Control of Rainwater Harvesting Systems: Optical-Flow based Radar Nowcasting
(2026) In TVVR 5000 VVRM05 20261Division of Water Resources Engineering
- Abstract (Swedish)
- Urban water systems in southern Sweden face increasing pressure from both short-
duration intense rainfall, which can cause pluvial flooding and pollutant discharges, and
more frequent dry periods, which motivate water reuse (IPCC 2023; Schimanke et al.
2022; Canedo Rosso et al. 2025). Rainwater harvesting (RWH) tanks can contribute to
both goals (Campisano et al. 2017), but their detention benefit is reduced when storage
is unavailable ahead of storms. Forecast-based control (FBC) can create capacity
through pre-storm releases (Kerkez et al. 2016; Xu et al. 2022), but its viability depends
on precipitation forecast skill at short lead times.
This thesis investigates whether high-resolution radar-based nowcasting can provide
... (More) - Urban water systems in southern Sweden face increasing pressure from both short-
duration intense rainfall, which can cause pluvial flooding and pollutant discharges, and
more frequent dry periods, which motivate water reuse (IPCC 2023; Schimanke et al.
2022; Canedo Rosso et al. 2025). Rainwater harvesting (RWH) tanks can contribute to
both goals (Campisano et al. 2017), but their detention benefit is reduced when storage
is unavailable ahead of storms. Forecast-based control (FBC) can create capacity
through pre-storm releases (Kerkez et al. 2016; Xu et al. 2022), but its viability depends
on precipitation forecast skill at short lead times.
This thesis investigates whether high-resolution radar-based nowcasting can provide
forecast information of sufficient quality to support FBC of RWH systems in the Malmö
area. Using the PySTEPS framework (Pulkkinen et al. 2019), four extrapolation-based
nowcasting configurations were generated from Swedish radar composites and evaluated
for 19 intense precipitation events at 0.5 km / 5 min resolution, with additional verifi-
cation at the 2 km / 15 min configuration used by SMHI’s operational KNEP product
(SMHI 2025a). Forecasts were compared against Eulerian and Lagrangian persistence
baselines using continuous, categorical, and spatial verification metrics over lead times
of 5–90 min.
Nowcast skill depends strongly on season and event duration. Winter events retain
useful correlation beyond 60 min for 3 h durations, whereas convective summer events
decorrelate within about 15–30 min, consistent with the conventional decorrelation
framing based on Pearson correlation (Germann et al. 2002). At the lead times rele-
vant to control, S-PROG at native resolution and ANVIL at KNEP-equivalent reso-
lution maintain probability of detection above 0.7 and frequency bias close to unity;
the latter supports the operational choice of ANVIL for KNEP (Falahat 2026). The
direct effect of increasing radar resolution within a fixed method is modest, but the
best-performing method changes with resolution: S-PROG performs best at native
resolution, while ANVIL performs best at the coarser operational configuration.
Interpreted against required tank drawdown times from parallel emptying-time sim-
ulations (Sundstedt 2026), the results indicate that 30–90 min nowcast horizons can
support operational pre-storm releases under permissive outflow strategies and urban-
scale outflow constraints, while stricter constraints and strategies requiring a fully
empty tank at event onset require longer horizons and motivate blending with numer-
ical weather prediction. (Less) - Popular Abstract (Swedish)
- Södra Sverige brottas med två motsatta vattenproblem, ibland under samma sommar. Skyfall
dränker gatubrunnar och översvämmar gator och källare; torrperioder sänker vattennivåerna och
tvingar fram uppmaningar att spara på vattnet. En och samma anläggning skulle kunna hjälpa
mot båda: regnvattentanken. Den samlar takvatten till sådant som toalettspolning och bevattning
- och avlastar därmed dricksvattnet under torka - samtidigt som det lediga utrymmet i tanken
kan fånga upp nästa skyfall innan det når gatan.
Haken är att tanken inte kan göra båda sakerna samtidigt. För att förse med vatten bör den
vara full; för att dämpa översvämning bör den vara tom. Lösningen är att styra den aktivt: om
man visste att ett skyfall var en timme bort... (More) - Södra Sverige brottas med två motsatta vattenproblem, ibland under samma sommar. Skyfall
dränker gatubrunnar och översvämmar gator och källare; torrperioder sänker vattennivåerna och
tvingar fram uppmaningar att spara på vattnet. En och samma anläggning skulle kunna hjälpa
mot båda: regnvattentanken. Den samlar takvatten till sådant som toalettspolning och bevattning
- och avlastar därmed dricksvattnet under torka - samtidigt som det lediga utrymmet i tanken
kan fånga upp nästa skyfall innan det når gatan.
Haken är att tanken inte kan göra båda sakerna samtidigt. För att förse med vatten bör den
vara full; för att dämpa översvämning bör den vara tom. Lösningen är att styra den aktivt: om
man visste att ett skyfall var en timme bort kunde man släppa ut en del av det lagrade vattnet i
förväg och göra plats precis i tid. Detta kallas prognosstyrning, och allt hänger på en enda sak -
om regnprognosen för de närmaste timmarna går att lita på.
Med väderradardata från SMHI i ovanligt hög upplösning - en halv kilometer, var femte minut - tog
jag fram korttidsprognoser för 19 verkliga regntillfällen kring Malmö, både intensiva sommarskyfall
och bredare vinterregn. För varje prognos mätte jag hur långt fram den förblev tillförlitlig.
Två saker stack ut. För det första beror tillförlitligheten starkt på vilken sorts regn det är. Bredare
vinterregn kan förutsägas upp till ungefär en timme i förväg; intensiva sommarskyfall - just
de tillfällen som orsakar översvämning - tappar tillförlitlighet redan inom omkring 15 minuter,
eftersom sådana skurar växer och dör ut för snabbt för att följas. Besvärligt nog är prognoserna
som svårast precis när det står som mest på spel.
För det andra: genom att lägga dessa prognoshorisonter mot hur lång tid en tank behöver för att
tömmas - beräknat i ett parallellt examensarbete - kunde jag besvara den praktiska frågan: räcker
prognosen för att agera på? Svaret är ett villkorat ja. För mindre tankar och rimliga utflödeskrav
är en timmes framförhållning gott om tid. För stora tankar eller stränga krav, där tanken måste
vara helt tom innan skyfallet, räcker inte radarn ensam - den skulle behöva kombineras med
vanliga väderprognosmodeller som ser längre fram.
Resultatet är en kartläggning av när smarta, prognosstyrda regnvattentankar är värda att satsa på
- och när de ännu inte är det. I takt med att Sverige svänger allt kraftigare mellan översvämning
och torka är infrastruktur som tyst klarar båda, i stället för att vara byggd för bara det ena,
precis vad våra städer kommer att behöva. (Less) - Popular Abstract
- Southern Sweden faces two opposite water problems, sometimes in the same summer. Cloudbursts
overwhelm city drains and flood streets and basements; dry spells lower reservoirs and force
water-saving appeals. A single piece of infrastructure could help with both: the rainwater
harvesting tank. It collects roof runoff for uses such as toilet flushing and garden watering -
easing demand on drinking water during droughts - while the empty space in the tank can catch
the next downpour before it reaches the street.
The catch is that a tank cannot do both jobs at once. To supply water it should be full; to
prevent flooding it should be empty. The way out is to control it actively: if you knew a storm
was an hour away, you could release... (More) - Southern Sweden faces two opposite water problems, sometimes in the same summer. Cloudbursts
overwhelm city drains and flood streets and basements; dry spells lower reservoirs and force
water-saving appeals. A single piece of infrastructure could help with both: the rainwater
harvesting tank. It collects roof runoff for uses such as toilet flushing and garden watering -
easing demand on drinking water during droughts - while the empty space in the tank can catch
the next downpour before it reaches the street.
The catch is that a tank cannot do both jobs at once. To supply water it should be full; to
prevent flooding it should be empty. The way out is to control it actively: if you knew a storm
was an hour away, you could release some stored water in advance, making room just in time.
This is forecast-based control, and it lives or dies by one thing - whether the rain forecast for the
next hour can be trusted.
Using weather radar data from the Swedish Meteorological and Hydrological Institute at unusually
fine detail - half a kilometre, every five minutes - I generated short-term rain forecasts for 19 real
rainfall events around Malmö, covering both intense summer cloudbursts and broader winter rain.
For each forecast I measured how far ahead it stayed reliable.
Two things stood out. First, that reliability depends heavily on the kind of rain. Broad winter
rain can be predicted up to about an hour ahead; intense summer cloudbursts - the very events
that cause flooding - lose reliability within roughly 15 minutes, because such storms grow and
fade too quickly to follow. Awkwardly, forecasting is hardest exactly when the stakes are highest.
Second, by lining up these forecast horizons with how long a tank needs to empty - estimated
in a parallel thesis - I could answer the practical question: is the forecast good enough to act
on? The answer is a conditional yes. For modest tanks and gentle drainage limits, an hour of
warning is plenty. For large tanks or strict limits, where the tank must be completely empty
before the storm, radar alone is not enough - it would need to be combined with conventional
weather models that see further ahead.
The result is a mapping of when smart, forecast-driven rainwater tanks are worth deploying
- and when they are not yet. As Sweden swings more sharply between flooding and drought,
infrastructure that quietly handles both, instead of being designed for just one, is exactly what
cities will need. (Less)
Please use this url to cite or link to this publication:
https://lup.lub.lu.se/student-papers/record/9244097
- author
- Jansson, Melker LU
- supervisor
- organization
- course
- VVRM05 20261
- year
- 2026
- type
- H2 - Master's Degree (Two Years)
- subject
- keywords
- Nowcasting, Radar, Rainwater, Harvesting
- publication/series
- TVVR 5000
- report number
- TVVR26/5020
- ISSN
- 1101-9824
- language
- English
- additional info
- Examiner: Magnus Persson
- id
- 9244097
- date added to LUP
- 2026-07-07 23:18:02
- date last changed
- 2026-07-07 23:18:02
@misc{9244097,
abstract = {{Urban water systems in southern Sweden face increasing pressure from both short-
duration intense rainfall, which can cause pluvial flooding and pollutant discharges, and
more frequent dry periods, which motivate water reuse (IPCC 2023; Schimanke et al.
2022; Canedo Rosso et al. 2025). Rainwater harvesting (RWH) tanks can contribute to
both goals (Campisano et al. 2017), but their detention benefit is reduced when storage
is unavailable ahead of storms. Forecast-based control (FBC) can create capacity
through pre-storm releases (Kerkez et al. 2016; Xu et al. 2022), but its viability depends
on precipitation forecast skill at short lead times.
This thesis investigates whether high-resolution radar-based nowcasting can provide
forecast information of sufficient quality to support FBC of RWH systems in the Malmö
area. Using the PySTEPS framework (Pulkkinen et al. 2019), four extrapolation-based
nowcasting configurations were generated from Swedish radar composites and evaluated
for 19 intense precipitation events at 0.5 km / 5 min resolution, with additional verifi-
cation at the 2 km / 15 min configuration used by SMHI’s operational KNEP product
(SMHI 2025a). Forecasts were compared against Eulerian and Lagrangian persistence
baselines using continuous, categorical, and spatial verification metrics over lead times
of 5–90 min.
Nowcast skill depends strongly on season and event duration. Winter events retain
useful correlation beyond 60 min for 3 h durations, whereas convective summer events
decorrelate within about 15–30 min, consistent with the conventional decorrelation
framing based on Pearson correlation (Germann et al. 2002). At the lead times rele-
vant to control, S-PROG at native resolution and ANVIL at KNEP-equivalent reso-
lution maintain probability of detection above 0.7 and frequency bias close to unity;
the latter supports the operational choice of ANVIL for KNEP (Falahat 2026). The
direct effect of increasing radar resolution within a fixed method is modest, but the
best-performing method changes with resolution: S-PROG performs best at native
resolution, while ANVIL performs best at the coarser operational configuration.
Interpreted against required tank drawdown times from parallel emptying-time sim-
ulations (Sundstedt 2026), the results indicate that 30–90 min nowcast horizons can
support operational pre-storm releases under permissive outflow strategies and urban-
scale outflow constraints, while stricter constraints and strategies requiring a fully
empty tank at event onset require longer horizons and motivate blending with numer-
ical weather prediction.}},
author = {{Jansson, Melker}},
issn = {{1101-9824}},
language = {{eng}},
note = {{Student Paper}},
series = {{TVVR 5000}},
title = {{Control of Rainwater Harvesting Systems: Optical-Flow based Radar Nowcasting}},
year = {{2026}},
}