diff --git a/NEWS.md b/NEWS.md
index 608d9455d88aa94ea29989b9094612653c7e8896..b470ccbed0c599d11b99a7c7b4457a179ae136f1 100644
--- a/NEWS.md
+++ b/NEWS.md
@@ -4,6 +4,39 @@
 
 ### 1.6.10.9000 Release Notes (2021-03-02)
 
+#### New features
+
+- `CreateInputsModel()` gains a `QupstrUnit` argument in order to manage the unit of the flow in the `Qupstream` argument in case of the use of a semi-distributed version of a hydrolohical model. ([#110](https://gitlab.irstea.fr/HYCAR-Hydro/airgr/-/issues/110))
+- `RunModel_Lag()` gains a `QcontribDown` argument containing the time series of the runoff contribution of the downstream sub-basin in case of the use of a semi-distributed version of a hydrolohical model. ([#109](https://gitlab.irstea.fr/HYCAR-Hydro/airgr/-/issues/109))
+
+
+#### Bug fixes
+
+- Fixed bug in `RunModel`. The `RunModel_Lag()` can now be passed to the `FUN_MOD` argument. ([#108](https://gitlab.irstea.fr/HYCAR-Hydro/airgr/-/issues/108))
+- Fixed bug in `RunModel_Lag()`. The function no longer returns two values for a single time step run. ([#102](https://gitlab.irstea.fr/HYCAR-Hydro/airgr/-/issues/102))
+- Fixed bug in `RunModel_Lag()`. The `StateEnd` value is now correct when there are more than a single upstream basin. ([#103](https://gitlab.irstea.fr/HYCAR-Hydro/airgr/-/issues/103))
+- Fixed bug in `RunModel_Lag()`.  The `StateEnd` value is now correct if upstream flow units is the mm/time step. ([#104](https://gitlab.irstea.fr/HYCAR-Hydro/airgr/-/issues/104))
+- Fixed bug in `RunModel_CemaNeigeGR5H()`. The solid precipitation are now take into account in the GR5H model. ([#105](https://gitlab.irstea.fr/HYCAR-Hydro/airgr/-/issues/105))
+- Fixed bug in `RunModel_CemaNeige()` and `CreateInputsModel()`. `RunModel_CemaNeige()` now runs at the hourly time step. ([#106](https://gitlab.irstea.fr/HYCAR-Hydro/airgr/-/issues/106))
+- Fixed the 'param_optim' vignette. The starting points used for the multi-start approach are now in the transformed space.([#101](https://gitlab.irstea.fr/HYCAR-Hydro/airgr/-/issues/101))
+
+
+#### Major user-visible changes
+- `LengthHydro` must now be set in kilometers (not anymore in meters) in the `CreateInputsModel()` function. ([#112](https://gitlab.irstea.fr/HYCAR-Hydro/airgr/-/issues/112))
+- `TransfoParam_GR5H()` now use the same transformation as `TransfoParam_GR4H()` fo the X1 parameter. The previous transformation set by Ficchì seems unnecessary as it provokes irrealistically high X1 values. ([#50](https://gitlab.irstea.fr/HYCAR-Hydro/airgr/-/issues/50))
+
+
+#### Minor user-visible changes
+
+- The `RunModel*()` functions now run faster. The computation times are significantly shorter for long times series with many time steps (e.g. hourly times series), due to a better management of the missing values in and out the Fortran codes. Only simulation computation times have been improved (it is largely invisible to the user for calibration computation times). ([#113](https://gitlab.irstea.fr/HYCAR-Hydro/airgr/-/issues/113))
+- The external calibration algorithms used in 'param_optim' and 'param_mcmc' vignettes now  run faster.  The `RunModel_*()` used during the parameter estimation process now run faster because the outputs contain only the simulated flows (see the `Outputs_Sim` argument in the `CreateRunOptions()` help page).
+
+
+#### Version control and issue tracking
+
+- The CI pipelines now fail when the checks return a warning message (and not just an error message). ([#86](https://gitlab.irstea.fr/HYCAR-Hydro/airgr/-/issues/86))
+- The reverse dependencies packages (e.g. the 'airGRteaching' or the 'airGRdatassim') are now checked by the CI pipelines. ([#86](https://gitlab.irstea.fr/HYCAR-Hydro/airgr/-/issues/86))
+
 ____________________________________________________________________________________