VARIATIONS IN LOCAL 3D CLOUD STRUCTURE IMPACT THE INTERPRETATION OF NASA CERES AND MODIS DATA FOR TOP-OF-ATMOSPHERE RADIATION STUDIES FROM CLOUDS. CURRENTLY CLOUD RADIATIVE FEEDBACK MECHANISMS ARE ONE OF THE LARGEST SOURCES OF UNCERTAINTY IN GLOBAL CLIMATE MODELS. MUCH OF THIS UNCERTAINTY RESULTS FROM LACK OF KNOWLEDGE OF CLOUD PARTICLE PHASE AND VERTICAL AND HORIZONTAL STRUCTURE.HOWEVER THE CERES SPATIAL RESOLUTION OF 20 KM DOES NOT PERMIT RADIATION VARIATIONS BY MOST CLOUDS TO BE RESOLVED AND BOTH CERES AND MODIS ARE INSENSITIVE TO THE VERTICAL STRUCTURE AND PHASE DISTRIBUTION IN MANY CLOUDS. SURFACE-BASED DATA ON THE 3-D CLOUD STRUCTURE AND PHASE FOR AN ENSEMBLE OF CLOUDS IS THUS NEEDED DURING SATELLITE OVERPASSES TO BE ABLE TO INTERCOMPARE RADIATIVE TRANSFERMODELS BASED ON MODIS AND OTHER SATELLITE DATA WITH CERES DATA TO IMPROVE THE 3-D PARAMETERIZATION OF GCM CLOUD MODELS. SUCH DATA CAN BE OBTAINED USING A MULTI-SENSOR ARRAY OF GROUND-BASED VISIBLE-LIGHT CAMERAS. CLOSELY RELATED THE FORECASTING OF SOLAR INSOLATION AND ASSOCIATED CLOUD COVER ON TIME SCALES OUT TO ~1 HOUR AHEAD AND WITH A SPATIAL RESOLUTION OF ~100 METERS IS RECOGNIZED TO BE VALUABLE FOR STABILIZING POWER GRIDS WITH HIGH PENETRATIONS OF SOLAR PHOTOVOLTAIC ENERGY. DATA FOR CLOUD-ADVECTION BASED SOLAR INSOLATION FORECASTING CAN CONCEIVABLY BE OBTAINED USING GEOSTATIONARY SENSORS BUT NOT WITH THEREQUISITE SPATIAL RESOLUTION NOR LATENCY OF SECONDS NEEDED TO ADEQUATELY PREDICT MOTION DURING HIGH RAMP RATE EVENTS. MOREOVER THE BOTTOM-UP PERSPECTIVE VIEW OF CLOUDS PROVIDES DATA THAT IS MORE STRONGLY CORRELATED WITH CLOUD-INDUCED FLUCTUATIONS IN SOLAR POWER THAN DATA OBTAINED FROM TOP-DOWN. THE DEVELOPMENT OF ADVANCED ELECTRIC GRID MANAGEMENT PRACTICES FOR IMPROVED INTEGRATION OF RENEWABLE SOLAR ENERGY WOULD BENEFIT FROM A COMPLEMENTARY MEANS OF OBSERVING CLOUDS USING A MULTI-SENSOR ARRAY OF LOW-LATENCYGROUND-BASED CAMERAS. THE ABOVE DATA NEEDS FOR BOTH 3D CLOUD RADIATION MODEL VALIDATION AND SOLAR FORECASTING CAN BE ADDRESSED USING A NETWORK OF LOWCOST UPWARD-LOOKING VISIBLE LIGHT CCD SKY CAMERAS POSITIONED AT ~2 KM SPACING OVER AN AREA OF ~30-60 KM IN SIZE AND SYNCHRONIZED TO ACQUIRE AND REPORT IMAGERY ON 30 SECOND INTERVALS. USING MODERN ELECTRONIC INTEGRATION AND FABRICATION METHODS SUCH CAMERAS CAN BE MANUFACTURED IN QUANTITY AND DEPLOYED AT A MARGINAL COST OF ~$200-$400 EACH AND OPERATED UNATTENDED FOR YEARS USING EXISTING COMMUNICATIONS INFRASTRUCTURE TO TRANSMIT IMAGERY WITH EXTREMELY LOW (<5 S) LATENCY. HOWEVER THE INITIAL INVESTMENT FOR SUCH A DISTRIBUTED SENSOR NETWORK IS TOO LARGE TO JUSTIFY WITHOUT A TRIAL PHASE BY WHICH TO UNDERSTAND THE POTENTIAL UTILITY OF UP-LOOKING MULTISENSORY VISIBLE IMAGERY IN SOLAR FORECASTING AND CLOUD RADIATION SCIENCE. TO DEVELOP THE DATA SETS NECESSARY TO OPTIMALLY DESIGN A MULTI-SENSOR CLOUD CAMERA NETWORK WE PROPOSE TO ORGANIZE A TEAM OF ~100-200 CITIZEN SCIENTISTS USING (PHASE A PROTOTYPE) SELF-OWNED PDA CAMERAS AND (PHASE B IMPLEMENTATION) OFFERING SENSOR SITES WITH WIFI ACCESS TO COLLECT UNIQUE DISTRIBUTED DATA SETS SUITABLE FOR MODIS-CERES CLOUD RADIATION SCIENCE AND SOLAR FORECASTING ALGORITHM DEVELOPMENT. THE DATA SETS WILL BE OBSERVED OVER BOULDER CO COVERING ~10-15% OF BOULDER COUNTY AT 30-SECOND RESOLUTION AND SPECIFICALLY DURING INTERVALS OF AQUA ASCENDING AFTERNOON (~1:40 PM LOCAL) OVERPASSES. THE DATA SET WILL BE USED IN 3D CLOUD RECONSTRUCTION AND FORECASTING ALGORITHMS TO DEMONSTRATE A PROTOTYPE DISTRIBUTED NETWORK. THE PROJECT WILL BENEFIT FROM THE SUPPORT OF THE CITY OF BOULDER CO FOR BOTH ORGANIZING THE CITIZEN TEAMS NECESSARY FOR THE DATA COLLECTION AND FOR SITING THE NETWORK STATIONS DURING AN ANTICIPATED CITIZEN-FOCUSED FOLLOW-ON IMPLEMENTATION PHASE. A LOW-COST AND ROBUST SENSOR DESIGN SUITABLE FOR LARGE SCALE FABRICATION AND LONG TERM DEPLOYMENT WILL BE DEVELOPED DURING THE PROTOTYPING PHASE.