Physical risk of assets
Which assets are most at risk, and why?
Wildfire, flood and heat for every plant, line or substation, today and in 2041-2070; for grids, checked against real faults.
- Insurers
- Funds and banks
- Distribution companies
Maps · data · reports on the territory
The territory,
tile by tile.
We turn public satellite, climate and power grid data into answers: where to connect, which assets are exposed and what can fail. Using physical models, statistics and machine learning, at the scale of your project and with the reliability of every figure.
Why TESELA
Anyone can plot public data. What we do is turn it into intelligence: answers you can check.
Models, not just maps
We combine sources, downscale them with physics and machine learning and validate the result against what actually happened: faults, wildfires, measurements.
At the scale of the decision
From 9-30 km cells down to whatever the data allows, and from there to what you decide on: an asset, a plot, a grid node or a municipality.
With a confidence score
Every figure carries a score from 0 to 10 and what it has been checked against. If the score is low, we tell you which data would improve it.
Public data first
We start from public data: it belongs to everyone and anyone can check what we say.
Copernicus satellites and services, the CNMC, REE, MITECO, the land registry, OpenStreetMap. Every source is cited with its date, and on top of that comes our work: cleaning, combining, modelling and validating.
When a study calls for it, the client can provide their own data: measurements, fault history, asset inventory. It is used only for their study, never mixed with anything else and never leaves it.
We start with Spain, but almost all the sources are European. The same work can be done in any EU country.
Downscaling
Coarse
From the model to your plot.
This is how a global reanalysis sees your area: a few cells per province.
Illustration using a TESELA map · the final resolution is whatever the data allows
Anatomy of a TESELA figure
- Question
- Is a transmission line more likely to fail in the year fire burns along its route?
- Result
- 28 % of lines had a fault, versus 12 % when the route does not burn. The probability almost doubles (×1.9).
- Data
- REE quality-of-service audits 2021-2025 (3,285 line-years), EFFIS wildfires, OpenStreetMap routes.
- Procedure
- A line-by-line model that separates fire from line length and voltage; the risk model was frozen in 2020 and tested against what happened afterwards.
- Margin
- p = 0,001: casi imposible que sea casualidad.
- Score
- 8/10
- Why it is not a 10
- Five years is short for rare events, and OpenStreetMap is not REE's inventory.
- What would raise it
- The utility's own line inventory and more years of fault records.
Method
Every figure has its score.
This is our guarantee standard. Every result is delivered with the procedure that produces it, the data it comes from, its margin and a score from 0 to 10 that tells you how far you can trust it. If the score is low, we tell you what would raise it. We would rather give you an honest figure than a spectacular one.
- From result to data, with no gaps. Each report describes its method step by step, so you can trace any figure back to the data behind it.
- Every piece of data, with its origin. A public source cited with its date, or your own data if you have provided it. We always say which is which.
- Caveats, in writing. If we are not fully convinced by a piece of data, we say so, with its score and the reason why.
- And how to improve it. Which data or what work would raise the score, so you can decide whether it is worth it.
No black box, no half-truths: data, a result with its reliability and a proposal to improve it.
8 /10
Fire and faults. Checked against 3,285 line-years of REE data.
6 /10
2041-2070 projection. Five climate models; depends on the scenario.
6 /10
Wind. With annual data it does not show up in the faults; that does not mean it does not matter.
DownloadEvery source, with its date.
ScaleAs far as the data allows.
OverlayIndicators per asset or plot.
ValidationAgainst what actually happened.
DeliveryReport, GIS and web.
Recent maps
This is TESELA.
Maps made with public data, dated and sourced. Each one answers a question and can be redone for your area. Click to see it in full.
Power grid · CNMC, 02-11-2026
The grid, node by node
Only 45 of the 937 transmission nodes have room for renewables: 1,413 MW. Reliability 9/10.
Risk · EFFIS, Copernicus · 2026
Where it burns and where summer dries out
Wildfire recurrence and summer dryness, cell by cell.
Rail · timetables for 10-11-2026, OpenStreetMap
Trains and protection, section by section
Trains per day and protection system (ASFA, ERTMS, LZB) for each section.
Grid and land · Nov 2026
Garoña node sheet
Suitable land within 25 km, free of protected areas, burnt or flood-prone zones.
News
What is changing around you.
Rules, data releases and decisions that affect the grid, land and risk. Each item comes with its source, date and score.
2Nov 2026
Power gridSource: CNMCScore 9/10CNMC publishes November grid access capacity
Only 45 of the 937 transmission-grid nodes have room for new renewable generation: 1,413 MW across Spain. The CNMC warns these are indicative values that cannot simply be added up.
8Oct 2026
Just transitionSource: MITECOScore 9/10MITECO's consultation on just-transition grid nodes closes
The prior public consultation on the 15 just-transition grid nodes closed on 8 October. The order would allocate capacity first come, first served, with local socio-economic commitments; it is not yet approved and sets no MW per node.
Services
What we can answer for you.
For those who make decisions about the territory: energy, insurance, funds, infrastructure, public administration and agriculture. We show what already has a published report and what is done on request.
Family 1
Grid and connection
Access capacity per node With example
Where is there room left on the grid, and how does it change each month? Node by node, with its land, its competition and its procedure.
See an example (PDF, in Spanish) →Where to site a plant With example
Suitable land around a node, free of protected areas, burnt or flood-prone zones, and what it costs to get there.
See an example (PDF, in Spanish) →Batteries and large loads On request
Nodes with capacity, free land nearby and where there is already a lot of solar: the places where storage or connecting a large load makes sense.
Family 2
Risk and climate
Physical risk of a grid With example
Which hazard really explains the faults, section by section, checked against REE's history and projected to 2041-2070.
See an example (PDF, in Spanish) →Risk for plants and portfolios With example
Fire, flood and heat for each plant, with return periods, today and at mid-century.
See an example (PDF, in Spanish) →Climate assessment for the EU Taxonomy On request
The physical risk and adaptation analysis required by Appendix A, with traceable data, scenarios and method.
Family 3
Territory and infrastructure
Wildfire near your lines With example
Which sections have burnable land nearby and how often it burns, to prioritise vegetation clearing and patrols. Sample for Cantabria.
See an example (PDF, in Spanish) →Rail network Map available
Traffic per section and level of signalling and protection, based on public timetables and OpenStreetMap.
Heat, water and land On request
Urban heat by neighbourhood, drought and water stress, agrivoltaics: the indicator at the scale of the plot or the municipality.
Family 4
On request: your question
What do you need to know about what is happening around you? We talk it through openly, look for the public data available, enrich it with yours if you have any, and tell you whether your question can be answered with data, and how reliably.
- We talkWhat decision you need to make and what you are missing to make it.
- We searchWhat public data exists, from when and at what resolution.
- We combineWith your data, if you provide it, only for your study.
- We answerWith the result, its score and how to improve it.
With example: there is a TESELA report you can download. On request: we do it with your area and your data.
Your question isn't here? If it has coordinates, it can be mapped.
Delivery formats
This is what you get.
Four formats depending on what you need: a one-off report, a sheet per site, a tailored study or monthly monitoring. Open each one to see what you would receive.
Power grid and land
Can you connect here, and on what land?
Access capacity at each node, free and unrestricted land around it, and what it costs to get there. With an alert when something changes.
- Developers
- Batteries
- Data centres
Tailored studies
Your question, with whatever data it takes.
We combine whatever public sources are needed, and yours if you want to provide them, bring them down to the scale the data allows and calculate the indicator you need.
- Consultancies
- Public administrations
- Agriculture and water
Monitoring
What has changed in your area this month?
Your layers kept up to date: a dated refresh of each source and an alert when something that affects you changes. No need to commission the study again.
- Asset portfolios
- Development teams
Case study · Spain's transmission grid
What really explains grid faults?
The question every grid operator asks in the face of climate change: which threat really brings lines down, and how much worse will it get? We answered it with public data, line by line, for all 683 lines at 220 and 400 kV.
Sources: REE (lines and faults), Copernicus, EFFIS, OpenStreetMap · TESELA, November 2026
1 · What we measure
28 % versus 12%
of lines had a fault in a year when fire burned inside their route. The probability almost doubles (×1.9).
Score 8/10
2 · What we cannot see
Wind and rain
Measured year by year, they do not explain the faults. Their failures are local and last hours; a 25 km grid cannot capture them.
Score 6/10
3 · What is coming
+35 % a +74 %
more fire-related faults in 2041-2070, depending on the scenario. 10% of the lines hold half the risk: that is where investment pays off.
Score 6/10
Behind TESELA
Abel Tortosa-Andreu
Scientist and engineer. Twenty years designing his own methods to predict what the atmosphere will do and turn it into energy decisions.
An astrophysicist by training, with doctoral research in solar physics at the Instituto de Astrofísica de Canarias. Since then he has built operational forecasting systems, invented algorithms that have made their way into industry and brought high-performance numerical computing to real energy problems. His hallmark: understanding the problem before touching the data and, when the method needed does not exist, creating it.
High-performance numerical computing
Magnetohydrodynamic simulation of the Sun on MareNostrum, parallel code with MPI and GPU computing with CUDA from its early days: accelerated finite volume and finite element methods for engineering.
Forecasting systems
Operational numerical weather prediction, wind and solar forecasting for Europe, Asia and Latin America, a wind-ramp forecasting system presented at the EGU, and electricity demand forecasting with model ensembles.
Proprietary algorithms
Re-Modeling, a long-term extrapolation method for wind time series later sold to Vestas, and its 4D extension; self-learned power curves and machine-learning gap-filling models.
From problem to solution
He has led forecasting teams and worked on both sides of the table: with those who commission a study and make decisions with it, and with those who code it. That is why he designs every answer around the decision behind it.
- Solar physics and supercomputingInstituto de Astrofísica de Canarias
Radiative magnetohydrodynamic simulation of the Sun on MareNostrum (BSC), with publications in Astronomy & Astrophysics.
- High-performance computing on GPUsCIMNE
GPU-accelerated finite volume and finite element algorithms for engineering simulation.
- Numerical prediction and forecastingMeteosim Truewind · AWS Truepower · UL
From operational model developer to head of forecasting and director of information services for Europe, the Middle East and Africa.
- Director of ForecastingVortex
R&D in atmospheric modelling, geostatistics and data science. Created Re-Modeling and a predictive system for wind farms.
- Wind fleet data and performanceHolaluz · GE Vernova
Demand forecasting and, at GE Vernova, predictive analytics with SCADA, fleet indicators and cloud-based machine learning.
- Wind resource, on the developer sideA renewable energy developer
Senior wind resource engineer and developer: long-term extrapolation, data gap-filling and wind farm design.
- Energy R&D and TESELAMadrid
All of the above, together: physics, data at scale, validated models and maps you understand at first glance.
Contact
What decision do you need to make?
Tell us about the project and the area. We will reply with a proposal and, if it fits, a sample map with your data.
To start, tell us
- The questionWhat you need to decide, and by when.
- The placeA point, a plot, a network or a portfolio of assets.
- Your dataIf you have them: measurements, failure history, inventory. They are optional and used only for your study.
We reply in Spanish, Catalan, French or English.