Although Earth’s surface parameters obtained from satellite data have become more and more precise, it is still difficult to guarantee temporal consistency and spatial continuity for large-scale geoscience phenomena. Developing new Earth observation platforms is a feasible way to improve the consistency and continuity of such data. As the planet’s only natural satellite, the Moon has special advantages as a platform for observing Earth, including long lifetime, whole disk view, tectonic stability and unique perspective. After presenting the observation geometry constructed by using the ephemeris, this paper mainly discusses the characteristics of a lunar platform and the proper Moon-based sensors, as well as the scientific objectives of Moon-based Earth observation. Solid Earth dynamics, the energy budget of Earth, Earth’s environmental elements and the Earth-space environment are four potential applications analysed in this paper.

Spatially location and working status of pollution sources are very important pieces of information for environment protection. Waste gas produced by fossil fuel consumption in the industry is mainly discharged to the atmosphere through a chimney. Therefore, detecting the distribution of chimneys and their working status is of great significance to urban environment monitoring and environmental governance. In this paper, we use an open access dataset BUAA-FFPP60 and the faster regions with convolutional neural network (Faster R-CNN) algorithm to train the preliminarily detection model. Then, the trained model is used to detect the chimneys in three high-resolution remote sensing images of Google Maps, which is located in Tangshan city. The results show that a large number of false positive targets are detected. For working chimney detection, the recall rate is 77.27%, but the precision is only 40.47%. Therefore, two spatial analysis methods, the digital terrain model (DTM) filtering, and main direction test are introduced to remove the false chimneys. The DTM is generated by ZiYuan-3 satellite images and then registered to the high-resolution image. We set an elevation threshold to filter the false positive targets. After DTM filtering, we use principle component analysis (PCA) to calculate the main direction of each target image slice, and then use the main direction to remove false positive targets further. The results show that by using the combination of DTM filtering and main direction test, more than 95% false chimneys can be removed and, therefore, the detection precision is significantly increased.

The Moon provides us with a long-term, stable, and unique place for Earth observation. Space agencies of various countries, including the United States, China, and Italy, have made the realization of Moon-based Earth observation an objective of their lunar missions. To date, although some conceptual studies have been presented, an accurate geometric model for Moon-based Earth observation has not yet been described. This paper introduces such a geometric model, which connects the attitude of a Moon-based sensor with a corresponding field of view on the Earth’s surface. The aberration and light time correction are involved. Due to the lack of high-quality experimental data, one qualitative comparison with Chang’E lander images and another quantitative comparison with software output are made. The comparison results show good similarity. The overall model accuracy is evaluated to be better than 4″ at current stage and will be better than 1.5″ if the seleno-graphic position is accurately determined. In direct geolocation process, the aberration and light time will cause a total 0.7 km deviation on the ground near the sublunar point. In SAR range history simulation, the light time effect will lead to a linear error, as large as tens of meters, throughout the integration time.

To study the Earth’s energy balance and to extend exoplanet research, the Earth’s outward radiative flux and its radiance in the Moon-based view were simulated according to the Earth–Sun–Moon geometry model, with the help of ERA5. A framework was developed to identify the angular distribution model (ADM) of Earth’s surface and its scene types, according to the surface and atmospheric data from ERA5. Our simulation shows that the specific viewing geometry controls the periodical variations in the Moon-based view radiative flux and its radiance, which reflect the orbital period of the Moon. The seasonal variations in shortwave and longwave radiative flux follow the energy balance in general, which is probably influenced by the Earth albedo. The derived global ADM would help to identify the anisotropic factor of observations at DSCOVR. Our simulations prove that Moon-based observation is a valuable source for Earth observation and that the orbital information of exoplanets could be derived from the radiance observation.

The variation in the radiation budget at Earth’s top of the atmosphere (TOA) represents the most fundamental metric defining the status of global climate change. The accurate estimation of Earth’s shortwave radiant exitance is of critical importance to study Earth’s radiation budget (ERB) at TOA. Measuring Earth’s outgoing shortwave radiance (OSR) is a key point to estimate Earth’s shortwave radiant exitance. Compared with space-borne satellite systems, Moon-based sensors (MS) could provide large-scale, continuous, and long-term data for Earth radiation observations, bringing a new perspective on ERB. However, the factors affecting the estimation of Earth’s OSR in the lunar direction have not yet been fully explored, for example, anisotropic surface reflection and the effects of clouds and aerosols on radiation budget. In this work, we only focused on the influence of anisotropic surface reflection. To evaluate the extent of this influence, we constructed a model to estimate Earth’s OSR in the lunar direction (EOSRiLD), integrating the variables of anisotropic surface reflection (scene types, solar zenith angles, viewing zenith angles, and relative azimuth angles) and radiant flux in Moon-viewed sunlit regions. Then, we discussed it over three time periods (Earth’s rotation, revolution period, and synodic month cycle) and analyzed the impact of three variables (area of the Moon-viewed sunlit region, scene types, and incident-viewing angular bins) on anisotropic EOSRiLD. Our results indicate that EOSRiLD based on the assumptions of anisotropic and isotropic reflection is different but they all show the same monthly cycle change, which is related to the area of the Moon-viewed sunlit region. At the beginning and end of the lunar month, the differences between anisotropy and isotropy are greatest in each cycle; when it is close to the first half of each cycle, there is a small difference peak. Both anisotropy and isotropy are caused by the relative azimuth angles between the Sun and Moon. In conclusion, even if the Moon-based platform has a wider scope than space-borne satellites, the difference is still large between anisotropy and isotropy. Therefore, we still need to consider the anisotropic surface reflection based on the Moon-based observation.

In producing orthophoto mosaic in a large area from spaceborne synthetic aperture radar (SAR) images, SAR image matching from adjacent orbits is a technical difficulty due to the speckle noise and different imaging mechanism between azimuth and range direction. In this paper, an area-based method, SAR-Moravec, is proposed for SAR orthophoto matching from adjacent orbits in a large area. Compared with the classical area-based Moravec, the template of SAR-Moravec is characterized by more directions for speckle noise restraint and a main direction consistent with the azimuth. In order to get evenly distributed matching points with high accuracy, the grid control mechanism and Gaussian pyramid from coarse to fine are introduced in matching. The whole process contains three steps. First, the pyramid images are constructed by the down-sampling process. Second, under grid control, the feature points are evenly extracted by the modified template. Third, the transformation model is iteratively calculated from the first to the last layer of the pyramid. After the matching process layer-by-layer, the final matching points and transformation model can be obtained. In the experiments, we compare the SAR-Moravec with three widely used methods, including the Moravec, the SAR-scale invariant feature transform (SAR-SIFT), and the SAR-features from an accelerated segment test (SAR-FAST). The results indicate that the proposed method has the best global matching accuracy among these methods and the matching efficiency is better than SAR-SIFT and SAR-FAST methods in large area.