Forecasting and Nowcasting Cloud Cover and Their Evaluation Using Satellite Data

Online real-time forecast cloud verification: Surface Observations, Forecasts, and satellite imagery interpretation

Weather Impact Decision Aid (WIDA)Workshop

12-15 March 2012Reno, Nevada

Presented by :Matt Young

Darko Koracin, Desert Research Institute, 2215 Raggio Parkway, Department of Atmospheric Sciences, Reno, NV. Email: Darko.Koracin@dri.edu

Ramesh Vellore, Centre for Climate Change Research Indian Institute of Tropical Meteorology, Pashan Pune, India. Email: rameshv@tropmet.res.in

John Lewis, Desert Research Institute, 2215 Raggio Parkway, Department of Atmospheric Sciences, Reno, NV. Email: John.Lewis@dri.edu

Melanie Wetzel (ret), Desert Research Institute, Department of Atmospheric Sciences, Reno, NV

To develop and test a mesoscale forecasting system with sub-kilometer horizontal resolution to support the NOWCAST system for Fallon Naval Air Station (NAS).

To improve the accuracy of the forecasts and nowcasts by assimilating asynchronous data into the forecasting system.

To provide short-term and accurate predictions of cloud structure relevant aircraft operations in the complex terrain of western and central Nevada.

To develop methods of mesoscale ensemble forecasting to improve the applicability of the forecasts and nowcasts.

A special emphasis will be on the accuracy of the cloud ceiling forecasts and impact on aviation safety.

Data Analysis/Modeling to Support NOWCAST & Forecast Activities at the Fallon Naval Air Station

The main characteristics of the real time forecasting system include:

• Numerical model data results posted on the dedicated web site with password protection [URL: http://www.adim.dri.edu/] and are updated with model runs every 12 hrs when in operation.

• 24-hr forecasts provided for five interactive domains: (27 km, 9.0 km, 3.0 km 1.0 km .333km)

• the highest sub-kilometer resolution (330m) for the innermost domain is centered on the NAS Fallon runway.

• A new automatic pre-forecast data assimilation simultaneously using:

– Satellite data with new methodology to improve model initial conditions in the vicinity of the NAS Fallon area and initial boundary layer height over the entire model domain.

– Data from four meteorological stations that were set up in the NAS Fallon area during a previous ONR project.

– Doppler weather radar data.

The Support & Forecast Activities Data Analysis/Modeling System Features

• To support real-time forecasting and nowcasting at the NAS Fallon range complex, surface meteorological stations were installed at the following locations during the project period:

– B17 at Centroid in Fairview Valley (39°19'27" N, 118°13'22" W, 4235’MSL). Location : 23 miles ESE from NAS Fallon.

– B19 at Blowing Sand Mountains (39°08'31"N, 118°40'01"W, 3886’MSL). Location: 16 miles SSE from NAS Fallon.

– B20 at Carson Sink (38°54'40" N, 118°23'14" W, 3881’ MSL). Location: 31 miles NNE from NAS Fallon.

– EW71 at Edwards Creek Valley (39°31'57" N, 117°44'50" W, 5192’ MSL). Location: 11 miles NE of Cold Springs, NV in Edwards valley

B20

B17

B19

EW71

The Support & Forecast Activities Data Analysis/ Modeling System Surface Observation Sites

The optimum utilization of satellite data has become a new component in cloud nowcasting research to monitor and predict the short-term physical characteristics of boundary layer cloud and thermodynamic conditions in the vicinity of the cloud top (cloud top temp – CTT).

To estimate cloud-top height for low-level clouds, a method had been developed that combines the above-cloud information extracted from the satellite imagery and the below-cloud ground truth information obtained from weather station measurements . (Vellore et al. 2006)

Forecasting and Nowcasting Cloud Cover and Their Evaluation Using Satellite Data

Two distinct products are displayed:

1.Determination of cloud types.

2.Improvement in the surface temperature forecasts as a consequence of more realistic initial boundary layer structure.

Regarding a new method of satellite data assimilation

Fig. (b): Observed surface temperature, satellite-measured cloud top temperature, and simulated time series of air temperature at the surface for the simulations initialized at 1200 UTC on 24 April 2006.

Indicated by Fig b, difference between the CCT and the measured surface temperature indicated that the clouds were absent at the time for which the simulated contours of the integrated liquid water path are shown in Fig. a.

Without assimilating satellite data, the model significantly overestimated cloudiness and underestimated the surface temperature (Fig. b).

The satellite data assimilation improved the forecasts of cloudiness and consequently air temperature in the boundary layer.

A schematic of the satellite data assimilation system is shown next two slides

No SAT assimilation SAT assimilation

24APR06 OBS KNFL 1800Z 36006kt 70 BKN 10.0mi 56F/39F 30.01INS

Consider only the pixels from within the circle for data assimilation

Channel 2 (3.9 microns)Channel 4 (10.7 microns)

Reno sounding (Tsfc, T400mb, T700mb

Average surface air temp(Reno and Fallon)Tavg

Cloud top temperature (CCT) =Channel 4 brightness temperaturee

Total number of pixels (NTOT) is a circleencompassing the stations Reno and

Fallon (radial distance=distance between Reno and Fallon Naval Air Station

(station ID:B17) Reno (39.56 N; 119.57 W)B17 (39.324 N; 118.223 W)

NNO =No. of cloud-free pixelsNLOW=No. of low-level cloud pixelsNMID=No. of mid-level cloud pixels

NHIGH=No. of high-cloud pixelsNTOT=Total number of pixels

=(NNO+NLOW+NMID+NHIGH)

If CTT>Tavg-4K NoYes

If CTT>T700 mb

If CTT< 233.15K(or)

CTT<T400 mb

NighttimeCh.4-Ch.2>2K

DaytimeCh.4-Ch.2 <-25K

Low-LevelClouds (NLOW)

MiddleLevel clouds

(NMID)

High clouds(NMID)

[mixed phase]

[Ice phase]

NoNoYesCloud

Identification

INPUTSGround Truth/

Satellite

Flowchart for low and middle cloud identification

and satellite data assimilation for real time

forecasting system

No clouds(NNO)

Base sounding at B17:Construction of

distance weighted sounding at B17 using

Reno and Elko soundings

Partially/fullycloudy (NLOW+NMID/NTOT

is 30% or greater than (NNO/NTOT)

(High clouds are not considered

Inadequate cloudiness/ No clouds

Yes

Assumption single cloud

layer

Cloud base: Adiabatic LCL (or) cloud ceiling

from ceilometer observations

Cloud top: Matching the CTT in

the sounding

Humidity between the cloud base and cloud top is > 80%

Nudge the modified temperature and

moisture soundings into the model analysis

Force the moisture nearly saturated

between the cloud base and cloud top

Model outputs/

verification

Data assimilation

Yes

Nudge the base sounding at B17

only into the model analysis

NO

Flowchart for low and middle cloud identification and satellite data assimilation for real time forecasting system

(Cont)

• Visualization systems can be a useful tool for a time-efficient and perception-rich search through large forecast outputs

• A person in virtual environment can interactively probe forecast results with respect to various model domains, times, and forecasted parameters.

• For the visualization to the left, white represents clouds and cyan represents areas of high turbulence kinetic energy.

Visualization of the real-time forecasts of turbulence and cloudiness in a virtual environment

Visualization of the real-time forecasts of turbulence and cloudiness over the runway for the Fallon NAS in

the CAVE (Cave Automatic Virtual Environment).

• This research summarizes a method and a strategy for evaluating cloud top heights based on a combination of infrared reflectance values from the geostationary (GOES) satellites with the ground-truth measurements.

• It should be mentioned that cloud-top-height (CTH) is not observed directly except with the use of either radiosonde data, satellite LIDAR (Light Detection And Ranging), aircraft LIDAR, or aircraft pireps if available. LIDAR is usually used for cloud base measurements only.

• The analysis towards improvement of cloud predictions in a real-time forecasting system follows a sequence of steps:

– Satellite data collection and quality control– Cloud classification– Data assimilation of cloud information into NWP models– Forecasting of cloud structure and evolution– Visualization and assessment of cloud products

• The main advantage of the developed method is to improve the inherently inaccurate initial and boundary conditions obtained from global and regional-scale models with coarse horizontal and vertical resolution that are required input to the high-resolution forecasting and nowcasting.

SUMMARY

QUESTIONS ???

Hoffman, R. N., and E. Kalnay, 1983: Lagged average forecasting, an alternative to Monte Carlo forecasting. Tellus, 35A, 100-118.

Lee, T. F., F. J. Turk, and K. Richardson, 1997: Stratus and fog products using GOES-8-9 3.9 μm data. Weather and Forecasting, 12, 664-677.

Lewis, J., 2005: Roots of ensemble forecasting. Mon. Wea. Rev., 133, 1865-1885.

Lewis, J., S. Lakshmivarahan, and S. Dhall, 2006: Dynamic Data Assimilation: A Least Squares Approach. Cambridge Univ. Press, 648.

Menzel, W. P., and Coauthors, 1998: Application of GOES 8/9 soundings to weather forecasting and nowcasting. Bulletin of the American Meteorological Society, 79, 2059-2077 Stauffer, D. R., and N. L. Seaman, 1990: Use of four-dimensional data assimilation in a limited area model. Part I: Experiments with synoptic scale data. Mon. Wea. Rev., 118, 1250-1277.

Vellore, R., D. Koracin, and M. Wetzel, 2006: Improvements in cloud predictions throughmodeling and visualization of the satellite imagery. Lectures in Computer Science, LNCS 4292, Vol. II, 544-553.

Vellore, R., D. Koracin, M. Wetzel, S. K. Chai, and Q. Wang, 2007: Challenges in mesoscale predictions of nocturnal stratocumulus-topped marine boundary layer and implications for operational forecasting. Weather and Forecasting, 22, 1101-1122. Willmott, C. J. 1982: Some comments on the evaluation of model performance. Bull. Amer. Met. Soc., 63, 1309-1313.

References: