The APOLLO cloud product statistics Web service

May 2013 1

The APOLLO cloud product statistics Web service

Introduction DLR and Transvalor are preparing a new Web service to disseminate the statistics of the APOLLO cloud physical parameters as a further help in the characterization of a solar site. This Web service preparation phase is done with the help of a grant from ESA1. A very important part of our work is to assess this future Web service in order to respond as much as possible to the requirements expressed by the renewable energy services and users. This document presents the APOLLO cloud physical parameters computations and the different statistics based on this computation which will be exposed to the user by the Web service. Based on this description, we invite all the interested potential users to fill in and send us back the very short questionnaire send in conjunction with the present document.

The APOLLO methodology at a glance The APOLLO methodology uses multiple spectral channels of the METEOSAT Second Generation satellite (MSG) to discriminate between different cloud types.

Below is presented picture of water/mixed phase clouds on the left image and thin ice clouds on the right image, as seen from the ground.

1 contract 4000107680/13/I-AM; Expansion of the market for EO Based Information Services in waste

management, renewable energy and ecosystem services assessment

The APOLLO cloud product statistics Web service

May 2013 2

Karlsruher Wolkenatlas, copyright B. Mühr

The APOLLO methodology delivers cloud mask, cloud classification, cloud optical depth, liquid and ice water path, and cloud top temperature and infrared emissivity as cloud parameter products for each MSG SEVIRI pixel in a temporal resolution of 15 minutes during daytime, for the period 2004-2012 (8 years). The covered zone is [60°N,60°S,60°E,60W], with a resolution of 3x3 km2 at the nadir of the satellite [0°, 0°]. This resolution is about 4x5 km2 to 5x6 km2 in Europe. The following parameters are computed and stored:

Cloud mask and snow

Cloud coverage (0-100%)

Cloud type (low, medium, high water/mixed phase clouds; thin ice clouds)

Cloud optical depth

Cloud top temperature Additionally, a cloud classification scheme delivers information on:

vertically extended cold, very thick cloud-layers thin clouds warm and thick water clouds multi-layer clouds stratiform clouds

Main statistics delivered by the Apollo WEB service and their possible use for a solar plant sitting or conduct

Cloud mask This very basic statistic shows that in the Ulm (Germany) location pixel considered, for the year 2010, about 21% of the time slots were clear, 78% cloudy and 1% with snow on the ground.

The APOLLO cloud product statistics Web service

May 2013 3

Cloud type The cloud type can also be represented as a global number distribution for all the time slots or as a 2D histogram showing the changes in cloud type as a function of the hour of the day. In Ulm, 35% of the day time, there are medium water clouds. The 2D histogram shows also that the medium water clouds are predominant. It shows also that the clear sky periods, were the irradiation will be at its maximum, are at the beginning of the morning or at the end of the day.

Cloud overcast type With the computed APOLLO data at the pixel location of interest, it is possible to classify the cloud overcast type and to use it for instance to optimize a PV tilt angle. In Armstrong et al., Renewable Energy, 2010, the optimum tilt angle (OTA) is following the rules:

cloud free regions of the sun belt with dominating beam part : o OTA = function (latitude)

regions where the diffuse part dominates (> 45°N) have smaller OTA: o OTA = function (latitude, frequency of clouds and their type)

With a classification into: clear, bright overcast, dark overcast and partly cloudy.

clear (1) = at least 90% cloud free pixels; a clear pixel is defined as having max 10% cloud coverage.

bright overcast (2) = more than 50% of cloudy pixels, but at least 20% of them are thin ice clouds only

dark overcast (3) = more than 50% of cloudy pixels, but more than 80% of them are not thin ice clouds

partly cloudy (4) = the rest

For Cologne (Germany) and Almeria (Spain), the statistics are shown below:

The APOLLO cloud product statistics Web service

May 2013 4

In Almeria, where there are CSP plants relying on DNI, one can thus verify that there are only 8% of situations of bright overcast (high thin ice clouds). This is a typical situation with a sharp decrease of the DNI due to the diffraction of the direct irradiation (and the increase of apparent the sun disk angle), even though there is no very noticeable decrease of the GHI.

Cloud scatteredness The analysis is extended further in a 49x49 pixel window around the location of interest. In this window, several values are computed, mostly the number of cloud elements and the cloud shape/complexity (fractal box dimension). This permits for each time slot to define the aspect of the cloud cover. Broken, scattered and isolated clouds are defined on the 49x49 pixels window as:

Broken: window cloud fraction > 50%

Scattered: (window cloud fraction in [10%, 50%]) OR (window cloud fraction<10% and number cloud>5)

Isolated: window cloud fraction<10% AND number cloud <=5

With these elements, a 2D histogram (for the cloudy time slots only) is plotted which gives a global information on the aspect of the cloud cover at the location when clouds are present in the zone. Below are two examples of the results, the first one for Cologne (Germany) and the second one for Almeria (Spain).

It is clearly visible that in Cologne, the cloudy days are mostly overcast or with a few large clouds or broken clouds. In Almeria, the scattered clouds or isolated clouds skies are more frequent. For the scattered cloud cases, the consequences on the PV or CSP systems can be multiple:

Overshooting of global irradiance o ramps in irradiance within seconds o power converter outside optimum efficiency range o reduction of system lifetime

Rapid changes in direct irradiance o non-linear effects in heat transfer fluids

Power output of distributed PV generation o change of average PV production on transformer level o ramps of PV generation cause difficulties in the electricity grid o how many shaded photovoltaic systems do we expect in an area in the next hour?

The APOLLO cloud product statistics Web service

May 2013 5

Ramps A ramp is a change from cloudy to clear and vice versa in the cloud mask at the pixel location, from one time slot to the next. The number of ramps per days is an important parameter for the system definition. It can reduce the performance of a CSP system or necessitate a short term electricity storage system even for a PV plant to ensure the stability of the electricity output and the safety of the inverters. The two figures below are showing the distribution of the number of ramps per day in 2005 for Cologne (Germany - 1283 ramps in total for the year) and Almeria (Spain – 1081 ramps in total for the year).

Description of the planned Web service and request for feedback The service is intended to be accessible through the SoDa web site or as an offline request for a given site. The initial result of a query on a location will be a compilation document (.pdf) with all the major statistics results for that location. The availability of the data will increase from a few test points (Version 0) at the very beginning up to a full MSG disk 1° resolution grid for the first version of the product (Version 1). Initially, in the testing phase, we are looking for interested potential users for a testing/feedback phase. This feedback is very important to deliver the most valuable results for the users. Via the short questionnaire joined to this document, we expect to receive first a generic feedback from all potentially interested users on the interest and the results delivered by the WEB service:

Do you understand the statistical results delivered by the service?

Would you request further statistics based on the APOLLO cloud physical parameters?

Do you see a potential use of this statistics in your assessment of a new site sitting or the management of an existing site?

Do you think you would/could use such a service (and how often)?

Would you be prepared to pay (and maybe how much?) for: o A yearly subscription for the whole MSG covered zone via the SoDa web site o A statistic for a single location from time to time as an offline service

In a second phase, in a few months from now, we will offer to an extended panel of interested customers to choose one or a few test sites and deliver to them these statistics. We want to validate the use of these statistics through the comparison against the experience of the customer at the site location (ground measured irradiance, electrical output, cloud observations or weather patterns, etc…).