For decades, scientists have struggled to separate how much carbon dioxide ecosystems take in through photosynthesis from how much they release through respiration. The conventional partitioning method assumes that respiration responds to temperature in the same way at night and in daylight, which can make photosynthesis look too high and mislead climate and ecosystem models. We built NEON SIFnet to provide a clearer answer. It measures a faint glow from leaves, called solar-induced chlorophyll fluorescence, which is emitted only when photosynthesis is happening. Using identical instruments and shared procedures, we collect continuous data at very different sites—a deciduous forest in Tennessee, an evergreen forest in Washington, a desert shrubland in New Mexico, and a prairie in North Dakota. By converting this light signal into estimates of photosynthesis and comparing them with the conventional partitioning method, we found that our new approach provides a more accurate picture of photosynthesis, capturing how it changes with the seasons, plant type, and water availability—ecosystem functions that the conventional methods miss. This improved data is essential for better understanding both the carbon and water cycles, giving us a more reliable tool to monitor the ecosystem as the climate changes.