money for fuel or inspections. By the third year, the radar starts to fail; by the fifth, it becomes a metal ornament that only serves as a perch for birds.
Planning the human factor: it’s not all machinery
Planning must also consider the people who will operate and maintain the radar. Training a radar technician is not a two‑week course; it takes years of experience. If the institution doesn’t plan for recruitment and training before the radar arrives, it will end up with a shiny machine and nobody who knows how to use it.
You must plan factory training trips, refresher courses, exchanges with other countries, and above all, a succession plan: the expert engineer will retire someday—who will take their place? That question must be answered during the planning phase, not when the expert is already gone and the antenna has stopped spinning.
Planning, in summary, means answering these six questions before the radar arrives:
- What phenomena are we going to monitor, and with what priority?
- Where do we place the radar to see them best, without creating interference or access problems?
- What technical specifications must the radar have for those phenomena and that location?
- What infrastructure (power, communications, roads, protection) must we install?
- Which spare parts and tools must we have from day one, and how will we renew them?
- Which people are we going to train, and how do we ensure continuity?
If all these answers are clear and documented, then the purchase—or the donation—will be a sound investment. If not, it will be an expensive adventure whose results will depend more on luck than on talent. And meteorology, dear reader, cannot depend on luck.
Organization
Organizing is putting the work into motion: handing out the blueprints, assigning the trades, and setting the rules of the job, because no matter how good the bricks are, the building won’t rise on its own.
From this section onward we’ll assume the radars have already been acquired—by donation or purchase—and are ready to work. It is assumed that before acquiring them, a study was made of the dangerous phenomena derived from weather situations such as floods, landslides, mudslides, destructive winds (whether linear like downbursts or rotational like tornadoes), violent hail, or any other. And it is also assumed that the radars were suitably sited for detecting and tracking the situations that cause them. So let’s move on to the organizational activities.
Verification and calibration
If you’re the kind of person who goes to the hardware store and buys a shiny new digital multimeter from a good brand, and thinks that makes it ready to measure, you might start using your radar’s data «as the manufacturer gives it to you.» Then come the complaints that the Particle Classifier shows snow in very warm areas—a real complaint from a student in one of my courses.
Not only must you verify every parameter upon radar reception, but you must also establish routines and schedules for verification and calibration at regular intervals. That’s about periodicity, but much more complicated is establishing the methodologies for verification and calibration of all the radar modules and of the radar as a whole. Sometimes the manufacturer supplies some very general indications.
I have held in my hands the routines they use in the factory and in field radars, and they are extensive, detailed documents. Those routines require adaptation to the radars, to field conditions, and to the instruments and methods available in the service—that is, meticulous customization.
Modus vivendi
There are countries where radars belong to two or more different organizations. Each bought them for different purposes. For example, in Panama, two radars from different organizations are used harmoniously for the benefit of the country. In Colombia, four radars are operated by IDEAM and five by Civil Aviation (assuming the one in Pasto is still alive—I can’t find news about it, nor do I know whether the three new radars acquired from Baron are operational).
ICAO and WMO share a document they call a Modus Vivendi to regulate interactions related to meteorology. At the country level, agreements and documents are also needed so that the radars of both organizations serve not only their own purposes but also those of the other.
Let me give you an example: wind shear is disastrous for aviation, but it’s generally of little concern to the general population. Similarly, floods rarely affect airports, because they were built safe from them.
Coordination and cooperation are needed for the country’s benefit, but that cannot be left to goodwill; detailed instructions are required for joint operation in the face of each type of phenomenon.
Mode of operation
I usually divide it into three aspects: the work regime, the scanning strategy, and the mode of operation (see my post Operación de los radares meteorológicos for details). There are hundreds of ways these aspects can be combined to make an observation. In the NEXRAD ecosystem they called it VCP (Volume Coverage Pattern—see my post VCP en NEXRAD: El misterio de explorar la atmósfera), i.e., the way we scan the atmosphere. Many elements of a radar’s behavior can be modified to explore the atmosphere.
In operational practice, it’s very cumbersome to configure a VCP on the fly for every phenomenon; you need to have them configured in advance, and not so many that you have to study before choosing one. Some dangerous phenomena have a life cycle of only 10‑15 minutes. You have to make the right decision about which VCP can best capture it in just a few minutes.
Exchanging images (of some products) between radars of different organizations is good. Exchanging primary data is even better. But if you don’t unify the modes of operation and ensure the same modes for the same phenomena, the exchange is just a mirage.
IT aspects
By IT aspects I mean organizing the data flow from the radars to the end users. If the data flow has no alternative paths, it can be interrupted at any moment. Without redundancy, it’s a «fair‑weather» system—and meteorology, precisely, must be a system that works in «bad weather.» No matter how secure the fiber optics from the radar to the end users are, at some stretch it can break, or run out of backup power. This happened to us in Cuba with the Pico San Juan radar during Hurricane Irma (read that post and draw your own conclusions). Note, dear reader, that I didn’t say «from the radar to the forecast center»—I said «to the end users.» Radar data should impact the end users directly (population, businesses, institutions, media).
Linked to that is data archiving. A radar surveillance system must have its own and duplicated data archive, in different locations, with protections, with duplicated backups. Otherwise, it’s a waste of the money invested in the radar—like throwing the investment in the trash. True, the primary use of radar data is operational forecasting, but the data archive is essential for studying climatology, for studying long‑term drought (by the absence of radar‑measured rainfall data), and for applying Big Data techniques to gain new knowledge.
It’s very good to put data on public servers (like AWS); but the data put there must be in a uniform, accessible, and known format. If it’s stored in a proprietary format, you’re using public foreign servers as a substitute for the institutional archive that should belong to the institution—that’s my opinion in light of my long experience. A bomb in AWS servers and the archive is gone. Does that sound far‑fetched? Amazon and other big players have built their data centers in the Middle East, and we all know what can happen there—besides, ownership of that information passes to the foreign operator.
The National Center for Atmospheric Research (NCAR) has placed the primary and secondary data from all 159 NWS radars on AWS servers. The formats of these data (Level II and Level III) are known, and the NWS has released the ORPG code for many years. We ourselves (LADETEC) have developed an application to read Level III data (LAMULA™ WebViewer), an application to read Level II data (LAMULA™ Process), and an application to convert primary data from any radar into Level II format (LAMULA™ ORPG)—all thanks to the fact that the data, the formats, and the programs are freely available. That, in my opinion, is the right and sustainable path to promote the use of radar information. Old radar data is invaluable for training and improving numerical forecast models for the country in question.
Staying tied to the manufacturer’s software is unsustainable in the long run, in addition to the inconveniences of after‑sales service delays. After some time, the manufacturer has no obligation to maintain compatibility with older versions. If that happens, the country or institution that stored the data in the manufacturer’s proprietary format either abandons them or pays a lot of money to someone to write rescue software.
Human factor and role definition
So far we’ve talked about machines, institutional agreements, and data flows. But the most fragile link—and at the same time the most valuable—in this whole chain is the person on the other side of the screen. It’s not enough to have the world’s most expensive radar if the operator can’t interpret a tornado echo or doesn’t distinguish between ground clutter and a real convective cell.
The organization must clearly define, in writing, the profiles of each position: the maintenance technician (who looks after the electrical and mechanical health of the radar), the operational meteorologist (who monitors data quality and decides changes in scanning mode), the shift supervisor (who authorizes the issuance of alerts and coordinates with other institutions), and the data archiving officer (who ensures that the information is not lost).
From the definition of positions and their work contents, you must define the training plans for each one.
But defining positions isn’t enough. You must establish clear chains of command for emergency situations. When the hurricane is 100 km off the coast and the radar starts giving intermittent faults, the operator cannot be calling their boss, who calls another boss, and so on. At that moment, the organization must have foreseen who makes which decision and within how many seconds.
Writing manuals: the radar’s «road book»
A radar is not operated by intuition or by the memory of those who have been there the longest. That memory retires, gets sick, or leaves for another institution. A serious organization stands on living documents, updated and accessible.
Here I refer to two types of manuals, distinct but complementary:
- Operations manual: contains step‑by‑step instructions for starting up the radar, configuring the different VCPs (scan modes), starting data recording, switching channels (H/V), and performing daily noise and gain checks. It’s the equivalent of the car owner’s manual, but specific to each radar model and the particular conditions of its installation (altitude, climate, local interference).
- Procedures manual: this is more strategic. It says what to do for each scenario. For example: «If reflectivity > 55 dBZ is detected within 20 km, switch to the severe‑storm VCP and notify the shift supervisor within 2 minutes.» Or «If the transmitter fails during a storm, activate the backup radar (if available) or fall back to the rain‑gauge network and satellite, and report the incident to technical support within the next 5 minutes.»
Writing these manuals takes time, effort, and above all, field testing. They are not drafted from an office far from the radar. They are built with the participation of the operators and technicians themselves, because they are the ones who know the quirks and virtues of their machine. Once written, they must be printed right next to the control station and also available in digital format on several devices. The perfect manual is useless if it’s locked in a drawer.
Training
Training is a continuous process. Annually, for each position, it is necessary to organize training for the staff, taking into account the competencies expected of them and their knowledge from other fields—therefore, it is done for each individual. I knew very good practices under the direction of Professor M. T. Abshaev at the Anti‑Hail Service in the North Caucasus. Both engineers and meteorologists were retrained every couple of years, because, as Abshaev told me, «we teachers ourselves learn new things every time.»
Technical support
Technical support for one or several radars takes one of two forms: (1) build your own team of engineers, or (2) contract the service to a third party (outsourcing, as they call it).
Good radar engineers are not abundant in the world, that’s the truth. Even radar manufacturers are not flush with good integration engineers—those who know all the systems and the whole assembly in detail. The radar is such a complicated machine that it takes a long time to fully assimilate. Manufacturers pay their engineers very well to keep them from leaving for other things—not necessarily even other radar firms. Since so many fields converge in radars (take a look at this post to convince yourself), those engineers leave and succeed brilliantly in any of them. So it’s not easy for any country/institution to have its own engineers.
Some countries/institutions choose to pay a third party and «not worry» (as if that were possible) about the technical service of their radars. They generally turn to a local technology company. That company, in turn, is forced to establish contact with the manufacturer. Here there is a whole spectrum of intermediaries and manufacturers. Things can go well, badly, or so‑so. But what I am sure of is that they will pay a lot of money, and as for the result—I wish them luck. Whether you have your own engineers or not, you should have a hybrid expert who knows radar technology and its meteorological applications, so they can demand proper service from whoever is contracted for technical support.
In my experience, the only sustainable option is to have the experts in‑house. The best examples: Argentina and Brazil.
Execution
Execution is when the storm knocks on your door. When the «movie director» shouts: Action!
Planning and organizing are desk exercises and meetings. Execution is the heat of battle. It’s the moment when the operator puts down their coffee, sits in front of the screens, and puts everything they prepared into practice. Execution is not improvisation; it’s the perfect staging of the manuals and the roles.
Execution consists of five synchronized movements:
1. Active monitoring and hot quality control.
The operator cannot passively watch the colors on the screen. They must be hunting for anomalies in real time: radio‑frequency interference, bird or insect echoes, attenuation by heavy rain that hides areas behind the storm. If they detect a spurious datum, they must have the skill and authority to apply filters or to null that measurement before it reaches the end user. It’s like being an airplane pilot: you don’t just watch the instruments; you listen to the engine and feel the vibrations.
2. Agile and decisive switching of scan modes (VCP).
We already defined the VCPs during organization. In execution, the operator must decide, in seconds, which one is most suitable. A severe phenomenon can be born, grow, and die in 10 to 15 minutes. If the operator takes 5 minutes to decide that they need to switch to a faster scan and lower elevations, they’ve already missed the storm’s development phase. Execution here is muscle memory: knowing which button to press and which parameter to modify without hesitation, even when the phone is ringing and the boss is looking over your shoulder.
3. Rigorous application of the manuals (SOPs) during a contingency.
This is where all the effort of writing the manuals pays off. When the radar goes blind in the middle of a hurricane—as happened to Cuba with the Pico San Juan radar—the operator has no time to invent. They must open the procedures manual and follow the action sequence: check power sources, restart the system, activate the alternate data link, notify technical support, and meanwhile switch to secondary data sources (satellite, rain gauges). Disciplined execution, step by step, is what separates a professional service from chaos.
4. Real‑time inter‑institutional coordination (the Modus Vivendi in action).
The agreements with Civil Aviation, Civil Defense, or IDEAM are already signed. But executing them means picking up the phone and saying: «Hey, our radar detects wind shear on the runway—activate your flight‑diversion protocol.» Or warning the mayor’s office: «We have 80 mm of accumulated rain in the river basin—evacuate the riverbanks in 30 minutes.» Waiting for the other party to call is not an option; proactive execution saves lives.
5. Generation and dissemination of the final product to the user.
Reflectivity or velocity data is cryptic to a firefighter or an ordinary citizen. Executing well means translating that information into a clear, georeferenced, actionable message: «Imminent flooding in neighborhood X, street Y, at 14:30.» And it’s not enough to upload it to a website. You must push it actively via audible alarms, mobile apps, text messages, and bulletins to the press. The end user doesn’t have to be a meteorologist; the radar’s execution ends when that user understands the danger and acts accordingly.
Control
Control is measuring to improve, auditing to avoid failure. The radar is neither an ornament nor a trophy.
Once it’s operating and all missions have been executed, you have to sit down calmly and evaluate the results. Control is the feedback loop of the system. Without control, organization and planning become obsolete, and execution becomes mechanical routine.
Control operates on three levels:
1. Control of the data flow.
The data flow must be checked and logged for statistical purposes. You check whether the radar executed the observation, and then you log its presence at every instance it must pass through—both the primary data and the derived products. In Cuba we developed software for this (VESTA|Supervisor) that automatically checked the entire flow and displayed a map of Cuba where each radar appeared as a small circle with a color code that showed at a glance what had happened with each observation. It also monitored the actions executed by the operators and left a text log of every comment, warning, or alert issued by each operator.
2. Control of the final data quality.
It’s not enough to calibrate the radar every so often. You must systematically compare radar‑estimated rainfall with ground‑based rain‑gauge data. If the radar says 50 mm fell and the gauges show 35 mm, there’s a deviation to correct in the algorithms. This cross‑check should be done weekly or monthly, and its results documented to adjust calibration parameters. Also, you must check the performance of the algorithms for hail, mesocyclones, tornadoes, and downbursts. Likewise, check the Particle Classifier, which must distinguish rain, hail, snow, ground targets, birds, and insects.
3. Control of compliance with procedures.
After each significant event (a hurricane, a hailstorm, a tornado), a lessons‑learned meeting should be held. Was the correct VCP applied? Did the operator follow the manual to the letter? Were there delays in notification? Did technical support respond within the agreed time? This is not about finding someone to blame, but about identifying deviations to correct the manuals and improve staff training. If a procedure is not followed because it’s impossible to follow, then the problem is in the manual, not in the person.
4. Control of archiving and data sustainability.
The data archive is not a drawer where you throw information and forget it. Control requires periodic audits of the backup system. Are the data being copied to the two different locations at the agreed frequency? Are the formats still readable? Have emergency restorations been tested? Many institutions store data for years without ever verifying that they can actually be read. Good control means performing restoration tests at least once a year.
Control closes the loop. Its findings must feed back into Planning: if spare parts ran out faster than expected, readjust the stock. If a manual was confusing, rewrite it. If an operator made an error, reinforce their training. Weather radar surveillance is not a project that ends; it’s a living process that perfects itself with each cycle.
So far we have gone through the entire life cycle of a weather radar surveillance service: we planned it with our feet on the ground, we organized it with workshop discipline, we executed it with the coolness of someone who knows that every second counts, and we controlled it with the critical eye of someone who is not satisfied with what «seems» to work.
If there is one idea I want you to burn into your memory after reading these lines, it is this: the radar is not a household appliance that you plug in and forget; it is a living system that demands daily attention, technical care, and a long‑term vision. He who buys a radar thinking he has bought «automatic weather safety» is buying, in reality, an expensive disappointment. He who buys a radar knowing that he has to plan, organize, execute, and control is buying a tool that can save lives, protect crops, and guide decision‑making for decades.
But the knowledge doesn’t end here. Every radar is a different world, every country has its unique phenomena, every institution has its limitations and strengths. What I have shared with you is the fruit of many years of work, of mistakes learned from and successes celebrated—but I’m sure there is much more to say and do.
That’s why I invite you to contact me. If you are in the process of acquiring a radar, if you already have one and don’t know where to start organizing it, if your radar is «alive but limping,» or if you simply want to exchange views with someone who has been in the trenches, I am at your disposal.
You can write to me at [email protected], visit our website at https://ladetec.com, or find me on professional and social networks. I don’t promise to have all the answers, but I do promise to listen carefully, to share what I know, and if I don’t know something, to look together for someone who can help.
Because weather radar surveillance is not a one‑person or one‑institution affair. It’s a collective effort, and the more we share knowledge, the better prepared we will all be when the storm—inevitably—arrives.
Thank you for reading this far. This post is long, but it’s the material of an entire course in which I try to explain my long practice with Cuban radars in the radar surveillance service. Now, put on your helmets, turn on the radar, and… happy surveillance!
