Spatio-temporal　variations　of　watershed　evolution,　a　process　under　rainfall　erosion,　soil　erosion,　and　kinetic　energy　variation,　are　diverse,　complex,　and　dynamic　phenomena.　In　this　study,　a　clear　evolution　process　of　the　loess　watershed　under　an　artificial　rainfall　experiment　was　monitored　with　close-range　photogrammetry.　Via　complex　network　theory,　we　constructed　the　watershed　weighted　complex　network　(WWCN)　to　simulate　watershed　spatial　structures　in　nine　evolution　stages.　On　this　basis,　we　fully　analyzed　and　discussed　the　watershed　structural　variation　during　the　watershed　evolution　process　from　the　aspect　of　node,　edges,　polygon,　and　overall　spatial　structure.　The　quantitative　indexes　showed　a　regular　variation　following　the　loess　watershed　evolution,　with　a　drastic　trend　in　the　early　and　active　evolution　stages,　whilst　a　slow　trend　in　the　stable　evolution　stages.　From　the　perspective　of　watershed　spatial　structure,　fresh　insights　into　the　watershed　structure　(such　as　the　closeness,　community　effect,　connectivity,　stability,　and　stability　of　the　watershed　network)　are　introduced　to　comprehensively　explain　the　watershed　evolution　process.　Besides,　we　found　watershed　structural　indexes　are　closely　related　to　the　typical　terrain　indexes,　which　may　suggest　that　watershed　structures　have　responses　to　characteristics　of　watershed　morphologies　during　the　watershed　evolution　process.　This　study　is　beneficial　to　deepen　further　understanding　of　the　formation　genesis　and　variation　laws　of　the　watershed　evolution　process　as　well　as　present　a　potential　method　for　landform　quantitative　analysis.　©　2023,　The　Author(s),　under　exclusive　licence　to　Springer-Verlag　GmbH　Germany,　part　of　Springer　Nature.